Datasets:

Hack90
/

europe_pmc_articles_part_2

Dataset card Viewer Files Files and versions Community

Split (1)

train · 1.2M rows

article_title stringlengths 4 3.51k	abstract stringlengths 13 23.5k	contributor stringlengths 29 482k	journal stringlengths 3 180	full_text stringlengths 21 5.74M
Ultra-high field magnetic resonance imaging of the basal ganglia and related structures	<p>Deep brain stimulation is a treatment for Parkinson's disease and other related disorders, involving the surgical placement of electrodes in the deeply situated basal ganglia or thalamic structures. Good clinical outcome requires accurate targeting. However, due to limited visibility of the target structures on routine clinical MR images, direct targeting of structures can be challenging. Non-clinical MR scanners with ultra-high magnetic field (7T or higher) have the potential to improve the quality of these images. This technology report provides an overview of the current possibilities of visualizing deep brain stimulation targets and their related structures with the aid of ultra-high field MRI. Reviewed studies showed improved resolution, contrast- and signal-to-noise ratios at ultra-high field. Sequences sensitive to magnetic susceptibility such as T2<sup>*</sup> and susceptibility weighted imaging and their maps in general showed the best visualization of target structures, including a separation between the subthalamic nucleus and the substantia nigra, the lamina pallidi medialis and lamina pallidi incompleta within the globus pallidus and substructures of the thalamus, including the ventral intermediate nucleus (Vim). This shows that the visibility, identification, and even subdivision of the small deep brain stimulation targets benefit from increased field strength. Although ultra-high field MR imaging is associated with increased risk of geometrical distortions, it has been shown that these distortions can be avoided or corrected to the extent where the effects are limited. The availability of ultra-high field MR scanners for humans seems to provide opportunities for a more accurate targeting for deep brain stimulation in patients with Parkinson's disease and related disorders.</p>	<contrib contrib-type="author"><name><surname>Plantinga</surname><given-names>Birgit R.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><xref ref-type="author-notes" rid="fn001"><sup></sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/156213"/></contrib><contrib contrib-type="author"><name><surname>Temel</surname><given-names>Yasin</given-names></name><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><xref ref-type="aff" rid="aff3"><sup>3</sup></xref><xref ref-type="author-notes" rid="fn001"><sup></sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/58959"/></contrib><contrib contrib-type="author"><name><surname>Roebroeck</surname><given-names>Alard</given-names></name><xref ref-type="aff" rid="aff4"><sup>4</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/13490"/></contrib><contrib contrib-type="author"><name><surname>Uludağ</surname><given-names>Kâmil</given-names></name><xref ref-type="aff" rid="aff4"><sup>4</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/2190"/></contrib><contrib contrib-type="author"><name><surname>Ivanov</surname><given-names>Dimo</given-names></name><xref ref-type="aff" rid="aff4"><sup>4</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/188797"/></contrib><contrib contrib-type="author"><name><surname>Kuijf</surname><given-names>Mark L.</given-names></name><xref ref-type="aff" rid="aff5"><sup>5</sup></xref></contrib><contrib contrib-type="author"><name><surname>ter Haar Romenij</surname><given-names>Bart M.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="aff" rid="aff6"><sup>6</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/158678"/></contrib>	Frontiers in Human Neuroscience	<sec sec-type="introduction" id="s1"><title>Introduction</title><p>The basal ganglia are a group of nuclei deep in the brain, which play an important role in specific motor, limbic, and associative processes (Temel et al., <xref rid="B92" ref-type="bibr">2005</xref>). Anatomically, they consist of the caudate nucleus-putamen (also referred to as striatum), external and internal globus pallidus (GPe and GPi, respectively), substantia nigra (SN), and the subthalamic nucleus (STN). Structural or functional impairments of basal ganglia structures can lead to neurological and psychiatric disorders, e.g., Parkinson's disease (PD) (Obeso et al., <xref rid="B79" ref-type="bibr">2008</xref>), dystonia (Wichmann and Dostrovsky, <xref rid="B99" ref-type="bibr">2011</xref>), Tourette's syndrome (Mink, <xref rid="B74" ref-type="bibr">2006</xref>), and obsessive-compulsive disorder (Maia et al., <xref rid="B69" ref-type="bibr">2008</xref>). Although most of the patients with basal ganglia diseases can be managed by drug and/or behavioral therapy, an increasing number of patients are referred to specialized teams for deep brain stimulation (DBS) (Lee et al., <xref rid="B62" ref-type="bibr">2007</xref>; Ackermans et al., <xref rid="B3" ref-type="bibr">2008</xref>; Limousin and Martinez-Torres, <xref rid="B64" ref-type="bibr">2008</xref>; Denys et al., <xref rid="B29" ref-type="bibr">2010</xref>). The main reasons for DBS referral include the proven benefit of DBS over best medical treatment (Deuschl et al., <xref rid="B30" ref-type="bibr">2006</xref>; Schuepbach et al., <xref rid="B90" ref-type="bibr">2013</xref>) or insufficient response to non-surgical therapies. DBS is a minimally invasive surgical procedure and involves the implantation of stimulating electrodes with millimeter precision into a specific brain target. The brain regions targeted most often are located in the basal ganglia, and include the ventral parts of the striatum (Malone et al., <xref rid="B70" ref-type="bibr">2009</xref>; Denys et al., <xref rid="B29" ref-type="bibr">2010</xref>), post-eroventral part of the GPi (Damier et al., <xref rid="B21" ref-type="bibr">2007</xref>; Lee et al., <xref rid="B62" ref-type="bibr">2007</xref>; Ackermans et al., <xref rid="B3" ref-type="bibr">2008</xref>), ventral and anterior parts of the pallidum (Ackermans et al., <xref rid="B3" ref-type="bibr">2008</xref>), the STN (Follett and Torres-Russotto, <xref rid="B38" ref-type="bibr">2012</xref>), and surrounding structures such as the ventrolateral and anterior parts of the thalamus (Fisher et al., <xref rid="B37" ref-type="bibr">2010</xref>).</p><p>Currently, there are three methods to locate the target for DBS: (a) using intraoperative neurophysiological mapping tools, (b) using stereotactic coordinates derived from <italic>post-mortem</italic> or magnetic resonance imaging (MRI) based atlases (indirect targeting), and (c) via direct visualization on individual magnetic resonance (MR) images (direct targeting). Combinations of these methods are generally used. Direct targeting has the advantage over indirect targeting in that it accounts for differences in individual anatomy, which is especially critical when small structures such as those in DBS are targeted. However, at standard clinical magnetic field strengths (1.5T and 3T) direct visualization often lacks contrast for very high precision DBS targeting. The increasing availability of ultra-high magnetic field (7T or higher) MR scanners promises direct, accurate visualization of target regions with a very high specificity. A better understanding of the structural and functional components of the basal ganglia and related structures at ultra-high resolution approaching the microscopic level, is not only expected to increase the accuracy of DBS, shorten surgery, and potentially improve the clinical outcomes (Yokoyama et al., <xref rid="B105" ref-type="bibr">2006</xref>; Wodarg et al., <xref rid="B102" ref-type="bibr">2012</xref>), but also to enhance our understanding of brain function and disease states. In this technology report, we present the current options for detailed visualization of deep-brain structures using multiple MRI contrasts at ultra-high magnetic field, based on a literature review.</p><p>English-language studies were searched on PubMed using combinations of title and abstract key words related to basal ganglia, thalamus, and ultra-high field MRI. Publications were selected by screening of titles and abstracts. Additional studies were found through the references cited in the selected articles.</p><p>In this technology report, anatomical structures are denoted in English, unless their Latin names are commonly used. In the first sections, we provide background information on the basic concepts of MRI, which we consider important to understand the different image types that can be obtained, and on the conventional methods of MR imaging of the basal ganglia. Subsequently, we review the current literature on <italic>in vivo</italic> and <italic>ex vivo</italic> (i.e., <italic>post-mortem</italic>) ultra-high field imaging of the basal ganglia and related structures.</p></sec><sec><title>Summary of the principles of magnetic resonance imaging</title><p>Whether and how well a certain brain structure is visible on an MR image depends on biophysical tissue parameters and MRI acquisition protocols (see Table <xref ref-type="table" rid="T1">1</xref>).</p><table-wrap id="T1" position="float"><label>Table 1</label><caption><p><bold>Important concepts in MR imaging</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="left" valign="top" rowspan="1" colspan="1"><bold>Variable</bold></th><th align="left" valign="top" rowspan="1" colspan="1"><bold>Definition</bold></th><th align="left" valign="top" rowspan="1" colspan="1"><bold>Specific for</bold></th><th align="left" valign="top" rowspan="1" colspan="1"><bold>Relevance</bold></th></tr></thead><tbody><tr><td align="left" valign="top" rowspan="1" colspan="1">T1</td><td align="left" valign="top" rowspan="1" colspan="1">Spin-lattice relaxation time</td><td align="left" valign="top" rowspan="1" colspan="1">Tissue</td><td align="left" valign="top" rowspan="2" colspan="1">Influences MR signal in tissue</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">T2</td><td align="left" valign="top" rowspan="1" colspan="1">Spin-spin relaxation time</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">T2<sup></sup></td><td align="left" valign="top" rowspan="1" colspan="1">T2<sup></sup> relaxation time</td><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">R1</td><td align="left" valign="top" rowspan="1" colspan="1">1/T1</td><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">R2</td><td align="left" valign="top" rowspan="1" colspan="1">1/T2</td><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">R2<sup></sup></td><td align="left" valign="top" rowspan="1" colspan="1">1/T2<sup></sup></td><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">TE</td><td align="left" valign="top" rowspan="1" colspan="1">Echo time</td><td align="left" valign="top" rowspan="1" colspan="1">Sequence</td><td align="left" valign="top" rowspan="3" colspan="1">Determines the generated contrast</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">TR</td><td align="left" valign="top" rowspan="1" colspan="1">Repetition time</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Flip angle</td><td align="left" valign="top" rowspan="1" colspan="1">Flip angle</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">χ</td><td align="left" valign="top" rowspan="1" colspan="1">Magnetic susceptibility</td><td align="left" valign="top" rowspan="1" colspan="1">Tissue</td><td align="left" valign="top" rowspan="1" colspan="1">Gives extra contrast to certain substances</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">SNR</td><td align="left" valign="top" rowspan="1" colspan="1">Signal-to-noise ratio</td><td align="left" valign="top" rowspan="1" colspan="1">Image</td><td align="left" valign="top" rowspan="2" colspan="1">Quantifies the quality of the image</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">CNR</td><td align="left" valign="top" rowspan="1" colspan="1">Contrast-to-noise ratio</td><td rowspan="1" colspan="1"/></tr></tbody></table></table-wrap><p>The relaxation times, T1, T2, and T2<sup></sup>, are time constants that describe magnetic spin interaction properties of nuclei, which depend, among other things, on the molecular composition and organization of the tissue and the strength of the main magnetic field. Often, the relaxation rates R1, R2, and R2<sup></sup> are used, defined as 1/T1, 1/T2, and 1/T2<sup></sup> respectively. MRI uses the dependencies of these relaxation times on tissue properties to generate contrast within an image.</p><p>The actual type of contrast is determined by the MRI pulse sequence that is used, such as spin-echo (SE) and gradient echo (GE) sequences. These MRI contrasts are sensitive to different biophysical properties of the tissue and it is a matter of intense research to quantitatively relate tissue composition and MRI contrasts. Thus, individual and combinations of MRI contrasts provide a window to examine microstructural properties of brain tissue. The different sequences are described by the combination of the properties of the gradient, radio-frequency pulses and timing parameters. Properties that are often varied are the echo time (TE), repetition time (TR), and flip angle. This can result in T1-, T2- or T2<sup></sup>-weighted images in which the contrast is mainly caused by differences in T1, T2, or T2<sup></sup> values of the tissue. The variability in sequences therefore facilitates optimization of the protocol for each structure of interest individually.</p><sec><title>Susceptibility weighted imaging</title><p>Susceptibility-weighted (SW) images can also be acquired (Haacke and Reichenbach, <xref rid="B43" ref-type="bibr">2011</xref>). These images are based on the principle that MR images are generally complex-valued, i.e., effectively two images are always acquired: a commonly used magnitude image that often directly displays the anatomical structures and a phase image that is usually disregarded. The phase image however is sensitive to the so-called magnetic susceptibility (χ). This property of tissues and substances alters the local magnetic field values. Paramagnetic materials have a positive χ and strengthen the magnetic field, and diamagnetic materials have a negative χ and weaken the magnetic field. Tissues with a susceptibility that differs from their surrounding structures, such as tissues with myelin and iron-containing substances, cause local deviations in the magnetic field inside and outside of the structures. This leads to local phase differences, which can then be extracted from the original phase images. In susceptibility-weighted imaging (SWI), these phase images are combined with the magnitude images, which can result in additional contrast, which particularly enhances the brain's (micro)vessels and the small deep brain structures.</p></sec><sec><title>Quantitative maps</title><p>Furthermore, post-processing techniques can be employed, to produce so-called T1, T2, T2<sup></sup>, or (quantitative) susceptibility maps, which display the quantitative T1, T2, T2<sup></sup>, or susceptibility values of each voxel in an image respectively. Sometimes R1-, R2- or R2<sup></sup>-values are computed instead, which are defined as 1/T1, 1/T2, and 1/T2<sup></sup> respectively.</p></sec><sec><title>Other techniques</title><p>In addition to structural imaging, diffusion-weighted imaging (DWI), which is directionally sensitive to water diffusion, gives complementary information (Le Bihan, <xref rid="B61" ref-type="bibr">2003</xref>). It can provide information on the location and orientation of neuronal fibers, aiding in visualization of these pathways (tractography) (Mori et al., <xref rid="B76" ref-type="bibr">1999</xref>) or super-resolution track-density imaging (TDI) (Calamante et al., <xref rid="B16" ref-type="bibr">2010</xref>). Furthermore, functional MRI (fMRI) can provide information on localized brain activity (Buxton, <xref rid="B14" ref-type="bibr">2013</xref>). Finally, DWI and fMRI can be used to compute the connectivity between two areas by computing the fiber paths between them (structural connectivity) or the correlation of functional activity (functional connectivity) respectively.</p></sec><sec><title>Field strength</title><p>In most DBS centers, the MR images are obtained from 1.5T or 3T MR scanners. However, in specialized neuroimaging centers, the possibilities of scanning at ultra-high field are increasingly being explored (Duyn, <xref rid="B34" ref-type="bibr">2012</xref>). Although the number keeps growing, at present an estimate of 61 human ultra-high field MR scanners has been installed or will be installed in the near future (see Table <xref ref-type="table" rid="T2">2</xref>). At ultra-high field alterations of physical properties can influence measurements both positively and negatively. Several issues including field strength dependent changes in relaxation times T1, T2, and T2<sup></sup>; increased B0 and B1 magnetic field inhomogeneities; and increased risks of tissue heating (Duyn, <xref rid="B34" ref-type="bibr">2012</xref>) make ultra-high field scanning more sensitive to inhomogeneous signal-to-noise ratios (SNR) and contrast-to-noise ratios (CNR), geometric distortions, and movement artifacts. This limits the use of T1-weighted, T2-weighted, and proton density weighted turbo spin echo (TSE) scan protocols that are commonly used in clinics (Hennig et al., <xref rid="B48" ref-type="bibr">1986</xref>). However, the alterations in relaxation times and the increased sensitivity to magnetic susceptibility have stimulated the focus of ultra-high field imaging to shift to susceptibility and T2<sup></sup>-dependent gradient echo sequences (Haase et al., <xref rid="B44" ref-type="bibr">2011</xref>). Furthermore, the SNR increases close to linearly with field strength, which offers the option to scan with higher spatial resolution (Vaughan et al., <xref rid="B96" ref-type="bibr">2001</xref>) and/or CNR in a shorter time (Duyn, <xref rid="B34" ref-type="bibr">2012</xref>). This makes ultra-high field MRI especially beneficial for detailed imaging of structures with altered magnetic susceptibility, such as the basal ganglia, myelin, and blood, which is also important for ultra-high field high-resolution fMRI imaging. Finally, side effects might occur during movement through the gradients of the strong field. The majority of the subjects have been reported to feel sensations when moving into or out of the bore, which was rated as unpleasant vertigo in 5–20% of the subjects (Glover et al., <xref rid="B41" ref-type="bibr">2007</xref>; Theysohn et al., <xref rid="B93" ref-type="bibr">2008</xref>) and a small number of subjects (approximately 3%) experienced a medium or strong metallic taste (Theysohn et al., <xref rid="B93" ref-type="bibr">2008</xref>).</p><table-wrap id="T2" position="float"><label>Table 2</label><caption><p><bold>Overview of ultra-high magnetic field (7T or higher) human MR scanners that have been installed or will be installed in the future according to the institutions' websites</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="left" valign="top" rowspan="1" colspan="1"><bold>Nr</bold></th><th align="left" valign="top" rowspan="1" colspan="1"><bold>Country</bold></th><th align="left" valign="top" rowspan="1" colspan="1"><bold>City</bold></th><th align="left" valign="top" rowspan="1" colspan="1"><bold>Institution, department</bold></th><th align="left" valign="top" rowspan="1" colspan="1"><bold>Manufacturer</bold></th><th align="center" valign="top" rowspan="1" colspan="1"><bold>Field strength (T)</bold></th><th align="left" valign="top" rowspan="1" colspan="1"><bold>Publication</bold></th></tr></thead><tbody><tr><td align="left" valign="top" rowspan="1" colspan="1">1</td><td align="left" valign="top" rowspan="1" colspan="1">Australia</td><td align="left" valign="top" rowspan="1" colspan="1">Melbourne</td><td align="left" valign="top" rowspan="1" colspan="1">Melbourne Brain Centre, Melbourne Brain Centre Imaging Unit</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">2</td><td align="left" valign="top" rowspan="1" colspan="1">Australia</td><td align="left" valign="top" rowspan="1" colspan="1">Brisbane</td><td align="left" valign="top" rowspan="1" colspan="1">University of Queensland, Centre for Advanced Imaging</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">3</td><td align="left" valign="top" rowspan="1" colspan="1">Austria</td><td align="left" valign="top" rowspan="1" colspan="1">Vienna</td><td align="left" valign="top" rowspan="1" colspan="1">Medical University of Vienna, MR Center of Excellence</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Hahn et al., <xref rid="B45" ref-type="bibr">2013</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">4</td><td align="left" valign="top" rowspan="1" colspan="1">Brazil</td><td align="left" valign="top" rowspan="1" colspan="1">Sao Paulo</td><td align="left" valign="top" rowspan="1" colspan="1">University of Sao Paulo</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">5</td><td align="left" valign="top" rowspan="1" colspan="1">Canada</td><td align="left" valign="top" rowspan="1" colspan="1">London</td><td align="left" valign="top" rowspan="1" colspan="1">Western University, Robarts Research Institute, Centre for Functional and Metabolic Mapping</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Goubran et al., <xref rid="B42" ref-type="bibr">2014</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">6</td><td align="left" valign="top" rowspan="1" colspan="1">Canada</td><td align="left" valign="top" rowspan="1" colspan="1">Toronto</td><td align="left" valign="top" rowspan="1" colspan="1">Toronto Western Hospital, Krembil Neuroscience Centre</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">China</td><td align="left" valign="top" rowspan="1" colspan="1">Beijing</td><td align="left" valign="top" rowspan="1" colspan="1">Chinese Academy of Sciences, State Key Laboratory of Brain and Cognitive Science</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">He et al., <xref rid="B47" ref-type="bibr">2014</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">8</td><td align="left" valign="top" rowspan="1" colspan="1">Denmark</td><td align="left" valign="top" rowspan="1" colspan="1">Copenhagen</td><td align="left" valign="top" rowspan="1" colspan="1">Hvidovre Hospital, Danish Research Centre for Magnetic Resonance</td><td align="left" valign="top" rowspan="1" colspan="1">Philips</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">9</td><td align="left" valign="top" rowspan="1" colspan="1">France</td><td align="left" valign="top" rowspan="1" colspan="1">Marseille</td><td align="left" valign="top" rowspan="1" colspan="1">Center for Magnetic Resonance in Biology and Medicine</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">10</td><td align="left" valign="top" rowspan="1" colspan="1">France</td><td align="left" valign="top" rowspan="1" colspan="1">Saclay</td><td align="left" valign="top" rowspan="1" colspan="1">Alternative Energies and Atomic Energy Commission, Life Sciences Division, Neurospin</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Boulant et al., <xref rid="B9" ref-type="bibr">2011</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">11</td><td align="left" valign="top" rowspan="1" colspan="1">France</td><td align="left" valign="top" rowspan="1" colspan="1">Saclay</td><td align="left" valign="top" rowspan="1" colspan="1">Alternative Energies and Atomic Energy Commission, Life Sciences Division, Neurospin</td><td align="left" valign="top" rowspan="1" colspan="1">Custom built</td><td align="center" valign="top" rowspan="1" colspan="1">11.7</td><td align="left" valign="top" rowspan="1" colspan="1">Vedrine et al., <xref rid="B97" ref-type="bibr">2014</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">12</td><td align="left" valign="top" rowspan="1" colspan="1">Germany</td><td align="left" valign="top" rowspan="1" colspan="1">Berlin</td><td align="left" valign="top" rowspan="1" colspan="1">Max-Delbrueck-Center for Molecular Medicine, Berlin Ultrahigh Field Facility</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Dieringer et al., <xref rid="B32" ref-type="bibr">2011</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">13</td><td align="left" valign="top" rowspan="1" colspan="1">Germany</td><td align="left" valign="top" rowspan="1" colspan="1">Bonn</td><td align="left" valign="top" rowspan="1" colspan="1">German Center for Neurodegenerative Diseases</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">14</td><td align="left" valign="top" rowspan="1" colspan="1">Germany</td><td align="left" valign="top" rowspan="1" colspan="1">Essen</td><td align="left" valign="top" rowspan="1" colspan="1">Erwin L. Hahn Institute for Magnetic Resonance Imaging</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Dammann et al., <xref rid="B22" ref-type="bibr">2011</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">15</td><td align="left" valign="top" rowspan="1" colspan="1">Germany</td><td align="left" valign="top" rowspan="1" colspan="1">Heidelberg</td><td align="left" valign="top" rowspan="1" colspan="1">German Cancer Research Center</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Hoffmann et al., <xref rid="B50" ref-type="bibr">2011</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">16</td><td align="left" valign="top" rowspan="1" colspan="1">Germany</td><td align="left" valign="top" rowspan="1" colspan="1">Jülich</td><td align="left" valign="top" rowspan="1" colspan="1">Research Centre Jülich, Institute of Neuroscience and Medicine</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">9.4</td><td align="left" valign="top" rowspan="1" colspan="1">Arrubla et al., <xref rid="B4" ref-type="bibr">2013</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">17</td><td align="left" valign="top" rowspan="1" colspan="1">Germany</td><td align="left" valign="top" rowspan="1" colspan="1">Leipzig</td><td align="left" valign="top" rowspan="1" colspan="1">Max Planck Institute for Human Cognitive and Brain Sciences,</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Deistung et al., <xref rid="B26" ref-type="bibr">2013a</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">18</td><td align="left" valign="top" rowspan="1" colspan="1">Germany</td><td align="left" valign="top" rowspan="1" colspan="1">Magdeburg</td><td align="left" valign="top" rowspan="1" colspan="1">Leibniz Institute for Neurobiology, Center for Advanced Imaging</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Hoffmann et al., <xref rid="B49" ref-type="bibr">2009</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">19</td><td align="left" valign="top" rowspan="1" colspan="1">Germany</td><td align="left" valign="top" rowspan="1" colspan="1">Tübingen</td><td align="left" valign="top" rowspan="1" colspan="1">Max Planck Institute for Biological Cybernetics</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">9.4</td><td align="left" valign="top" rowspan="1" colspan="1">Budde et al., <xref rid="B13" ref-type="bibr">2014</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">20</td><td align="left" valign="top" rowspan="1" colspan="1">Italy</td><td align="left" valign="top" rowspan="1" colspan="1">Pisa</td><td align="left" valign="top" rowspan="1" colspan="1">Imago7 Foundation</td><td align="left" valign="top" rowspan="1" colspan="1">GE</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Costagli et al., <xref rid="B20" ref-type="bibr">2014</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">21</td><td align="left" valign="top" rowspan="1" colspan="1">Japan</td><td align="left" valign="top" rowspan="1" colspan="1">Niigata</td><td align="left" valign="top" rowspan="1" colspan="1">University of Niigata, Center for Integrated Human Brain Science</td><td align="left" valign="top" rowspan="1" colspan="1">GE</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Kabasawa et al., <xref rid="B53" ref-type="bibr">2006</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">22</td><td align="left" valign="top" rowspan="1" colspan="1">Japan</td><td align="left" valign="top" rowspan="1" colspan="1">Morioka</td><td align="left" valign="top" rowspan="1" colspan="1">Iwate Medical University, Institute for Biomedical Sciences</td><td align="left" valign="top" rowspan="1" colspan="1">GE</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Sato and Kawagishi, <xref rid="B86" ref-type="bibr">2014</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">23</td><td align="left" valign="top" rowspan="1" colspan="1">Japan</td><td align="left" valign="top" rowspan="1" colspan="1">Suita City</td><td align="left" valign="top" rowspan="1" colspan="1">Center for Information and Neural Networks</td><td rowspan="1" colspan="1"/><td align="center" valign="top" rowspan="1" colspan="1">7</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">24</td><td align="left" valign="top" rowspan="1" colspan="1">Netherlands</td><td align="left" valign="top" rowspan="1" colspan="1">Leiden</td><td align="left" valign="top" rowspan="1" colspan="1">Leiden University Medical Center, C.J. Gorter Center for High Field Magnetic Resonance in the LUMC</td><td align="left" valign="top" rowspan="1" colspan="1">Philips</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Dzyubachyk et al., <xref rid="B35" ref-type="bibr">2013</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">25</td><td align="left" valign="top" rowspan="1" colspan="1">Netherlands</td><td align="left" valign="top" rowspan="1" colspan="1">Utrecht</td><td align="left" valign="top" rowspan="1" colspan="1">UMC Utrecht</td><td align="left" valign="top" rowspan="1" colspan="1">Philips</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">de Bresser et al., <xref rid="B24" ref-type="bibr">2013</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">26</td><td align="left" valign="top" rowspan="1" colspan="1">Netherlands</td><td align="left" valign="top" rowspan="1" colspan="1">Amsterdam</td><td align="left" valign="top" rowspan="1" colspan="1">Spinoza Centre for Neuroimaging</td><td align="left" valign="top" rowspan="1" colspan="1">Philips</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">27</td><td align="left" valign="top" rowspan="1" colspan="1">Netherlands</td><td align="left" valign="top" rowspan="1" colspan="1">Maastricht</td><td align="left" valign="top" rowspan="1" colspan="1">Maastricht University, Maastricht Brain Imaging Centre</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Ivanov et al., <xref rid="B52" ref-type="bibr">2014</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">28</td><td align="left" valign="top" rowspan="1" colspan="1">Netherlands</td><td align="left" valign="top" rowspan="1" colspan="1">Maastricht</td><td align="left" valign="top" rowspan="1" colspan="1">Maastricht University, Maastricht Brain Imaging Centre</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">9.4</td><td align="left" valign="top" rowspan="1" colspan="1">Cloos et al., <xref rid="B19" ref-type="bibr">2014</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">29</td><td align="left" valign="top" rowspan="1" colspan="1">Republic of Korea</td><td align="left" valign="top" rowspan="1" colspan="1">Icheon</td><td align="left" valign="top" rowspan="1" colspan="1">Gachon University of Medicine and Science, Neuroscience Research Institute</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Cho et al., <xref rid="B17" ref-type="bibr">2008</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">30</td><td align="left" valign="top" rowspan="1" colspan="1">Sweden</td><td align="left" valign="top" rowspan="1" colspan="1">Lund</td><td align="left" valign="top" rowspan="1" colspan="1">Lund University, Lund University Bioimaging Center</td><td align="left" valign="top" rowspan="1" colspan="1">Philips</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">31</td><td align="left" valign="top" rowspan="1" colspan="1">Switzerland</td><td align="left" valign="top" rowspan="1" colspan="1">Lausanne</td><td align="left" valign="top" rowspan="1" colspan="1">Centre d'Imagerie BioMédicale</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Kickler et al., <xref rid="B60" ref-type="bibr">2010</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">32</td><td align="left" valign="top" rowspan="1" colspan="1">Switzerland</td><td align="left" valign="top" rowspan="1" colspan="1">Zürich</td><td align="left" valign="top" rowspan="1" colspan="1">Swiss Federal Institute of Technology and University of Zurich, Institute for Biomedical Engineering</td><td align="left" valign="top" rowspan="1" colspan="1">Philips</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Wyss et al., <xref rid="B103" ref-type="bibr">2014</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">33</td><td align="left" valign="top" rowspan="1" colspan="1">UK</td><td align="left" valign="top" rowspan="1" colspan="1">Nottingham</td><td align="left" valign="top" rowspan="1" colspan="1">University of Nottingham, Sir Peter Mansfield Magnetic Resonance Centre</td><td align="left" valign="top" rowspan="1" colspan="1">Philips</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Lotfipour et al., <xref rid="B66" ref-type="bibr">2012</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">34</td><td align="left" valign="top" rowspan="1" colspan="1">UK</td><td align="left" valign="top" rowspan="1" colspan="1">Oxford</td><td align="left" valign="top" rowspan="1" colspan="1">University of Oxford, Oxford Centre for Functional Magnetic Resonance Imaging of the Brain</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Berrington et al., <xref rid="B7" ref-type="bibr">2014</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">35</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Auburn</td><td align="left" valign="top" rowspan="1" colspan="1">Auburn University, Magnetic Resonance Imaging Research Center</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Denney et al., <xref rid="B28" ref-type="bibr">2014</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">36</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Baltimore</td><td align="left" valign="top" rowspan="1" colspan="1">Kennedy Krieger Institute, FM Kirby Center for Functional Brain Imaging</td><td align="left" valign="top" rowspan="1" colspan="1">Philips</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Intrapiromkul et al., <xref rid="B51" ref-type="bibr">2013</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">37</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Bethesda</td><td align="left" valign="top" rowspan="1" colspan="1">National Institute of Health, Functional MRI Facility</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Gaitan et al., <xref rid="B40" ref-type="bibr">2013</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">38</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Bethesda</td><td align="left" valign="top" rowspan="1" colspan="1">National Institutes of Health, National Institute of Neurological Disorders and Stroke</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">11.7</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">39</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Boston</td><td align="left" valign="top" rowspan="1" colspan="1">Massachusetts General Hospital, Martinos Center for Biomedical Imaging</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Augustinack et al., <xref rid="B5" ref-type="bibr">2005</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">40</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Chapel Hill</td><td align="left" valign="top" rowspan="1" colspan="1">University of North Carolina</td><td rowspan="1" colspan="1"/><td align="center" valign="top" rowspan="1" colspan="1">7</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">41</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Chicago</td><td align="left" valign="top" rowspan="1" colspan="1">University of Illinois, Center for MR Research</td><td align="left" valign="top" rowspan="1" colspan="1">Custom built</td><td align="center" valign="top" rowspan="1" colspan="1">9.4</td><td align="left" valign="top" rowspan="1" colspan="1">Lu et al., <xref rid="B67" ref-type="bibr">2013</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">42</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Cleveland</td><td align="left" valign="top" rowspan="1" colspan="1">Cleveland Clinic</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">43</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Columbus</td><td align="left" valign="top" rowspan="1" colspan="1">Ohio State University, Department of Radiology</td><td align="left" valign="top" rowspan="1" colspan="1">Bruker</td><td align="center" valign="top" rowspan="1" colspan="1">8</td><td align="left" valign="top" rowspan="1" colspan="1">Bourekas et al., <xref rid="B10" ref-type="bibr">1999</xref>; Robitaille et al., <xref rid="B85" ref-type="bibr">1999</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">44</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Columbus</td><td align="left" valign="top" rowspan="1" colspan="1">Ohio State University, Department of Radiology</td><td align="left" valign="top" rowspan="1" colspan="1">Philips</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">45</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Minneapolis</td><td align="left" valign="top" rowspan="1" colspan="1">University of Minnesota, Center for Magnetic Resonance Research</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Abosch et al., <xref rid="B2" ref-type="bibr">2010</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">46</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Minneapolis</td><td align="left" valign="top" rowspan="1" colspan="1">University of Minnesota, Center for Magnetic Resonance Research</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">47</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Minneapolis</td><td align="left" valign="top" rowspan="1" colspan="1">University of Minnesota, Center for Magnetic Resonance Research</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">10.5</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">48</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Minneapolis</td><td align="left" valign="top" rowspan="1" colspan="1">University of Minnesota, Center for Magnetic Resonance Research</td><td align="left" valign="top" rowspan="1" colspan="1">Varian</td><td align="center" valign="top" rowspan="1" colspan="1">9.4</td><td align="left" valign="top" rowspan="1" colspan="1">Deelchand et al., <xref rid="B25" ref-type="bibr">2010</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">49</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Nashville</td><td align="left" valign="top" rowspan="1" colspan="1">Vanderbilt University, Institute of Imaging Science</td><td align="left" valign="top" rowspan="1" colspan="1">Philips</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Eapen et al., <xref rid="B36" ref-type="bibr">2011</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">50</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">New Haven</td><td align="left" valign="top" rowspan="1" colspan="1">Yale University, Magnetic Resonance Research Center</td><td align="left" valign="top" rowspan="1" colspan="1">Varian</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Pan et al., <xref rid="B81" ref-type="bibr">2010</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">51</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">New York</td><td align="left" valign="top" rowspan="1" colspan="1">New York University School of Medicine, Center for Biomedical Imaging</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Pakin et al., <xref rid="B80" ref-type="bibr">2006</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">52</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">New York</td><td align="left" valign="top" rowspan="1" colspan="1">Icahn School of Medicine at Mount Sinai, Translational and Molecular Imaging Institute</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">53</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Philadelphia</td><td align="left" valign="top" rowspan="1" colspan="1">University of Pennsylvania, Center For Magnetic Resonance And Optical Imaging</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Bhagat et al., <xref rid="B8" ref-type="bibr">2011</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">54</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Pittsburgh</td><td align="left" valign="top" rowspan="1" colspan="1">University of Pittsburgh, Magnetic Resonance Research Center</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Moon et al., <xref rid="B75" ref-type="bibr">2013</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">55</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Portland</td><td align="left" valign="top" rowspan="1" colspan="1">Oregon Health & Science University, Advanced Imaging Research Center</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">56</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">San Francisco</td><td align="left" valign="top" rowspan="1" colspan="1">San Francisco Veterans Affairs Medical Center, Center for Imaging of Neurodegenerative Diseases</td><td align="left" valign="top" rowspan="1" colspan="1">Siemens</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">57</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">San Francisco</td><td align="left" valign="top" rowspan="1" colspan="1">University of California, Department of Radiology and Biomedical Imaging</td><td align="left" valign="top" rowspan="1" colspan="1">GE</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Metcalf et al., <xref rid="B73" ref-type="bibr">2010</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">58</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Dallas</td><td align="left" valign="top" rowspan="1" colspan="1">University of Texas Southwestern Medical Center, Advanced Imaging Research Center</td><td align="left" valign="top" rowspan="1" colspan="1">Philips</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Ren et al., <xref rid="B83" ref-type="bibr">2013</xref></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">59</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Iowa City</td><td align="left" valign="top" rowspan="1" colspan="1">University of Iowa, Iowa Institute for Biomedical Imaging</td><td align="left" valign="top" rowspan="1" colspan="1">GE</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">60</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Milwaukee</td><td align="left" valign="top" rowspan="1" colspan="1">Medical College of Wisconsin, Center for Imaging Research</td><td align="left" valign="top" rowspan="1" colspan="1">GE</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">61</td><td align="left" valign="top" rowspan="1" colspan="1">USA</td><td align="left" valign="top" rowspan="1" colspan="1">Stanford</td><td align="left" valign="top" rowspan="1" colspan="1">Stanford University, Richard M. Lucas Center for Imaging</td><td align="left" valign="top" rowspan="1" colspan="1">GE</td><td align="center" valign="top" rowspan="1" colspan="1">7</td><td align="left" valign="top" rowspan="1" colspan="1">Kerchner et al., <xref rid="B55" ref-type="bibr">2012</xref></td></tr></tbody></table><table-wrap-foot><p>Because not all scanners are operational yet, the last column refers to publications in which the mentioned scanner is used.</p></table-wrap-foot></table-wrap></sec></sec><sec><title>MRI of DBS targets at clinical field strengths of 1.5T and 3T</title><p>Direct visualization and targeting of DBS structures based on 1.5T or 3T MR images obtained in clinical practice can be challenging. Several studies compared different scanning sequences for the visibility of the STN (Kerl et al., <xref rid="B58" ref-type="bibr">2012a</xref>; Liu et al., <xref rid="B65" ref-type="bibr">2013</xref>), GPi (Nolte et al., <xref rid="B77" ref-type="bibr">2012</xref>; Liu et al., <xref rid="B65" ref-type="bibr">2013</xref>), GPe (Nolte et al., <xref rid="B77" ref-type="bibr">2012</xref>), and zona incerta (ZI) (Kerl et al., <xref rid="B57" ref-type="bibr">2012b</xref>) and showed that T2<sup></sup> (Kerl et al., <xref rid="B58" ref-type="bibr">2012a</xref>,<xref rid="B59" ref-type="bibr">c</xref>; Nolte et al., <xref rid="B77" ref-type="bibr">2012</xref>) and quantitative susceptibility maps (Liu et al., <xref rid="B65" ref-type="bibr">2013</xref>) outperformed T1- and T2-weighted images. Furthermore, 3T functional and structural connectivity maps have been measured in healthy volunteers to visualize the functional subdivision of the STN, although higher spatial resolution is expected to reveal a more detailed anatomy (Brunenberg et al., <xref rid="B12" ref-type="bibr">2012</xref>). Also, a literature review concluded that there is no consensus whether 1.5T and 3T MRI are reliable and accurate enough to be employed for direct targeting of the STN, due to serious shortcomings in the contrast between the STN and surrounding structures (Brunenberg et al., <xref rid="B11" ref-type="bibr">2011</xref>). Visualization of the small substructures in the thalamus at lower field strengths is even less straightforward, primarily due to lack of contrast. One study identified four large thalamic nuclei groups on 3T magnetization-prepared rapid acquisition of gradient echo (MPRAGE) images (Bender et al., <xref rid="B6" ref-type="bibr">2011</xref>) and another study identified the centromedian nucleus directly on 3T proton density weighted MR images (Kanowski et al., <xref rid="B54" ref-type="bibr">2010</xref>). The thalamus was also segmented at 1.5T and 3T using DWI (Wiegell et al., <xref rid="B100" ref-type="bibr">2003</xref>; Unrath et al., <xref rid="B95" ref-type="bibr">2008</xref>; Pouratian et al., <xref rid="B82" ref-type="bibr">2011</xref>; Mang et al., <xref rid="B71" ref-type="bibr">2012</xref>) or a combination of ten different sequences (Yovel and Assaf, <xref rid="B106" ref-type="bibr">2007</xref>).</p><p>Although several sequences have been investigated for the visualization of basal ganglia structures at clinical field strengths, DBS structures such as the motor part of the STN, and certain regions within the thalamus, such as the ventrolateral nuclei, need to be displayed more distinctively in order to rely on these images solely for targeting.</p></sec><sec><title>Ultra-high field imaging of the deep-brain structures</title><p>Several studies identified deep-brain (sub)structures at ultra-high field using different MRI contrasts. These studies, reviewed below, show the high potential of ultra-high field MRI to accurately identify and delineate thalamic, parathalamic and subthalamic nuclei. Table <xref ref-type="table" rid="T3">3</xref> shows detailed scanning parameters of the described studies, referred to by line numbers.</p><table-wrap id="T3" position="float"><label>Table 3</label><caption><p><bold>Overview of acquisition parameters used in the described studies</bold>.</p></caption><graphic xlink:href="fnhum-08-00876-i0001"/><table-wrap-foot><p>Ax, axial; Cor, coronal; EPI, echo planar imaging; FLAIR, fluid attenuation inversion recovery; FLASH, fast low angle shot; FLASH-HB, FLASH with high bandwidth; FOV, field of view; GE, gradient echo; GRASE, gradient and spin echo; M-GE, multi-echo GE; MPRAGE, magnetization prepared rapid gradient echo; MP2RAGE, magnetization prepared 2 rapid acquisition gradient echoes; PD, Parkinson's disease; PIAF, parallel imaging acceleration factor; Sag, sagittal; SD, spin density; SE, spin-echo; S-GE, single-echo GE; SPACE, sampling perfection with application of optimized contrasts using different flip angle evolutions; SWI, susceptibility weighted imaging; T1w, T1-weighted; T2w, T2-weighted; T2<sup></sup>w, T2<sup></sup>-weighted; TDI, track density imaging; TI, inversion time; TR, repetition time; TSE, turbo spin-echo; χ-map, susceptibility map. Bandwidths that were originally reported in kHz have been converted to Hz/pixel and are denoted with an asterisk (<sup></sup>).</p></table-wrap-foot></table-wrap><sec><title>Visualization of deep-brain structures at ultra-high field <italic>in vivo</italic></title><p>Since the installation of the first ultra-high field MR scanner, several studies investigated the visualization of deep-brain structures at ultra-high field <italic>in vivo</italic> (Table <xref ref-type="table" rid="T4">4</xref>).</p><table-wrap id="T4" position="float"><label>Table 4</label><caption><p><bold>Overview of the basal ganglia and related (sub)structures that have been identified using different protocols at ultra-high field MRI</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="left" valign="top" rowspan="1" colspan="1"><bold>Study</bold></th><th align="left" valign="top" rowspan="1" colspan="1"><bold>Image type</bold></th><th align="left" valign="top" rowspan="1" colspan="1"><bold>Findings</bold></th><th align="center" valign="top" rowspan="1" colspan="1"><bold>Line</bold></th></tr></thead><tbody><tr><td align="left" valign="top" rowspan="1" colspan="1">Bourekas et al., <xref rid="B10" ref-type="bibr">1999</xref></td><td align="left" valign="top" rowspan="1" colspan="1">T2<sup></sup>w</td><td align="left" valign="top" rowspan="1" colspan="1">GP, SN, and RN appear hypointense</td><td align="center" valign="top" rowspan="1" colspan="1">1</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Novak et al., <xref rid="B78" ref-type="bibr">2001</xref></td><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">GP, SN, and RN appear hypointense</td><td align="center" valign="top" rowspan="1" colspan="1">2</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Abduljalil et al., <xref rid="B1" ref-type="bibr">2003</xref></td><td align="left" valign="top" rowspan="1" colspan="1">GE Magnitude</td><td align="left" valign="top" rowspan="1" colspan="1">SN and RN appear hypointense</td><td align="center" valign="top" rowspan="1" colspan="1">3</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">GE Phase</td><td align="left" valign="top" rowspan="1" colspan="1">Substructures within SN and RN</td><td align="center" valign="top" rowspan="1" colspan="1">3</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Cho et al., <xref rid="B17" ref-type="bibr">2008</xref></td><td align="left" valign="top" rowspan="1" colspan="1">GE</td><td align="left" valign="top" rowspan="1" colspan="1">SN and RN in coronal plane hypointense</td><td align="center" valign="top" rowspan="1" colspan="1">4</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Cho et al., <xref rid="B18" ref-type="bibr">2010</xref></td><td align="left" valign="top" rowspan="1" colspan="1">Coronal GE</td><td align="left" valign="top" rowspan="1" colspan="1">Discrimination of STN and SN</td><td align="center" valign="top" rowspan="1" colspan="1">38</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Abosch et al., <xref rid="B2" ref-type="bibr">2010</xref></td><td align="left" valign="top" rowspan="1" colspan="1">SWI</td><td align="left" valign="top" rowspan="1" colspan="1">Clear delineation of STN</td><td align="center" valign="top" rowspan="1" colspan="1">7–8</td></tr><tr><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">Boundary between STN and SN</td><td rowspan="1" colspan="1"/></tr><tr><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">Lamina pallidi medialis and lamina pallidi incompleta</td><td rowspan="1" colspan="1"/></tr><tr><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">Vim, anterior and medial boundaries of pulvinar, boundary of the nucleus ventralis caudalis</td><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Eapen et al., <xref rid="B36" ref-type="bibr">2011</xref></td><td align="left" valign="top" rowspan="1" colspan="1">T2w and T2<sup></sup>w</td><td align="left" valign="top" rowspan="1" colspan="1">Subregions within RN</td><td align="center" valign="top" rowspan="1" colspan="1">9</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">T2<sup></sup>w</td><td align="left" valign="top" rowspan="1" colspan="1">Subregions within RN and SN</td><td align="center" valign="top" rowspan="1" colspan="1">10</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Schafer et al., <xref rid="B87" ref-type="bibr">2012</xref></td><td align="left" valign="top" rowspan="1" colspan="1">χ-map</td><td align="left" valign="top" rowspan="1" colspan="1">Boundary between STN and SN</td><td align="center" valign="top" rowspan="1" colspan="1">11</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Deistung et al., <xref rid="B27" ref-type="bibr">2013b</xref></td><td align="left" valign="top" rowspan="1" colspan="1">χ-map</td><td align="left" valign="top" rowspan="1" colspan="1">Subnuclei within the SN</td><td align="center" valign="top" rowspan="1" colspan="1">12</td></tr><tr><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">Discrimination of the STN from the SN and surrounding gray and white matter</td><td rowspan="1" colspan="1"/></tr><tr><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">Lamina pallidi medialis and lamina pallidi incompleta</td><td rowspan="1" colspan="1"/></tr><tr><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">Medullary lamina in RN</td><td rowspan="1" colspan="1"/></tr><tr><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">Vim, pulvinar, lateral and medial geniculate nucleus, dorsomedial nucleus and dorsal nuclei group</td><td rowspan="1" colspan="1"/></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">R2<sup></sup>-map</td><td align="left" valign="top" rowspan="1" colspan="1">Substructures in RN</td><td align="center" valign="top" rowspan="1" colspan="1">13</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Lenglet et al., <xref rid="B63" ref-type="bibr">2012</xref></td><td align="left" valign="top" rowspan="1" colspan="1">Tractography</td><td align="left" valign="top" rowspan="1" colspan="1">Projection based subdivisions of the SN, STN, GP and thalamus</td><td align="center" valign="top" rowspan="1" colspan="1">14</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Calamante et al., <xref rid="B15" ref-type="bibr">2012</xref></td><td align="left" valign="top" rowspan="1" colspan="1">TDI</td><td align="left" valign="top" rowspan="1" colspan="1">Signal intensity differences within thalamus</td><td align="center" valign="top" rowspan="1" colspan="1">15</td></tr><tr><td align="left" valign="top" colspan="4" rowspan="1"><bold><italic>POST-MORTEM</italic> STUDIES</bold></td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Rijkers et al., <xref rid="B84" ref-type="bibr">2007</xref></td><td align="left" valign="top" rowspan="1" colspan="1">T2w</td><td align="left" valign="top" rowspan="1" colspan="1">Visualization of the pulvinar, the lateral and medial geniculate bodies, cerebral peduncle, habenulointerpeduncular tract, periaquaductal gray, the medial lemniscus, the spinothalamic tract, the mammillothalamic tract, and the superior colliculus.</td><td align="center" valign="top" rowspan="1" colspan="1">16:18</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Soria et al., <xref rid="B91" ref-type="bibr">2011</xref></td><td align="left" valign="top" rowspan="1" colspan="1">T1w</td><td align="left" valign="top" rowspan="1" colspan="1">Visibility of SN and RN</td><td align="center" valign="top" rowspan="1" colspan="1">19</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Massey et al., <xref rid="B72" ref-type="bibr">2012</xref></td><td align="left" valign="top" rowspan="1" colspan="1">T2w</td><td align="left" valign="top" rowspan="1" colspan="1">Hypointense band between SN and STN</td><td align="center" valign="top" rowspan="1" colspan="1">21</td></tr><tr><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">High detailed visibility of STN and surrounding</td><td rowspan="1" colspan="1"/></tr><tr><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">Intensity differences between anteromedial and posterolateral part of STN</td><td rowspan="1" colspan="1"/></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">T2w</td><td align="left" valign="top" rowspan="1" colspan="1">Fibers of the subthalamic fasciculus</td><td align="center" valign="top" rowspan="1" colspan="1">20</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Foroutan et al., <xref rid="B39" ref-type="bibr">2013</xref></td><td align="left" valign="top" rowspan="1" colspan="1">FLASH GE</td><td align="left" valign="top" rowspan="1" colspan="1">High-detail images of SN, RN, putamen, and a clear separation of the GP into its external and internal part.</td><td align="center" valign="top" rowspan="1" colspan="1">22</td></tr></tbody></table><table-wrap-foot><p>The last column refers to the line of Table <xref ref-type="table" rid="T3">3</xref> that gives more details about the scan protocols used. FLASH, fast low angle shot; GE, gradient echo; GP, globus pallidus; RN, red nucleus; SN, substantia nigra; STN, subthalamic nucleus; SWI, susceptibility-weighted imaging; T1w, T1-weighted; T2w, T2-weighted; T2<sup></sup>w, T2<sup></sup>-weighted; TDI, track density imaging; Vim, ventral intermediate nucleus; χ-map, susceptibility map.</p></table-wrap-foot></table-wrap><p>In 1999, the basal ganglia were visualized at ultra-high field (8T) using a two-dimensional (2D) multi-slice GE sequence, where high-resolution (195 × 195 μm in-plane) T2<sup></sup>-weighted axial images of one volunteer were obtained in 13 min (Table <xref ref-type="table" rid="T3">3</xref>-1) (Bourekas et al., <xref rid="B10" ref-type="bibr">1999</xref>). On these images the globus pallidus (GP), SN and red nucleus (RN) appeared as hypointense regions. These findings were later confirmed in sagittally recorded slices with similar acquisition parameters (Table <xref ref-type="table" rid="T3">3</xref>-2) (Novak et al., <xref rid="B78" ref-type="bibr">2001</xref>). In 2003, the same group showed that on GE phase images (Table <xref ref-type="table" rid="T3">3</xref>-3), within the SN, the SN pars dorsalis and SN pars lateralis had a higher signal intensity than the matrix of the SN, and within the RN, the medullary lamella showed a higher signal intensity than the RN pars oralis (Abduljalil et al., <xref rid="B1" ref-type="bibr">2003</xref>). A few years later, again the SN and RN appeared hypointense on 7T axial, sagittal, and coronal GE images (Table <xref ref-type="table" rid="T3">3</xref>-4) (Cho et al., <xref rid="B17" ref-type="bibr">2008</xref>) and in 2010, 7T coronal GE images (Table <xref ref-type="table" rid="T3">3</xref>-38) were obtained on which the STN and SN could be well distinguished (Cho et al., <xref rid="B18" ref-type="bibr">2010</xref>).</p><p>A more detailed description of the visualization of the basal ganglia at 7T with three different scanning sequences, exploiting T1-weighted, T2-weighted and susceptibility-weighted imaging, was published in 2010 (Table <xref ref-type="table" rid="T3">3</xref>-5:8) (Abosch et al., <xref rid="B2" ref-type="bibr">2010</xref>). Using SWI, a clear delineation of the STN and the boundary dividing it from the SN were visualized in both axial and coronal planes (Figure <xref ref-type="fig" rid="F1">1A</xref>). Also, SWI allowed visualization of varying levels of contrast within the RN and two of the laminae within the GP (lamina pallidi medialis and incompleta), thus also distinguishing between the GPi and the GPe. Within the thalamus, it showed intensity variations corresponding to the locations of the ventral intermediate nucleus (Vim), the anterior and medial boundaries of the pulvinar, and the boundary of the nucleus ventralis caudalis as identified with the Schaltenbrand and Wahren atlas (Schaltenbrand et al., <xref rid="B89" ref-type="bibr">1977</xref>).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Examples of structures identified at ultra-high field. (A)</bold> Adopted with permission from Abosch et al. (<xref rid="B2" ref-type="bibr">2010</xref>). Ultra-high field (7T) susceptibility-weighted axial and coronal images show a clearly delineated subthalamic nucleus (STN), a boundary between the STN and substantia nigra, and heterogeneous signal intensity in the red nucleus. <bold>(B)</bold> Adopted with permission from Deistung et al. (<xref rid="B27" ref-type="bibr">2013b</xref>). Axial 7T susceptibility map displaying (a) the head of the caudate nucleus, (b) anterior limb of the internal capsule, (c) putamen, (d) external capsule, (e) anterior commissure, (f) external globus pallidus, (g) lamina pallidi medialis, (h) pallidum mediale externum, (i) lamina pallidi incompleta, (j) pallidum mediale internum, (k) posterior limb of internal capsule, (l) subthalamic nucleus, and (m) red nucleus. <bold>(C)</bold> Adopted with permission from Deistung et al. (<xref rid="B27" ref-type="bibr">2013b</xref>). Ultra-high field (7T) susceptibility maps of inferior (C,E) and superior (H,J) sections of the thalamus. (E,J) show overlays of substructures of the thalamus according to the Schaltenbrand et al. (<xref rid="B89" ref-type="bibr">1977</xref>) on the images shown in (C,H) respectively. The pulvinar (Pu.l) can be distinguished from (C,E) and the dorsomedial nucleus (M) and dorsal nuclei group (D.o and D.im) can be seen in (H,J).</p></caption><graphic xlink:href="fnhum-08-00876-g0001"/></fig><p>In 2011, Eapen et al., imaged several deep-brain structures with two different sequences at 7T: T2- and T2<sup></sup>-weighted gradient and spin-echo (GRASE) and T2<sup></sup>-weighted GE (Table <xref ref-type="table" rid="T3">3</xref>-9:10) (Eapen et al., <xref rid="B36" ref-type="bibr">2011</xref>). Both GRASE and GE scans showed a clear distinction between the densely and the poorly vascularized regions of the RN, but only the GE scan also showed signal intensity differences within the SN, possibly representing the SN pars compacta and SN pars reticulata. In two later studies, susceptibility maps were investigated. Using a multi-echo GE sequence (Table <xref ref-type="table" rid="T3">3</xref>-11), a boundary between the STN and the SN was shown (Schafer et al., <xref rid="B87" ref-type="bibr">2012</xref>). The use of susceptibility maps generated from three single-echo GE phase data sets with different head positions (Table <xref ref-type="table" rid="T3">3</xref>-12) also facilitated detailed visualization of structures (Deistung et al., <xref rid="B27" ref-type="bibr">2013b</xref>). It provided discrimination between the subnuclei within the SN, and allowed for accurate discrimination of the STN from the SN and surrounding gray matter and white matter. Furthermore, within the GP, these maps showed the lamina pallidi medialis and lamina pallidi incompleta (Figure <xref ref-type="fig" rid="F1">1B</xref>). The RN displayed substructures in the susceptibility maps, facilitating identification of the medullary lamella, and the RN pars oralis and RN pars dorsomedialis showed a significantly increased susceptibility, compared to the RN pars caudalis. Finally, within the thalamus clear intensity variations were observed on these susceptibility maps corresponding to the Vim, pulvinar, lateral and medial geniculate nucleus, dorsomedial nucleus, and dorsal nuclei group as identified with the Schaltenbrand and Wahren atlas (Schaltenbrand et al., <xref rid="B89" ref-type="bibr">1977</xref>) (Figure <xref ref-type="fig" rid="F1">1C</xref>).</p><p>In two other studies by Kerl et al., investigating the STN and ZI with different sequences at 7T (Kerl et al., <xref rid="B59" ref-type="bibr">2012c</xref>, <xref rid="B56" ref-type="bibr">2013</xref>), a distinction between the STN and the SN and ZI and a clear boundary dividing the rostral ZI from the internal capsule, STN and the pallidofugal fibers could be seen on T2<sup></sup>-weighted images and the latter also on coronal SW images.</p><p>Finally, two studies employed DWI properties to identify substructures within the DBS related structures. In one study, DWI (Table <xref ref-type="table" rid="T3">3</xref>-14) was used to estimate the pathways between seven regions of interest: caudate nucleus, putamen, GPe, GPi, SN, STN, and thalamus (Lenglet et al., <xref rid="B63" ref-type="bibr">2012</xref>). Seven pathways could be successfully identified: the nigrostriatal, nigropallidal, nigrothalamic, subthalamopallidal, pallidothalamic, striatopallidal, and thalamostriatal pathway. These projections were also used to create subparcellations of the SN, possibly corresponding to the SN pars reticulata and SN pars compacta; subdivisions of the STN into a dorsolateral and ventromedial part; subdivisions of the GPe into medial, lateral and rostro-ventral parts; subdivisions of the GPi into laterocaudal, rostral, and mid portions; and many subdivisions within the thalamus. In another study, 7T DWI (Table <xref ref-type="table" rid="T3">3</xref>-15) was used to construct track-density images of the thalamus (Calamante et al., <xref rid="B15" ref-type="bibr">2012</xref>). These showed high-resolution (200 μm isotropic) substructures within the thalamus with clear intensity differences, not only related to track-density, but also to the directionality of the fibers.</p></sec><sec><title>Visualization of deep-brain structures at ultra-high field <italic>ex vivo</italic></title><p>When scanning <italic>ex vivo</italic>, even higher resolution and higher SNR can be obtained due to the possibility of longer scan times and less movement artifacts. Although fixed tissue may suffer from altered tissue properties, such as decreases in T1 and T2 (Tovi and Ericsson, <xref rid="B94" ref-type="bibr">1992</xref>) and a decreased diffusion coefficient (D'Arceuil et al., <xref rid="B23" ref-type="bibr">2007</xref>), which is especially challenging for DWI, it also has great advantages over <italic>in vivo</italic> MRI. Several studies employed <italic>ex-vivo</italic> imaging for investigating the deep-brain structures at ultra-high field (Table <xref ref-type="table" rid="T4">4</xref>).</p><p>In 2007, the STN and its surroundings were explored at 9.4T with a T2-weighted sequence (Table <xref ref-type="table" rid="T3">3</xref>-16:18) in a <italic>post-mortem</italic> brain sample (Rijkers et al., <xref rid="B84" ref-type="bibr">2007</xref>). Acquiring a high in-plane resolution of 100 × 100 μm, not only the most prominent structures of the basal ganglia were visualized, but also the pulvinar, the lateral and medial geniculate bodies, cerebral peduncle, habenulointerpeduncular tract (fasciculus retroflexus), periaquaductal gray, the medial lemniscus, the spinothalamic tract, the mammillothalamic tract, and the superior colliculus.</p><p>Three <italic>post-mortem</italic> brain stems have also been imaged at 7T for 119 min, acquiring 150 × 150 μm images. On these T1-weighted images (Table <xref ref-type="table" rid="T3">3</xref>-19), the RN and SN, which displayed heterogeneous signal intensity, could be visualized (Soria et al., <xref rid="B91" ref-type="bibr">2011</xref>). Even higher in-plane resolutions of 44 × 44 and 88 × 88 μm (Table <xref ref-type="table" rid="T3">3</xref>-20:21) were achieved in a different study after scanning <italic>post-mortem</italic> brain samples for 72 and 10 h respectively (Massey et al., <xref rid="B72" ref-type="bibr">2012</xref>). The obtained T2-weighted images facilitated visualization of the STN, SN, RN, ZI, and thalamus but also allowed a highly detailed identification of many smaller structures surrounding the STN. Furthermore, a hypointense signal band was seen between the SN and STN facilitating easy separation of the two structures. Also the anteromedial part of the STN was relatively hypointense compared to the posterolateral portion, which might be related to the subdivision of the STN in a limbic, associative and sensorimotor part. On the 44 × 44 μm resolution images even the fibers of the subthalamic fasciculus were visualized accurately.</p><p>Finally, one study that focused on differences in T2 and T2<sup></sup> values and iron content between <italic>post-mortem</italic> brains of progressive supranuclear palsy patients and controls, showed high-resolution (50 μm isotropic) fast low-angle shot (FLASH) GE images (Table <xref ref-type="table" rid="T3">3</xref>-22), displaying with much detail the SN, RN, putamen, and the GP with a clear separation into the GPe and GPi (Foroutan et al., <xref rid="B39" ref-type="bibr">2013</xref>).</p><p>These studies show that ultra-high field MRI can aid substantially in the identification of small (sub)structures including the separation between the STN and SN and the laminae within the GP both <italic>ex vivo</italic> and <italic>in vivo</italic>.</p></sec><sec><title>Comparison between sequences for ultra-high field imaging</title><p>In addition to the qualitative description of the visibility of deep-brain structures with ultra-high field MRI, comparisons between different sequences and image reconstruction methods have been made (see Table <xref ref-type="table" rid="T5">5</xref>).</p><table-wrap id="T5" position="float"><label>Table 5</label><caption><p><bold>Overview of comparative studies at ultra-high field</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="left" valign="top" rowspan="1" colspan="1"><bold>Study</bold></th><th align="left" valign="top" rowspan="1" colspan="1"><bold>Sequences</bold></th><th align="left" valign="top" rowspan="1" colspan="1"><bold>Line</bold></th><th align="left" valign="top" rowspan="1" colspan="1"><bold>Measure</bold></th><th align="left" valign="top" rowspan="1" colspan="1"><bold>Findings</bold></th></tr></thead><tbody><tr><td align="left" valign="top" rowspan="1" colspan="1">Abduljalil et al., <xref rid="B1" ref-type="bibr">2003</xref></td><td align="left" valign="top" rowspan="1" colspan="1">GE magnitude</td><td align="left" valign="top" rowspan="1" colspan="1">3</td><td align="left" valign="top" rowspan="1" colspan="1">Qualitative</td><td align="left" valign="top" rowspan="2" colspan="1">Phase images show additional structures to magnitude images</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">GE SWI</td><td align="left" valign="top" rowspan="1" colspan="1">3</td><td rowspan="1" colspan="1"/></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">GE phase</td><td align="left" valign="top" rowspan="1" colspan="1">3</td><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">Magnitude + Phase ≥ SWI</td></tr><tr><td align="left" valign="top" rowspan="2" colspan="1">Wharton and Bowtell, <xref rid="B98" ref-type="bibr">2010</xref></td><td align="left" valign="top" rowspan="1" colspan="1">MO χ-map</td><td align="left" valign="top" rowspan="1" colspan="1">23</td><td align="left" valign="top" rowspan="1" colspan="1">Artifacts and Δχ</td><td align="left" valign="top" rowspan="2" colspan="1">Least noise related artifact and most accurate Δχ in MO</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">RSO χ-map</td><td align="left" valign="top" rowspan="1" colspan="1">23</td><td rowspan="1" colspan="1"/></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">TSO χ-map</td><td align="left" valign="top" rowspan="1" colspan="1">23</td><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">MO≈RSO≈TSO</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Abosch et al., <xref rid="B2" ref-type="bibr">2010</xref></td><td align="left" valign="top" rowspan="1" colspan="1">T1w</td><td align="left" valign="top" rowspan="1" colspan="1">5</td><td align="left" valign="top" rowspan="1" colspan="1">Qualitative</td><td align="left" valign="top" rowspan="1" colspan="1">SWI > T2w > T1w</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">T2w</td><td align="left" valign="top" rowspan="1" colspan="1">6</td><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">SWI</td><td align="left" valign="top" rowspan="1" colspan="1">7:8</td><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Eapen et al., <xref rid="B36" ref-type="bibr">2011</xref></td><td align="left" valign="top" rowspan="1" colspan="1">T2w + T2<sup></sup>w</td><td align="left" valign="top" rowspan="1" colspan="1">9</td><td align="left" valign="top" rowspan="1" colspan="1">CNR of RN/VTA</td><td align="left" valign="top" rowspan="1" colspan="1">T2w + T2<sup></sup>w > T2<sup></sup>w</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">T2<sup></sup>w</td><td align="left" valign="top" rowspan="1" colspan="1">10</td><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Schafer et al., <xref rid="B87" ref-type="bibr">2012</xref></td><td align="left" valign="top" rowspan="1" colspan="1">T2<sup></sup>w</td><td align="left" valign="top" rowspan="1" colspan="1">11</td><td align="left" valign="top" rowspan="1" colspan="1">CNR</td><td align="left" valign="top" rowspan="1" colspan="1">χ-map > T2<sup></sup>w > T2<sup></sup>-map</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">T2<sup></sup>-maps</td><td align="left" valign="top" rowspan="1" colspan="1">11</td><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">χ-map</td><td align="left" valign="top" rowspan="1" colspan="1">11</td><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="2" colspan="1">Kerl et al., <xref rid="B59" ref-type="bibr">2012c</xref>, <xref rid="B56" ref-type="bibr">2013</xref></td><td align="left" valign="top" rowspan="1" colspan="1">T1w</td><td align="left" valign="top" rowspan="1" colspan="1">24</td><td align="left" valign="top" rowspan="1" colspan="1">SNR STN</td><td align="left" valign="top" rowspan="1" colspan="1">T2<sup></sup>w<sup>‡</sup> > T1w<sup>‡</sup> > SWI-MIP<sup>‡</sup> > SWI cor<sup>‡</sup> > T2w</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">T2w</td><td align="left" valign="top" rowspan="1" colspan="1">25</td><td align="left" valign="top" rowspan="1" colspan="1">CNR STN</td><td align="left" valign="top" rowspan="1" colspan="1">T2<sup></sup>w<sup>‡</sup> > SWI-MIP<sup>‡</sup> > T2 > SWI > T1w</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">T2<sup></sup>w</td><td align="left" valign="top" rowspan="1" colspan="1">26:28</td><td align="left" valign="top" rowspan="1" colspan="1">SNR rZI</td><td align="left" valign="top" rowspan="1" colspan="1">T2<sup></sup>w<sup>‡</sup> > SWI-MIP<sup>‡</sup>>T1<sup>‡</sup>>SWI>T2w</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">SWI</td><td align="left" valign="top" rowspan="1" colspan="1">29</td><td align="left" valign="top" rowspan="1" colspan="1">CNR rZI</td><td align="left" valign="top" rowspan="1" colspan="1">T2<sup></sup>w<sup>‡</sup> > SWI-MIP<sup>‡</sup> >T2>SWI>T1w</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">SWI-MIP</td><td align="left" valign="top" rowspan="1" colspan="1">29</td><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="2" colspan="1">Deistung et al., <xref rid="B27" ref-type="bibr">2013b</xref></td><td align="left" valign="top" rowspan="1" colspan="1">GE magnitude</td><td align="left" valign="top" rowspan="1" colspan="1">12</td><td align="left" valign="top" rowspan="1" colspan="1">Qualitative</td><td align="left" valign="top" rowspan="1" colspan="1">χ-map showed most detail</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">GE phase</td><td align="left" valign="top" rowspan="1" colspan="1">12</td><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">χ-map</td><td align="left" valign="top" rowspan="1" colspan="1">12</td><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">R2<sup></sup>-map</td><td align="left" valign="top" rowspan="1" colspan="1">13</td><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="2" colspan="1">Deistung et al., <xref rid="B26" ref-type="bibr">2013a</xref></td><td align="left" valign="top" rowspan="1" colspan="1">T2w</td><td align="left" valign="top" rowspan="1" colspan="1">30</td><td align="left" valign="top" rowspan="1" colspan="1">CNR SN</td><td align="left" valign="top" rowspan="1" colspan="1">χ-map > R2<sup></sup>-map > T2w > R1-map</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">R1-map</td><td align="left" valign="top" rowspan="1" colspan="1">31</td><td align="left" valign="top" rowspan="1" colspan="1">CNR RN</td><td align="left" valign="top" rowspan="1" colspan="1">χ-map > R2<sup></sup>-map > T2w > R1-map</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">R2<sup></sup>-map</td><td align="left" valign="top" rowspan="1" colspan="1">32</td><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">χ-map</td><td align="left" valign="top" rowspan="1" colspan="1">32</td><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/></tr></tbody></table><table-wrap-foot><p>The third column refers to the line of Table <xref ref-type="table" rid="T3">3</xref> that gives more details about the scan protocols used. Sequences that give significantly better results than T2-weighted images are denoted with a double dagger (<sup>‡</sup>). CNR, contrast-to-noise ratio; cor, coronal; GE, gradient echo; MIP, minimum intensity projection; MO, multi-orientation; RN, red nucleus; RSO, regularized single-orientation; rZI, rostral part of zona incerta; SNR, signal-to-noise ratio; STN, subthalamic nucleus; SWI, susceptibility-weighted imaging; T1w, T1-weighted; T2w, T2-weighted; T2<sup></sup>w, T2<sup></sup>-weighted; TSO, threshold based single orientation; VTA, ventral tegmental area; χ-map, susceptibility map.</p></table-wrap-foot></table-wrap><p>In a previously mentioned study from 2003, magnitude, phase-weighted magnitude (SWI), and phase images of a GE dataset (Table <xref ref-type="table" rid="T3">3</xref>-3), were compared for their capability to visualize (sub)structures (Abduljalil et al., <xref rid="B1" ref-type="bibr">2003</xref>). On magnitude images the SN and RN showed up hypointense and on phase images, substructures within the SN could be distinguished as well. The combined magnitude and phase images added little extra to the magnitude and phase images separately.</p><p>Later, in 2010, the mean susceptibility difference (Δχ) between compartments in an agar phantom, and between white matter and deep-brain structures of healthy subjects were compared among three different susceptibility mapping methods applied to GE FLASH images acquired at 7T (Table <xref ref-type="table" rid="T3">3</xref>-23) (Wharton and Bowtell, <xref rid="B98" ref-type="bibr">2010</xref>). The mapping methods consisted of (a) a multi-orientation method using images acquired with differing head positions, (b) a regularized single-orientation method, and (c) a threshold-based single-orientation method. Although all three methods showed large Δχ in the GP, SN, RN, internal capsule, putamen and caudate nucleus, the multi-orientation method resulted in the least noise related artifacts and good estimation of Δχ values in the phantom.</p><p>In another 2010 study, T1-weighted, T2-weighted, and SW imaging (Table <xref ref-type="table" rid="T3">3</xref>-5:8) were compared (Abosch et al., <xref rid="B2" ref-type="bibr">2010</xref>). Most structures were identified in the SW images (see Table <xref ref-type="table" rid="T4">4</xref>), followed by the T2-weighted images (Figure <xref ref-type="fig" rid="F2">2</xref>). The T1-weighted images showed no obvious structures. Eapen et al., also quantitatively compared their T2 + T2<sup></sup>- and T2<sup></sup>-weighted images (Table <xref ref-type="table" rid="T3">3</xref>-9:10) (Eapen et al., <xref rid="B36" ref-type="bibr">2011</xref>). No difference between both sequences could be found in the CNR between the SN and ventral tegmental area (VTA) and between the SN and RN, but in the T2 + T2<sup></sup>-weighted images, the CNR between RN and VTA was significantly better than in the T2<sup></sup>-weighted images.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Ultra-high field (7T) T1-weighted (A,D,G), T2-weighted (B,E,H), and susceptibility-weighted (C,F,I) images at different levels.</bold> Adopted with permission from Abosch et al. (<xref rid="B2" ref-type="bibr">2010</xref>). The susceptibility-weighted images show the highest detail followed by the T2-weighted images.</p></caption><graphic xlink:href="fnhum-08-00876-g0002"/></fig><p>In 2012, again differently reconstructed images derived from a multi-echo GE sequence (Table <xref ref-type="table" rid="T3">3</xref>-11) were compared, consisting of T2<sup></sup>-weighted magnitude images, T2<sup></sup>-maps, and susceptibility maps (Schafer et al., <xref rid="B87" ref-type="bibr">2012</xref>). In most subjects, the CNR between the SN and STN was highest in the susceptibility maps, suggesting that these are most suitable for differentiating the STN from the SN. The SNR of the STN and the rostral part of the ZI (rZI) and the CNR between these structures and white matter, imaged with different sequences, were investigated in two recent studies that compared T1-weighted GE, T2-weighted TSE, T2<sup></sup>-weighted FLASH and SW images (Table <xref ref-type="table" rid="T3">3</xref>-24:29) (Kerl et al., <xref rid="B59" ref-type="bibr">2012c</xref>, <xref rid="B56" ref-type="bibr">2013</xref>). Furthermore, minimum intensity projections (MIPs) of the SW images were computed. After adjusting the SNR and CNR for differences in voxel size, they were highest on the T2<sup></sup>-weighted images for both structures. Furthermore, the SNRs of both structures on the T2<sup></sup>-weighted, T1-weighted, SWI-MIP (and for the STN also on the coronal SW images) were significantly higher than those of the T2-weighted images. The CNRs of both structures on the T2<sup></sup>-weighted and for the rZI also on the SWI-MIP images were also significantly higher than on the T2-weighted images. Also, a 2013 study compared image reconstruction techniques at 7T consisting of (a) magnitude, (b) frequency, and (c) susceptibility maps derived from GE scans (Table <xref ref-type="table" rid="T3">3</xref>-12), and (d) R<sup></sup><sub>2</sub> maps derived from multi-echo GE scans (Table <xref ref-type="table" rid="T3">3</xref>-13) (Deistung et al., <xref rid="B27" ref-type="bibr">2013b</xref>). Qualitative analysis by a neuroanatomist revealed that susceptibility maps in general facilitated the most detailed visualization of structures. Finally, in a recent study by the same group, the CNR between several brain stem structures and their surroundings were compared between sequences (Table <xref ref-type="table" rid="T3">3</xref>-30:32) (Deistung et al., <xref rid="B26" ref-type="bibr">2013a</xref>). For the RN and the SN, the CNR of the R2<sup></sup>-map and the susceptibility map outperformed those of the R1-map and the T2-weighted image.</p><p>Although comparison between studies is difficult due to the differences in scanning conditions, the majority of these studies show that sequences that are sensitive to magnetic susceptibility such as SWI and T2<sup></sup> related images are most suitable for targeting basal ganglia structures and their subdivisions in DBS at ultra-high field.</p></sec><sec><title>Comparison between field strengths</title><p>In addition to comparisons between different sequences, some studies compared similar sequences between different field strengths (see Table <xref ref-type="table" rid="T6">6</xref>). In a 2008 study, the difference between a 7T GE image (Table <xref ref-type="table" rid="T3">3</xref>-4) and a 1.5 T image was briefly treated (Cho et al., <xref rid="B17" ref-type="bibr">2008</xref>). Visual inspection showed that the 7T image displayed better contrast, SNR and resolution. However, comparison is difficult because the acquisition parameters of the 1.5T image were unfortunately not provided. In the same year, T2<sup></sup>-weighted GE images were investigated, acquired at several echo times at three different field strengths: 1.5T, 3T, and 7T (Table <xref ref-type="table" rid="T3">3</xref>-33:35) (Yao et al., <xref rid="B104" ref-type="bibr">2009</xref>). This showed that increasing field strength resulted in a higher influence of iron on the value of R2<sup></sup>, making this contrast useful for iron-rich deep-brain structures, such as the GP, RN, SN, and putamen (Hallgren and Sourander, <xref rid="B46" ref-type="bibr">1958</xref>). A thorough quantitative investigation of the visibility of the STN related to field strength was performed in 2010 (Cho et al., <xref rid="B18" ref-type="bibr">2010</xref>), comparing the contrast between the STN and a baseline (containing the ZI and thalamus), the contrast between the STN and SN, the SNR in gray matter areas, and the slope of signal increase between STN and baseline among 1.5T, 3T, and 7T T2<sup></sup>-weighted GE images (Table <xref ref-type="table" rid="T3">3</xref>-36:38). At higher field strengths, the STN, SN, putamen, GPi, and GPe could be visualized while the boundaries of these structures were unclear on the 1.5T images (Figure <xref ref-type="fig" rid="F3">3</xref>). Furthermore, all quantitative measures increased with field strength, and the SNR and contrast were significantly improved at 7T compared to 1.5 and 3T. Finally, the two studies by Kerl et al., investigating the STN and rZI at 7T (Kerl et al., <xref rid="B59" ref-type="bibr">2012c</xref>, <xref rid="B56" ref-type="bibr">2013</xref>) were additionally performed at 3T. Again, they compared the SNR and CNR of these structures between different sequences: T1-weighted MPRAGE, T2-weighted fluid attenuated inversion recovery (FLAIR), T2-weighted sampling perfection with application of optimized contrasts using different flip angle evolutions (SPACE), two T2<sup></sup>-weighted 2D FLASH (FLASH2D) sequences, and SW images and their MIPs (Table <xref ref-type="table" rid="T3">3</xref>-39:46) (Kerl et al., <xref rid="B58" ref-type="bibr">2012a</xref>,<xref rid="B57" ref-type="bibr">b</xref>). This makes it possible to compare the SNRs and CNRs of the different studies between field strengths, when adjusted for voxel size, although it should be noted that for the T1- and T2-weighted images different sequences were used between field strengths. For both structures, the SNRs of the T2<sup></sup>-weighted, SWI-MIP and SW images of the 7T images were higher than those of the 3T images, but the SNRs of the 3T T2-weighted SPACE image and T1-weighted images were higher at 3T than at 7T. However, the CNRs of both structures were substantially higher on all the 7T sequences than on the corresponding 3T sequences.</p><table-wrap id="T6" position="float"><label>Table 6</label><caption><p><bold>Overview of studies that compare scan protocols between field strengths</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="left" valign="top" rowspan="1" colspan="1"><bold>Study</bold></th><th align="left" valign="top" rowspan="1" colspan="1"><bold>Sequence</bold></th><th align="left" valign="top" rowspan="1" colspan="1"><bold>Line</bold></th><th align="left" valign="top" rowspan="1" colspan="1"><bold>Measure</bold></th><th align="left" valign="top" rowspan="1" colspan="1"><bold>Findings</bold></th></tr></thead><tbody><tr><td align="left" valign="top" rowspan="1" colspan="1">Cho et al., <xref rid="B17" ref-type="bibr">2008</xref></td><td align="left" valign="top" rowspan="1" colspan="1">1. 1.5T</td><td align="center" valign="top" rowspan="1" colspan="1">4</td><td align="left" valign="top" rowspan="1" colspan="1">Qualitative</td><td align="left" valign="top" rowspan="2" colspan="1">7T has better contrast, SNR and resolution than 1.5T</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">2. 7T T2<sup></sup>w</td><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Yao et al., <xref rid="B104" ref-type="bibr">2009</xref></td><td align="left" valign="top" rowspan="1" colspan="1">1. 1.5T T2<sup></sup>w</td><td align="center" valign="top" rowspan="1" colspan="1">33</td><td align="left" valign="top" rowspan="1" colspan="1">R2<sup></sup></td><td align="left" valign="top" rowspan="2" colspan="1">R2<sup></sup> becomes more sensitive to iron with increasing field strength</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">2. 3T T2<sup></sup>w</td><td align="center" valign="top" rowspan="1" colspan="1">34</td><td rowspan="1" colspan="1"/></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">3. 7T T2<sup></sup>w</td><td align="center" valign="top" rowspan="1" colspan="1">35</td><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Cho et al., <xref rid="B18" ref-type="bibr">2010</xref></td><td align="left" valign="top" rowspan="1" colspan="1">1. 1.5T T2<sup></sup>w</td><td align="center" valign="top" rowspan="1" colspan="1">36</td><td align="left" valign="top" rowspan="1" colspan="1">Contrast</td><td align="left" valign="top" rowspan="1" colspan="1">7T<sup>‡</sup>>3T>1.5T</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">2. 3T T2<sup></sup>w</td><td align="center" valign="top" rowspan="1" colspan="1">37</td><td align="left" valign="top" rowspan="1" colspan="1">Slope of signal increase</td><td align="left" valign="top" rowspan="1" colspan="1">7T>3T>1.5T</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">3. 7T T2<sup></sup>w</td><td align="center" valign="top" rowspan="1" colspan="1">38</td><td align="left" valign="top" rowspan="1" colspan="1">SNR</td><td align="left" valign="top" rowspan="1" colspan="1">7T<sup>‡</sup>>3T>1.5T</td></tr><tr><td align="left" valign="top" rowspan="1" colspan="1">Kerl et al., <xref rid="B58" ref-type="bibr">2012a</xref>,<xref rid="B57" ref-type="bibr">b</xref>,<xref rid="B59" ref-type="bibr">c</xref>, <xref rid="B56" ref-type="bibr">2013</xref></td><td align="left" valign="top" rowspan="1" colspan="1">1. 3T T1w</td><td align="center" valign="top" rowspan="1" colspan="1">39</td><td align="left" valign="top" rowspan="1" colspan="1">SNR</td><td align="left" valign="top" rowspan="1" colspan="1">3T T1w > 7T T1w</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">2. 3T T2w FLAIR</td><td align="center" valign="top" rowspan="1" colspan="1">40</td><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">7T T2<sup></sup>w > 3T T2<sup></sup>w</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">3. 3T T2w SPACE</td><td align="center" valign="top" rowspan="1" colspan="1">41</td><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">7T SWI-MIP > 3T SWI-MIP axial</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">4. 3T T2<sup></sup>w</td><td align="center" valign="top" rowspan="1" colspan="1">41:45</td><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">3T T2w SPACE > 7T T2w > 3T T2w FLAIR</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">5. 3T SWI</td><td align="center" valign="top" rowspan="1" colspan="1">46</td><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">7T SWI > 3T SWI</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">6. 3T SWI-MIP</td><td align="center" valign="top" rowspan="1" colspan="1">46</td><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">7. 7T T1w</td><td align="center" valign="top" rowspan="1" colspan="1">24</td><td align="left" valign="top" rowspan="1" colspan="1">CNR</td><td align="left" valign="top" rowspan="1" colspan="1">7T T2<sup></sup>w > 3T T2<sup></sup>w</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">8. 7T T2w TSE</td><td align="center" valign="top" rowspan="1" colspan="1">25</td><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">7T SWI-MIP > 3T SWI-MIP</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">9. 7T T2<sup></sup>w</td><td align="center" valign="top" rowspan="1" colspan="1">26:28</td><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">7T T2w > 3T T2w SPACE > 3T T2w FLAIR</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">10. 7T SWI</td><td align="center" valign="top" rowspan="1" colspan="1">29</td><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">7T SWI > 3T SWI</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">11. 7T SWI-MIP</td><td align="center" valign="top" rowspan="1" colspan="1">29</td><td rowspan="1" colspan="1"/><td align="left" valign="top" rowspan="1" colspan="1">7T T1 > 3T T1</td></tr></tbody></table><table-wrap-foot><p>The third column refers to the line of Table <xref ref-type="table" rid="T3">3</xref> that gives more details about the scan protocols used. Sequences that significantly improve imaging at 7T compared to 1.5T and 3T are denoted with a double dagger (‡). CNR, contrast-to-noise ratio; FLAIR, fluid attenuated inversion recovery; MIP, minimum intensity projection; rZI, rostral part of zona incerta; SNR, signal-to-noise ratio; SPACE, sampling perfection with application of optimized contrasts using different flip angle evolutions; STN, subthalamic nucleus; SWI, susceptibility-weighted imaging; T1w, T1-weighted; T2w, T2-weighted; T2<sup></sup>w, T2<sup></sup>-weighted; TSE, turbo spin-echo.</p></table-wrap-foot></table-wrap><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Coronal T2<sup></sup>-weighted images obtained at 7.0T (A), 3.0T (B), and 1.5T (C).</bold> Adapted with permission from Cho et al. (<xref rid="B18" ref-type="bibr">2010</xref>). Visual inspection shows clearer identification of the substantia nigra (SN), subthalamic nucleus (STN), internal globus pallidus (GPi), external globus pallidus (GPe), and putamen (Pu) at 7T compared to 3T and 1.5T.</p></caption><graphic xlink:href="fnhum-08-00876-g0003"/></fig><p>These studies suggest that 7T MRI can better facilitate accurate targeting of deep brain structures than 1.5T or 3T MRI.</p></sec></sec><sec sec-type="discussion" id="s2"><title>Discussion</title><p>Accurate visualization of deep-brain structures is important to improve our understanding of their anatomy, connectivity and function, and for improved surgical targeting for DBS in movement and psychiatric disorders. To date, targeting based on direct visualization of DBS targets with T2-weighted 1.5T or 3T MRI can be difficult. However, studies at ultra-high field showed good visibility of these structures on SW images based on T2<sup></sup> and phase contrast. Structures that have been identified at ultra-high field include: a separation between the STN and SN (Abosch et al., <xref rid="B2" ref-type="bibr">2010</xref>; Cho et al., <xref rid="B18" ref-type="bibr">2010</xref>; Massey et al., <xref rid="B72" ref-type="bibr">2012</xref>; Schafer et al., <xref rid="B87" ref-type="bibr">2012</xref>; Deistung et al., <xref rid="B27" ref-type="bibr">2013b</xref>); the lamina pallidi medialis and lamina pallidi incompleta within the GP (Abosch et al., <xref rid="B2" ref-type="bibr">2010</xref>; Deistung et al., <xref rid="B27" ref-type="bibr">2013b</xref>); a subdivision of the STN in two halves (Lenglet et al., <xref rid="B63" ref-type="bibr">2012</xref>; Massey et al., <xref rid="B72" ref-type="bibr">2012</xref>); subdivisions of the SN possibly representing the SN pars reticulata and SN pars compacta (Eapen et al., <xref rid="B36" ref-type="bibr">2011</xref>; Lenglet et al., <xref rid="B63" ref-type="bibr">2012</xref>; Deistung et al., <xref rid="B27" ref-type="bibr">2013b</xref>); substructures in the RN (Abosch et al., <xref rid="B2" ref-type="bibr">2010</xref>; Eapen et al., <xref rid="B36" ref-type="bibr">2011</xref>) including the medullary lamella (Abduljalil et al., <xref rid="B1" ref-type="bibr">2003</xref>; Deistung et al., <xref rid="B27" ref-type="bibr">2013b</xref>), RN pars oralis (Abduljalil et al., <xref rid="B1" ref-type="bibr">2003</xref>), and RN pars caudalis (Deistung et al., <xref rid="B27" ref-type="bibr">2013b</xref>); and several regions in the thalamus (Lenglet et al., <xref rid="B63" ref-type="bibr">2012</xref>) including the Vim (Abosch et al., <xref rid="B2" ref-type="bibr">2010</xref>; Deistung et al., <xref rid="B27" ref-type="bibr">2013b</xref>), the pulvinar (Deistung et al., <xref rid="B27" ref-type="bibr">2013b</xref>) and its anterior and medial boundaries (Abosch et al., <xref rid="B2" ref-type="bibr">2010</xref>), the boundary of the nucleus ventralis caudalis (Abosch et al., <xref rid="B2" ref-type="bibr">2010</xref>), the lateral and medial geniculate nucleus (Deistung et al., <xref rid="B27" ref-type="bibr">2013b</xref>), the dorsomedial nucleus (Deistung et al., <xref rid="B27" ref-type="bibr">2013b</xref>) and the dorsal nuclei group (Deistung et al., <xref rid="B27" ref-type="bibr">2013b</xref>). Furthermore, 7T T2<sup></sup>-weighted and SW images have displayed improved CNR, SNR and resolution in the deep-brain regions, compared to 1.5T and 3T images (Cho et al., <xref rid="B18" ref-type="bibr">2010</xref>; Kerl et al., <xref rid="B58" ref-type="bibr">2012a</xref>,<xref rid="B57" ref-type="bibr">b</xref>,<xref rid="B59" ref-type="bibr">c</xref>, <xref rid="B56" ref-type="bibr">2013</xref>).</p><p>Based on a descriptive evaluation of different MR images, more and smaller structures can be identified on T2<sup></sup>-weighted, GE phase, SW images, and susceptibility and R2<sup></sup> maps than on T1- and T2-weighted images (Abduljalil et al., <xref rid="B1" ref-type="bibr">2003</xref>; Abosch et al., <xref rid="B2" ref-type="bibr">2010</xref>; Kerl et al., <xref rid="B59" ref-type="bibr">2012c</xref>, <xref rid="B56" ref-type="bibr">2013</xref>; Deistung et al., <xref rid="B27" ref-type="bibr">2013b</xref>). Although quantitative comparison between studies is difficult due to variations in scan protocols, the CNRs of deep-brain structures on T2<sup></sup> and SW images and corresponding maps are generally higher than those of T2- and T1-weighted images (Kerl et al., <xref rid="B59" ref-type="bibr">2012c</xref>, <xref rid="B56" ref-type="bibr">2013</xref>). For the SNR, the same trend can be seen, although T1-weighted images seem to have a higher SNR than SW images (Kerl et al., <xref rid="B59" ref-type="bibr">2012c</xref>, <xref rid="B56" ref-type="bibr">2013</xref>).</p><sec><title>Perspectives</title><p>The improved visualization of the basal ganglia with ultra-high field MRI discussed here provides good perspectives for clinical practice. The clear delineation of DBS target structures and their possible subdivisions may aid in more accurate targeting, which may reduce negative side effects and shorten surgery duration, or it may even allow surgery under general anesthesia. Furthermore, ultra-high field MRI also shows potential for more accurate diagnosis and monitoring of basal ganglia diseases due to, for example, improved identification of the SN pars compacta and SN pars reticulata, which may in its turn facilitate improved patient specific treatments.</p><p>In addition, ultra-high field MRI promises to be a versatile tool in clinically oriented research of the deep brain nuclei. It might help us to improve our current understanding of the functionality of the healthy basal ganglia and its disease processes with high resolution functional MRI and connectivity analyses.</p></sec><sec><title>Recommendations</title><p>When in the end considering the optimal scan protocol for visualizing the DBS targets for clinical purposes at ultra-high field, both image quality and practical requirements need to be taken into account. In terms of hardware, it is recommended to use a head coil with a high number of receive channels (i.e., 16 or higher). This has been shown to improve the SNR (de Zwart et al., <xref rid="B31" ref-type="bibr">2004</xref>; Wiggins et al., <xref rid="B101" ref-type="bibr">2006</xref>) which is also reflected from the studies described in Table <xref ref-type="table" rid="T3">3</xref>. In terms of scan protocol, based on the described literature, we recommend to use a 3D multi-echo GE sequence with an isotropic resolution of 0.5 mm<sup>3</sup> and partial brain coverage. The 3D sequence facilitates small and isotropic voxel sizes, which ensures good resolution in every plane which is important for distinguishing the STN from the SN. From the multi-echo GE scan, both T2<sup></sup>-weighted and susceptibility weighted images as well as T2<sup></sup>-maps, R2<sup></sup>-maps, and susceptibility maps can be computed, which were shown in the reviewed literature to display best basal ganglia visibility. Since the basal ganglia are located within the same axial oblique slab of approximately 4–5 cm thickness, we advise to shorten scan time by covering only this part of the brain. If more time reduction is required, partial Fourier imaging, elliptical k-space coverage, or parallel imaging can be considered as well.</p><p>To support these guidelines, Figure <xref ref-type="fig" rid="F4">4</xref> shows an example of a T2<sup></sup>-weighted image and an R2<sup></sup>-map created with these recommendations. The images were obtained by scanning a healthy volunteer on a 7T MR scanner (Magnetom 7T, Siemens, Erlangen, Germany) at Scannexus (Maastricht, The Netherlands) using a 32-channel phased-array coil (Nova Medical, Wilmington, United States) with a multi-echo 3D GE sequence. Scan time was reduced to 12 min and 23 s by partial brain coverage and 75% partial Fourier imaging (other scanning parameters can be found in Table <xref ref-type="table" rid="T3">3</xref>–line 47:48). On these 0.5 mm<sup>3</sup> isotropic resolution images, the STN can be distinguished from the SN in the coronal plane, and the three laminae of the GP can be identified.</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>Ultra-high field (7T) axial (A,B,E,F) and coronal (C,D,G,H) T2<sup></sup>-weighted images (A–D) and R2<sup></sup>-maps (E–H).</bold> Panels <bold>(B,D,F,H)</bold> show the anatomical structures that can be identified with the Schaltenbrand and Wahren atlas (Schaltenbrand and Wahren, <xref rid="B88" ref-type="bibr">2005</xref>): (a) caudate nucleus, (b) anterior limb of internal capsule, (c) putamen, (d) lamina pallidi lateralis, (e) external globus pallidus, (f) lamina pallidi medialis, (g) pallidum mediale externum, (h) lamina pallidi incompleta, (i) pallidum mediale internum, (j) inferior thalamic peduncle, (k) anterior commissure, (l) prothalamus, (m) fornix, (n) third ventricle, (o) hypothalamus, (p) posterior limb of internal capsule, (q) subthalamic nucleus, (r) red nucleus, (s) substantia nigra, (t) internal globus pallidus. (Courtesy D. Ivanov).</p></caption><graphic xlink:href="fnhum-08-00876-g0004"/></fig><p>When planning a DBS surgery, the MR images are often registered to CT images, resulting in images that both display the stereotactic frame from the CT image as well as contrast within the brain. This registration may be more reliable, however if a whole brain MR image is available as an intermediate step. Abosch et al. (<xref rid="B2" ref-type="bibr">2010</xref>) showed that it is already possible to perform 1 mm<sup>3</sup> whole brain T1-weighted imaging in 3.5 min, which may be a good candidate for coregistration.</p></sec><sec><title>Limitations</title><p>Despite these promising results concerning accurate and high-resolution visualization of the small deep brain (sub)structures, several issues still need to be addressed before they can routinely be employed in direct targeting for DBS.</p><p>Firstly, ultra-high field images have an increased risk of geometrical distortions compared to 1.5T images. The severity of these distortions at 7T in deep-brain regions has been investigated in several studies. One study compared the coordinates of marker points in a phantom imaged with 1.5T and 7T MRI to their locations on computed tomography (CT) images (Dammann et al., <xref rid="B22" ref-type="bibr">2011</xref>). The maximum distortion in either x-, y-, or z-direction at 7T was 1.6 mm, which was slightly larger than at 1.5T (0.9 mm). Furthermore, the fewest distortions were observed in the center of the phantom. In another study the distortions in an anthropomorphic phantom between T2<sup></sup>-weighted 7T MR and CT images were investigated, revealing a maximum deviation of 0.78 mm (Cho et al., <xref rid="B18" ref-type="bibr">2010</xref>). Finally, registration of 7T T1- and T2-weighted images of the midbrain of PD patients to 1.5T T1- and T2-weighted images showed that mainly rigid body transformations were required and that scaling and skew deformations were small (Duchin et al., <xref rid="B33" ref-type="bibr">2012</xref>). Furthermore, the midbrain region, containing many DBS targets, required the least correction. Quantitative comparison showed that the distances of the T2-weighted images were significantly less than 1 mm suggesting that affine registration of T1- and T2-weighted 7T images to CT images can already provide MR images with midbrain distortions comparable to those of 1.5T images. These few studies suggest that at 7T images can be acquired with distortions smaller than 1 mm in the deep-brain areas.</p><p>Secondly, some of the mentioned imaging techniques pose additional challenges in the clinical context. Most studies were performed on young and healthy volunteers. In patients, movement during image acquisition can be less controlled, counteracting the gain in SNR and spatial specificity obtained with ultra-high field. However, newer techniques, such as prospective motion correction might remedy this problem (Maclaren et al., <xref rid="B68" ref-type="bibr">2013</xref>). This approach monitors movement in the scanner with high accuracy and corrects the new image acquisition adaptively according to the new head position. That is, even with large head movements—as observed in many patients—the resulting images are already coregistered and movement artifact free.</p><p>In addition, the availability of ultra-high field MR scanners is currently limited. Firstly, the number of scanners that have been installed in the world is limited itself (see Table <xref ref-type="table" rid="T2">2</xref>), which is inherent to its high cost in purchase and in operation. Secondly, due to the novel status of ultra-high field MRI, safety precautions regarding metallic objects are often more strict than on 3T systems and the use of ultra-high field MRI is currently only allowed for research purposes.</p><p>Finally, direct targeting in DBS suffers from brain shift, intra-operative deformation of the brain compared to preoperative MR images due to difference in head position and cerebrospinal fluid loss. Without compensation for this, it will eventually still limit targeting accuracy. However, this effect is independent of the magnetic field strength and even the pre-operative imaging modality.</p></sec></sec><sec sec-type="conclusion" id="s3"><title>Conclusion</title><p>Ultra-high field MRI can reliably and accurately display subdivisions within the basal ganglia and related structures, which especially benefits from T2<sup>*</sup>- and phase-related contrasts. If the limitations concerning image distortions and the availability of the scanners are solved, these technical advances have the potential to improve accuracy of targeting in DBS surgery and the clinical outcome.</p><sec><title>Conflict of interest statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec></sec>
Do intentions for action penetrate visual experience?	Could not extract abstract	<contrib contrib-type="author"><name><surname>Briscoe</surname><given-names>Robert E.</given-names></name><xref ref-type="author-notes" rid="fn001"><sup>*</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/122406"/></contrib>	Frontiers in Psychology	<p>A now-famous study by Aglioti et al. (<xref rid="B1" ref-type="bibr">1995</xref>) involves a graspable version of the Ebbinghaus illusion (Figure <xref ref-type="fig" rid="F1">1</xref>). Aglioti and colleagues constructed a 3D version of the illusion, using thin solid disks. Subjects were asked to pick up the central disk on the left if the two central disks appeared identical in size, and to pick up the central disk on the right if they appeared different in size. The experimenters varied the relative sizes of the two target disks randomly so that in some trials physically different disks appeared perceptually identical in size, while in other trials physically identical disks appeared perceptually different in size. In selecting a disk in either trial condition, Milner and Goodale observe, “subjects indicated their susceptibility to the visual illusion” (<xref rid="B4" ref-type="bibr">1995/2006</xref>, p. 168): that is, their <italic>choice</italic> of which disk to pick up was determined by its apparent size rather than its real one. Nonetheless, the effect of the illusion was significantly less pronounced with respect to action, as measured by maximum grip aperture (MGA) in prehension, than with respect to conscious perceptual estimation (PE), as measured by the distance between thumb and forefinger in a manual estimate of disk size. Although the disk surrounded by small circles in the illusion display typically <italic>looks</italic> about 10% larger than the disk surrounded by large circles, the increase in MGA when reaching for the former disk exhibited a magnitude of around only 6%.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>The Ebbinghaus Illusion</bold>. The disk on the left looks typically looks about 10% smaller than the disk on the right.</p></caption><graphic xlink:href="fpsyg-05-01265-g0001"/></fig><p>According to proponents of the dual systems model of visual processing (Milner and Goodale, <xref rid="B4" ref-type="bibr">1995/2006</xref>), the illusion has a different effect on visual awareness than on visually guided grasping because the former makes use of different sources of visuospatial information than the latter. On this model, how the size of an object appears in conscious vision should not influence grip aperture, and, conversely, how the size of the object is represented by motor systems that guide grasping should not influence representation of its size in conscious vision.</p><p>At variance with this idea, however, Vishton et al. (<xref rid="B6" ref-type="bibr">2007</xref>) (experiment 3) found that the act of reaching for a disk in a 3D version of Ebbinghaus illusion significantly diminished the magnitude of the effect on subsequent PE for several minutes after reaching trials had ended (5.74% for PE vs. 6.10% for grasping). Strikingly, they also found (experiment 2) that when subjects were <italic>merely informed</italic> prior to engaging in PE trials that they would subsequently be required to grasp the disk that appeared larger, the effect of the illusion on PE was significantly diminished (6.18% for PE vs. 5.54% for grasping). “Simply listening to a description of a reaching task,” Vishton and co-authors write, “seems to affect size perception” (Vishton et al., <xref rid="B6" ref-type="bibr">2007</xref>, p. 718).</p><p>These findings suggest that the phenomenal contents of visual experience can be cognitively penetrated: high-level information originating outside of the visual system seems to modulate the way an object's size visually appears. There are different possible mechanisms whereby such penetration might occur. Vishton and co-authors propose that “intending to reach for a target changes how the reacher perceives it” and that “action choice changes the nature of visual size perception” (p. 718). But how does action selection have this effect? One possibility (a) is that an abstract, high-level intention to act—either a “distal” or “proximal” intention in the sense of Pacherie (<xref rid="B5" ref-type="bibr">2008</xref>)—somehow exerts a direct influence on PE, say, by changing the relative weightings assigned by the visual system to sources of depth information such as binocular disparity, vergence, accommodation, and relative size. Since size estimation depends, in part, on perceived distance in depth, this could explain the influence of intention on perception. A second possibility (b) is that the relevant effect is brought about via lower-level motor representations that implement and provide kinematic and dynamical specification for the subject's high-level intention. This would arguably still count as a case of cognitive penetration if the lower-level, action-specifying motor representations carried information from the subject's high-level intention that influenced relative cue weighting or other visual computations. As Wu (<xref rid="B7" ref-type="bibr">2013</xref>) writes, “The key [to cognitive penetration of vision by intention] is not directness of link but (internal) informational transfer of an appropriate kind” (p. 662). A third possibility (c) looks to motor imagery elicited in the course of both experiments for the source of penetration. Possibility (c), however, is not entirely distinct from (a) and (b), since there is evidence that internally rehearsing the performance of an action activates representations at all levels in the motor processing hierarchy (for reviews, see Decety and Grèzes, <xref rid="B2" ref-type="bibr">2006</xref>; Jeannerod, <xref rid="B3" ref-type="bibr">2006</xref>). A final possibility (d) is that the effect is not due to motor representations at all, but rather to the subject's <italic>beliefs</italic> concerning the action that she has been requested to perform.<xref ref-type="fn" rid="fn0001"><sup>1</sup></xref> Future studies will have to investigate which, if any, of these four explanations best accounts for the intriguing effects that Vishton and his co-authors have reported.</p><sec><title>Conflict of interest statement</title><p>The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec>
Mapping Epileptic Activity: Sources or Networks for the Clinicians?	<p>Epileptic seizures of focal origin are classically considered to arise from a focal epileptogenic zone and then spread to other brain regions. This is a key concept for semiological electro-clinical correlations, localization of relevant structural lesions, and selection of patients for epilepsy surgery. Recent development in neuro-imaging and electro-physiology and combinations, thereof, have been validated as contributory tools for focus localization. In parallel, these techniques have revealed that widespread networks of brain regions, rather than a single epileptogenic region, are implicated in focal epileptic activity. Sophisticated multimodal imaging and analysis strategies of brain connectivity patterns have been developed to characterize the spatio-temporal relationships within these networks by combining the strength of both techniques to optimize spatial and temporal resolution with whole-brain coverage and directional connectivity. In this paper, we review the potential clinical contribution of these functional mapping techniques as well as invasive electrophysiology in human beings and animal models for characterizing network connectivity.</p>	<contrib contrib-type="author"><name><surname>Pittau</surname><given-names>Francesca</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/82145"/></contrib><contrib contrib-type="author"><name><surname>Mégevand</surname><given-names>Pierre</given-names></name><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/49881"/></contrib><contrib contrib-type="author"><name><surname>Sheybani</surname><given-names>Laurent</given-names></name><xref ref-type="aff" rid="aff3"><sup>3</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/183924"/></contrib><contrib contrib-type="author"><name><surname>Abela</surname><given-names>Eugenio</given-names></name><xref ref-type="aff" rid="aff4"><sup>4</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/81716"/></contrib><contrib contrib-type="author"><name><surname>Grouiller</surname><given-names>Frédéric</given-names></name><xref ref-type="aff" rid="aff5"><sup>5</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/110817"/></contrib><contrib contrib-type="author"><name><surname>Spinelli</surname><given-names>Laurent</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/189745"/></contrib><contrib contrib-type="author"><name><surname>Michel</surname><given-names>Christoph M.</given-names></name><xref ref-type="aff" rid="aff3"><sup>3</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/3353"/></contrib><contrib contrib-type="author"><name><surname>Seeck</surname><given-names>Margitta</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/189724"/></contrib><contrib contrib-type="author"><name><surname>Vulliemoz</surname><given-names>Serge</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="corresp" rid="cor1">*</xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/109772"/></contrib>	Frontiers in Neurology	<sec sec-type="introduction" id="S1"><title>Introduction</title><p>Epilepsy is one of the most frequent chronic neurological disorder, with an incidence of 50/100,000/year and a prevalence of 0.5–1% (<xref rid="B1" ref-type="bibr">1</xref>, <xref rid="B2" ref-type="bibr">2</xref>). One third of these patients are drug resistant (<xref rid="B3" ref-type="bibr">3</xref>). Focal seizures are classically considered to be caused by an abnormal neuro-electrical activity of a focal epileptogenic zone and a subsequent spreading to other brain regions. This concept is intimately linked to the correlation between ictal signs and symptoms, electro-physiological activity, and structural lesion [anatomo-electro-clinical correlation (<xref rid="B4" ref-type="bibr">4</xref>)]. Furthermore, this hypothesis is crucial to select drug-resistant focal epilepsy patients for surgery, a widely accepted effective therapy (<xref rid="B5" ref-type="bibr">5</xref>, <xref rid="B6" ref-type="bibr">6</xref>). The aim of epilepsy surgery is to remove the epileptogenic zone with the preservation of the eloquent areas (<xref rid="B7" ref-type="bibr">7</xref>).</p><p>Recent progress in neuro-imaging and electro-physiology suggests that focal seizures and focal epilepsies are actually related to an abnormal function of a network of cortical and subcortical brain structures rather than to a single epileptogenic region (<xref rid="B8" ref-type="bibr">8</xref>–<xref rid="B14" ref-type="bibr">14</xref>). The occurrence of epileptic activity is due to the abnormal neuronal activity of these connected regions and abnormal interactions between them (epileptic network). This conceptual shift is reflected in the new terminology proposal for seizures and epilepsies of the International League against Epilepsy, which proposes “<italic>focal</italic>” as indicating seizures arising primarily “within networks limited to one hemisphere and that may be discrete or more widely distributed” (<xref rid="B15" ref-type="bibr">15</xref>). Generalized seizures are considered as “originating within and rapidly engaging, bilaterally distributed networks” of cortical and subcortical regions. Inside these networks, some brain regions are responsible for seizure initiation and propagation, whereas other nodes are more remotely involved, their activity modulating, or being modulated by the epileptic discharge.</p><p>There is increasing evidence that epileptic activity strongly interacts with physiological brain networks, notably the so-called “resting-state networks” (RSNs) (<xref rid="B8" ref-type="bibr">8</xref>, <xref rid="B16" ref-type="bibr">16</xref>). A RSN is a set of brain regions that shows temporal correlations in their activity (as measured by hemodynamic or electrical signals) and that are functionally related. They are observed during rest but correspond to the networks revealed in different behavioral and cognitive task (e.g., attention, vision, etc.). This has led to the new concept that the apparently resting spontaneous brain activity shows continuous interaction among brain networks responsible for various classes of sensory/behavioral functions (<xref rid="B17" ref-type="bibr">17</xref>). RSNs are highly organized in space, reproducible from subject to subject, and differ with aging and between genders (<xref rid="B18" ref-type="bibr">18</xref>).</p><p>In this paper, we review the converging evidence from different brain mapping techniques in human and animal models that epilepsy is related to the dysfunction of a large-scale brain networks, with alterations of physiological brain networks. We will particularly focus on the clinical impact of this new view of epilepsy as a network disease.</p></sec><sec sec-type="methods" id="S2"><title>Methods</title><p>An electronic literature search was conducted for articles on this topic regarding human and animal subjects. Sources searched included PubMed and relevant books. Words used in the search included the text words and subject headings of epilep, functional connect, resting-state functional network*, temporal epilepsy, extra-temporal epilepsy, electroencephalogram (or EEG), simultaneous functional MRI (fMRI) and EEG (or EEG-fMRI), electric and magnetic source imaging (or MSI, ESI), intracranial EEG (or iEEG or sEEG), cortico-cortical evoked potential, and single-pulse electrical stimulation. The words were searched independently and in combination. For each citation considered, the abstract was read (when available), and articles were excluded if they were outside the scope of the review. Studies published only in abstract form, letters, and technical reports were excluded. The bibliography of each of the retrieved papers was examined to identify relevant references that could have been missed by electronic search. The findings were described taking into account the limit of words and the critical insight of the authors.</p></sec><sec id="S3"><title>How to Measure Resting-State Networks? Functional Connectivity</title><p>Functional interactions between brain regions activity, can be characterized in several ways. On the one hand, functional connectivity (FC), the most widely used metrics, measures the statistical dependency between different signals obtained by correlation analysis. However, such strategy does not account for the direction of the information flow and cannot therefore infer causality relationships. On the other hand, effective or directed connectivity investigates directional relationships and aims at describing causal influences. Effective connectivity can be investigated using model-driven techniques such as structural equation modeling (<xref rid="B19" ref-type="bibr">19</xref>) and dynamic causal modeling (DCM) (<xref rid="B20" ref-type="bibr">20</xref>), data-driven techniques such as Granger-causal modeling (<xref rid="B21" ref-type="bibr">21</xref>), or by recording the response of remote areas to focal stimulation of a given brain region [cortico-cortical evoked potentials (<xref rid="B22" ref-type="bibr">22</xref>)]. Connectivity studies can be applied among a set of predefined relevant brain regions selected by the investigator, between one seed region and the rest of the brain or at the whole-brain scale, using the spatial resolution of the recording technique. A detailed description of the various approaches used for measuring connectivity is beyond the scope of this review and the reader is referred to studies comparing various approaches to better understand the specific limitations of each technique (<xref rid="B23" ref-type="bibr">23</xref>–<xref rid="B25" ref-type="bibr">25</xref>). The results obtained by such connectivity analysis between all pairs of brain regions can be represented in so-called connectivity matrices. Graph topological analysis is then increasingly applied to reduce the complexity of the data and extract meaningful characteristics of the networks (<xref rid="B26" ref-type="bibr">26</xref>).</p><sec id="S3-1"><title>Blood oxygen level dependent signal and physiological resting-state networks</title><p>The concept of brain networks originated, and has largely benefited, from the use of resting-state fMRI. fMRI detects blood oxygen level dependent (BOLD) signal change reflecting metabolically active brain areas not only in relation to a specific physiologic or pathologic event (<xref rid="B27" ref-type="bibr">27</xref>) but also in resting-state (RS) condition (resting-state-fMRI or RS-fMRI).</p><p>Biswal and colleagues demonstrated for the first time (1995) that brain regions that are functionally related, show temporal correlations in the low frequency component of the BOLD signal. In other words, fMRI FC detects zones that exhibit correlated BOLD fluctuations and, as a result, belong to the same functional network (<xref rid="B28" ref-type="bibr">28</xref>). Studies in monkeys (<xref rid="B29" ref-type="bibr">29</xref>) and in human beings (<xref rid="B30" ref-type="bibr">30</xref>) suggest that FC is related to neuronal processes.</p><p>Functional connectivity can be measured while the subject is performing a behavioral and cognitive task (task-related FC), or while the subject is not performing any specific task (RS-FC). The RSN that is mainly activated in condition of resting wakefulness and deactivated in task performing is called default-mode network (DMN) (<xref rid="B31" ref-type="bibr">31</xref>). This physiological RSN is involved in self-referential thoughts and consciousness (<xref rid="B32" ref-type="bibr">32</xref>, <xref rid="B33" ref-type="bibr">33</xref>). The concept of “resting” is debatable. Usually, subjects are instructed to lie down in the scanner with the eyes closed, and are invited to not sleep.</p><p>Different methods have been developed to extract RSNs, some requiring an “<italic>a priori</italic> hypothesis,” like seed-based approach (<xref rid="B34" ref-type="bibr">34</xref>), other do not [i.e., independent component analysis (<xref rid="B35" ref-type="bibr">35</xref>), or bootstrap analysis (<xref rid="B36" ref-type="bibr">36</xref>)]. The description of the methodological aspects is outside the scope of this review. Other papers can be consulted (<xref rid="B14" ref-type="bibr">14</xref>, <xref rid="B37" ref-type="bibr">37</xref>, <xref rid="B38" ref-type="bibr">38</xref>).</p></sec><sec id="S3-2"><title>EEG/MEG and physiological resting-state networks</title><p>Functional connectivity algorithms similar to those used for fMRI BOLD signals can be applied to MEG or EEG current-density estimations in the source space, revealing brain areas that are synchronized in specific frequency bands. As with fMRI, such analysis can be applied to task-related (<xref rid="B39" ref-type="bibr">39</xref>), as well as to spontaneous resting-state activity (<xref rid="B40" ref-type="bibr">40</xref>, <xref rid="B41" ref-type="bibr">41</xref>). The unique advantage of EEG/MEG connectivity analysis is the high temporal resolution that allows studying fast fluctuations within large-scale network interactions and fast switches between resting-state networks.</p><p>FC analysis of EEG/MEG considers the time-course of electro-magnetic signals and looks at correlations of oscillating networks (<xref rid="B42" ref-type="bibr">42</xref>). Beyond this view of temporal oscillations, EEG recordings can be considered as time-series of scalp potential maps that vary across time with the temporal resolution in the order of milliseconds (<xref rid="B43" ref-type="bibr">43</xref>). Several studies have shown that spontaneous EEG signals can be explained by the alternation of periods of stable topography, lasting almost 100 ms, very reproducible across subjects, and modifiable by neurological (<xref rid="B44" ref-type="bibr">44</xref>) or psychiatric impairment (<xref rid="B45" ref-type="bibr">45</xref>). These periods are called microstates and can be identified throughout an individual’s life (<xref rid="B46" ref-type="bibr">46</xref>) suggesting that they might be mediated by predetermined anatomical connections. During rest, four different microstates are consistently observed, and they can be considered as “basic building blocks” of spontaneous mental activities (<xref rid="B47" ref-type="bibr">47</xref>). A recent review on this topic is available (<xref rid="B48" ref-type="bibr">48</xref>).</p><p>It has been shown (<xref rid="B49" ref-type="bibr">49</xref>) that the temporal dynamic of EEG microstates have hemodynamic correlates that can be measured with EEG-fMRI and that each physiological microstate map corresponds to one of the well-described BOLD RS network. Such clear correlates between EEG and BOLD are less well found when looking at classical power fluctuations in specific EEG frequency bands (<xref rid="B50" ref-type="bibr">50</xref>). This finding strongly suggests that the EEG microstates can be the candidates for the electro-physiological signatures of fMRI RSNs. Scale-invariance of the alternation between microstates has been demonstrated to be the base of this coupling over such a wide temporal scale (<xref rid="B51" ref-type="bibr">51</xref>).</p></sec></sec><sec id="S4"><title>Evidence for Brain Networks Involved in Epileptic Activity</title><p>As described above, FC at the whole-brain level can be studied with EEG, MEG, fMRI, iEEG, or the combination of these techniques. They have been applied to patients with focal or generalized epilepsy to characterize spatial and temporal properties of epileptic networks.</p><sec id="S4-3"><title>EEG and MEG-based connectivity in epilepsy</title><p>EEG and MEG are appealing non-invasive techniques for estimating brain connectivity in epilepsy because they measure neuro-electrical activity more directly than fMRI and can potentially offer a higher temporal resolution.</p><p>Several studies using concordance with intracranial recordings or post-operative outcome have established that electric and magnetic source imaging (ESI, MSI) are reliable techniques for estimating the localization of the cortical generators of epileptic activity (<xref rid="B52" ref-type="bibr">52</xref>–<xref rid="B55" ref-type="bibr">55</xref>) and these techniques now offer a much more convincing strategy to investigate connectivity directly between the activity of cortical regions. Therefore, both ESI and MSI studies will be discussed together hereunder. Studies using connectivity analysis in the sensor space are not discussed here because of their severe limitations of interpretation due to important caveats related to sensor cross-talk, volume conduction, and reference choice of the electromagnetic signals (<xref rid="B56" ref-type="bibr">56</xref>). The projection of the signal in source space requires the selection of a head model describing the propagation of the electromagnetic signal (forward problem) and an inverse solution (estimating the cortical activity from the EEG/MEG recording, inverse problem) (<xref rid="B48" ref-type="bibr">48</xref>, <xref rid="B57" ref-type="bibr">57</xref>, <xref rid="B58" ref-type="bibr">58</xref>). A variety of head models exists, from template averaged normal brain to highly sophisticated realistic models based on individual anatomy, and they have been used in epilepsy imaging and cognitive neurosciences. Validation in patients with invasive EEG or surgical resection showed that the individual anatomy was important for the localization accuracy (<xref rid="B54" ref-type="bibr">54</xref>), but that the performance of highly sophisticated models did not outperform less computer-intensive models also based on individual anatomy, as these were disturbed by the presence of brain lesions in patients with epilepsy (<xref rid="B59" ref-type="bibr">59</xref>). Regarding inverse solutions, dipole models consider a single or a few equivalent dipole(s) as sources of the EEG/MEG signals of sources distributed in the whole cortex (<xref rid="B48" ref-type="bibr">48</xref>). While both approaches might yield complimentary results for localizing epileptic sources (<xref rid="B60" ref-type="bibr">60</xref>), distributed sources are best suited to the study of connectivity between cortical patches at a large brain scale.</p><p>The analysis of interictal epileptic discharges has principally aimed at localizing epileptic generators in the context of pre-surgical evaluations rather than studying large brain networks. Case reports or small MEG series showed promising results for the localizing value of the regions with high information outflow, estimated by connectivity analysis (<xref rid="B61" ref-type="bibr">61</xref>–<xref rid="B63" ref-type="bibr">63</xref>). In addition, based on development in cognitive neurosciences, the background activity measured by MEG and EEG in the classical frequency bands has also been used as a substrate to estimate abnormal connectivity in patients with epilepsy and correlate it with clinical variables. In patients with brain tumors, increased theta-band connectivity and more profound network alterations were associated with a higher number of epileptic seizures (<xref rid="B64" ref-type="bibr">64</xref>) and there is higher post-operative network improvement in patients who become seizure free (<xref rid="B65" ref-type="bibr">65</xref>).</p><p>In generalized epilepsies, connectivity studies have highlighted a network of hyperconnected anterior regions in photosensitive patients (<xref rid="B66" ref-type="bibr">66</xref>). Network analysis using graph theory in five patients with absence epilepsy showed a build-up of connectivity changes occurring before the onset of generalized spike-wave discharges (<xref rid="B67" ref-type="bibr">67</xref>). This shows the potential of such a technique for our understanding of the large-scale brain networks underlying hyperexcitability and interictal to ictal transition. A similar approach has been applied to iEEG recordings of interictal to ictal transition in patients with focal cortical dysplasia (<xref rid="B12" ref-type="bibr">12</xref>).</p><p>Another study used co-occurrence of MEG interictal spikes to build graphs of connectivity between the estimated sources of these spikes. In seven patients also investigated with stereotactic iEEG, the connections revealed by MEG were confirmed by iEEG (<xref rid="B68" ref-type="bibr">68</xref>).</p><p>Similarly to fMRI studies, future work will need to distinguish between transient connectivity alterations related to interictal discharges, that are known to be associated with subtle cognitive impairment (<xref rid="B69" ref-type="bibr">69</xref>), and deeper connectivity changes based on background activity alterations. The tools are now available to benefit from the high temporal resolution of EEG/MEG to further investigate these issues and this field has recently attracted an intense interest. While MEG offers advantages over EEG for longitudinal studies of post-operative cases, due to its insensitivity to skull defects, the development of long-term high-density EEG system, its greater versatility compared to MEG and its potential combination with fMRI will be precious for recording seizures and exploring network changes leading to their initiation, spread, and termination.</p></sec><sec id="S4-4"><title>EEG-fMRI connectivity in epilepsy</title><p>Simultaneous EEG and fMRI (EEG-fMRI) detects hemodynamic changes in the brain related to events of interest identified in the EEG (<xref rid="B70" ref-type="bibr">70</xref>). Combining high temporal resolution of EEG signal with high spatial resolution of BOLD images, EEG-fMRI has been shown to be useful to characterize various forms of focal and generalized epileptic abnormalities (hereunder called “spikes” for practical reasons) (<xref rid="B71" ref-type="bibr">71</xref>). EEG-fMRI helps to localize epileptic focus in patients with drug-resistant focal epilepsy candidate for surgery (<xref rid="B72" ref-type="bibr">72</xref>, <xref rid="B73" ref-type="bibr">73</xref>). From the first publications (<xref rid="B74" ref-type="bibr">74</xref>, <xref rid="B75" ref-type="bibr">75</xref>), EEG-fMRI has demonstrated that BOLD responses to a focal spike can be multifocal, also present at a distance from the presumed focus (Figure <xref ref-type="fig" rid="F1">1</xref>), corroborating the concept of epileptic network (<xref rid="B9" ref-type="bibr">9</xref>). Studying such networks can inform about patients’ prognosis after surgery. While focal responses predict a good post-operative outcome, diffuse results are associated with a poor outcome, probably reflecting that a larger network is involved in the epileptogenic zone (<xref rid="B76" ref-type="bibr">76</xref>, <xref rid="B77" ref-type="bibr">77</xref>). Epileptic activity can also be detected in the absence of spikes and fMRI analysis based on EEG topography can reveal epileptogenic networks (<xref rid="B78" ref-type="bibr">78</xref>).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Interictal network revealed by EEG-fMRI</bold>. Patients with non-lesional right frontal epilepsy. Marked events inside the EEG-fMRI session: right frontal spikes with phase reversal at F4 (on the left: longitudinal bipolar montage from 204 channels EEG). On the middle: scalp voltage map of the spike (204 electrodes, viewed from the top) with the maximal right frontal negativity (blue). BOLD response (<italic>t</italic>-value = 4; <italic>p</italic> < 0.05 corrected for family-wise-error) has maximal activation in the spike topography but other clusters with inferior statistical values are present in the contra-lateral homologous region and in the thalamus.</p></caption><graphic xlink:href="fneur-05-00218-g001"/></fig><p>BOLD responses to a neural event are usually detected with a delay of 4–6 s (<xref rid="B79" ref-type="bibr">79</xref>). Nevertheless, hemodynamic changes to spikes can have different peak times (<xref rid="B80" ref-type="bibr">80</xref>), and can occur before the spike is visible on the scalp (<xref rid="B81" ref-type="bibr">81</xref>). Dynamic analysis of BOLD response (<xref rid="B82" ref-type="bibr">82</xref>, <xref rid="B83" ref-type="bibr">83</xref>) can tell us which brain areas are first activated, by comparing early BOLD response vs. late BOLD response. However, this analysis does not address the concept of causality and the sluggishness and variability of BOLD responses prevent a more accurate investigation into the temporal dynamics and directionality of the connections (<xref rid="B24" ref-type="bibr">24</xref>). Causality within epileptic network can be addressed by effective connectivity approaches like Dynamic Causal Modeling (DCM) (<xref rid="B37" ref-type="bibr">37</xref>, <xref rid="B84" ref-type="bibr">84</xref>).</p><p>The combination of ESI with EEG-fMRI can offer complementary information for improving each single technique (Figure <xref ref-type="fig" rid="F2">2</xref>). Although EEG-fMRI and ESI measure different signals (hemodynamic the first, electrical the second), the concordance between ESI performed during fMRI recordings can allow distinguishing between hemodynamic changes related to spike onset vs. propagation, adding important temporal information to the limited fMRI temporal resolution (<xref rid="B85" ref-type="bibr">85</xref>, <xref rid="B86" ref-type="bibr">86</xref>).</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Techniques using different types of signal are concordant in localizing the epileptic focus in a patient with right orbito-frontal focal cortical dysplasia</bold>. On the top-line: ESI (256 electrodes, simplified realistic head model lSMAC, distributed inverse solution LORETA) performed on right frontal spikes (FP2-F8). On the bottom-line: EEG-fMRI performed on the same type of events recorded inside the scanner.</p></caption><graphic xlink:href="fneur-05-00218-g002"/></fig><p>EEG-fMRI studies can give insights about epileptogenesis. Interictal spikes of different types of epilepsy (frontal, temporal, and posterior quadrant), are associated with deactivation in the precuneus and posterior cingulate cortex (<xref rid="B10" ref-type="bibr">10</xref>), regions involved in the DMN (Figure <xref ref-type="fig" rid="F3">3</xref>). Other physiological RSNs could be affected by spikes: this interaction and its clinical consequences need to be clarified in future studies. A common involvement of the cingulate gyri in temporal lobe and frontal lobe epilepsy was reported (<xref rid="B10" ref-type="bibr">10</xref>), probably resulting from rapid spread of epileptic activity originating from the temporal and frontal areas, which both involve the limbic system.</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Interictal involvement of DMN in focal epilepsy</bold>. Patients with right hemispheric extended periventricular nodular heterotopia. Marked epileptic events inside the scanner: right posterior temporal spikes with phase reversal at P8 (on the left: 256 channel EEG; referential montage Fz-Cz). BOLD increase is concordant with the spike topography (topographic map on the middle), whereas BOLD decrease is present in the regions of default-mode network (DMN).</p></caption><graphic xlink:href="fneur-05-00218-g003"/></fig><p>A specific area, localized in the medial orbito-frontal gyrus (piriform cortex), called “area tempestas”, seems to be involved in the genesis or propagation of epileptic activity (<xref rid="B87" ref-type="bibr">87</xref>, <xref rid="B88" ref-type="bibr">88</xref>) in focal epilepsies. A DCM study supported the hypothesis of a causal link between hemodynamic changes in this structure and a specific type of reflex epilepsy, although in a single patient (<xref rid="B89" ref-type="bibr">89</xref>). Several other findings seem to corroborate the important role of the area tempestas: (i) its decrease in benzodiazepine receptor (<xref rid="B87" ref-type="bibr">87</xref>), (ii) its epileptogenic role in animal kindling models of temporal lobe epilepsy (TLE) (<xref rid="B90" ref-type="bibr">90</xref>–<xref rid="B92" ref-type="bibr">92</xref>), and (iii) its increase in gray matter volume in patient with frontal lobe epilepsy when compared to controls (<xref rid="B93" ref-type="bibr">93</xref>).</p><p>From a methodological point of view, multimodal combination between EEG/ESI, fMRI, and diffusion imaging tractography will allow exploring functional and structural connectivity at a finer spatio-temporal scale. Some initial small studies have highlighted the potential of these combinations (<xref rid="B94" ref-type="bibr">94</xref>–<xref rid="B96" ref-type="bibr">96</xref>).</p><sec id="S4-4-1"><title>fMRI and EEG-fMRI studies in focal epilepsy</title><p>Unfortunately, only few of the many studies on RS-FC have been done with the simultaneous recording of EEG. Spikes cause a transient cognitive impairment, by affecting, e.g., memory retrieval in rats (<xref rid="B97" ref-type="bibr">97</xref>), and memory maintenance and retrieval in human beings (<xref rid="B69" ref-type="bibr">69</xref>). Therefore, a more consistent use of simultaneous EEG while performing fMRI for RS or task-related studies in epileptic patients is needed to determine the influence of spikes on the determined BOLD networks. Indeed a study where EEG activity was monitored during a working memory-fMRI session (<xref rid="B98" ref-type="bibr">98</xref>) has shown that the task-related BOLD dramatically changes when spikes occur. Another advantage of the simultaneous recording of EEG in the scanner is that it allows monitoring the transition between different alertness states in order to assure that the subject is in RS and not in drowsiness/sleep state. A very recent review (<xref rid="B99" ref-type="bibr">99</xref>) has accurately discussed this issue and summarized the relevant studies.</p><sec id="S4-4-1-1"><title>Temporal lobe epilepsy</title><p>The majority of RS-fMRI studies in focal epilepsies have focused on TLE, which is the most common form of focal epilepsy in adults and offers the advantage of being one of the most homogeneous groups within the focal epilepsy syndromes.</p><p>Temporal lobe epilepsy has been the first epileptic syndrome to be considered as epileptogenic network (<xref rid="B100" ref-type="bibr">100</xref>) with relatively well characterized components encompassing different structures in the mesial temporal lobe (amygdala and hippocampus), adjacent cortex including enthorhinal cortex and lateral temporal cortex, and extra-temporal structures (i.e., thalamus and orbito-frontal cortex). fMRI connectivity studies (some with simultaneous EEG recording, others without) conducted by seeding the principal nodes of the mesial temporal network showed impaired connectivity with the other nodes of the network (<xref rid="B101" ref-type="bibr">101</xref>–<xref rid="B103" ref-type="bibr">103</xref>). Decreased connectivity is the most common finding among those studies. Nevertheless, there are reports of increased function of the unaffected hippocampus in patients with unilateral MTLE, both in the RS (<xref rid="B104" ref-type="bibr">104</xref>) and during task-related (<xref rid="B105" ref-type="bibr">105</xref>, <xref rid="B106" ref-type="bibr">106</xref>) acquisitions. Morgan et al. (<xref rid="B107" ref-type="bibr">107</xref>) have shown that RS cross-hippocampal FC is disrupted at the beginning of the disease and then increases linearly with epilepsy duration after 10 years, suggesting that length of disease influences FC patterns. Bettus et al. (<xref rid="B108" ref-type="bibr">108</xref>) studied the electro-physiological correlates of BOLD signal fluctuations in structures exhibiting epileptiform discharges, by measuring correlations between intracerebral EEG and resting-state fMRI in five patients with TLE. They found an increase in connectivity measured from the intracerebral EEG but a decrease of connectivity measured from the BOLD signal in regions with epileptiform abnormalities relative to non-affected areas. This discrepancy, obtained by measuring connectivity of two signals of different nature (electrical and hemodynamic), demonstrates the challenge in interpreting connectivity changes. It could also suggest an alteration of neurovascular coupling in TLE.</p><p>In unilateral mesial TLE (MTLE), the affected amygdala and hippocampus (and to a lesser extent on the contra-lateral side) are less connected between them and also with other consistent RSNs, such as the mesolimbic and the DMN, suggesting that these functional interictal changes explain cognitive and psychiatric impairments often found in patients with this type of epilepsy (<xref rid="B109" ref-type="bibr">109</xref>). Several fMRI studies, with and without EEG, have shown an abnormal FC between physiological consistent RSNs [i.e., language (<xref rid="B110" ref-type="bibr">110</xref>)] and MTL structures.</p><p><italic>Default-mode network</italic>. Laufs and colleagues (<xref rid="B8" ref-type="bibr">8</xref>) have shown that deactivation in DMN, involved in consciousness, is more frequent for spikes in patients with TLE than extra-TLE. Deactivation in the same regions in response to temporal spikes was also demonstrated by Kobayashi et al. (<xref rid="B111" ref-type="bibr">111</xref>) and by Fahoum et al. (<xref rid="B10" ref-type="bibr">10</xref>). Frings et al. (<xref rid="B112" ref-type="bibr">112</xref>) showed decreased DMN-hippocampus FC in MTLE patients compared to controls during an object-location memory task, underlying the importance of the intact connection between these structures to preserve memory. This concept was validated in post-surgical follow-up studies (see below). An impairment of the connections between DMN and MTL structures has been demonstrated also in RS with a seed-based fMRI analysis (<xref rid="B113" ref-type="bibr">113</xref>). The same group (<xref rid="B114" ref-type="bibr">114</xref>) combined fMRI RS-FC and diffusion tensor imaging, and showed that the decreased FC within the DMN in MTLE is correlated to abnormal structural connectivity. Although functional DMN connectivity is generally decreased in MTLE, few nodes can be hyperconnected and this may play a compensatory role for the loss of functional connections in other regions of the network (<xref rid="B115" ref-type="bibr">115</xref>). The same study, performed with an independent component analysis, has also shown distinct patterns of FC impairment with DMN between the left and right MTLE. The same difference has been further reported (<xref rid="B116" ref-type="bibr">116</xref>), suggesting that impaired cognition and memory in TLE may be different in right vs. left TLE. Morgan and colleagues (<xref rid="B117" ref-type="bibr">117</xref>) have identified a region in the ventral lateral nucleus of the right thalamus whose connectivity to the hippocampi separates left from right TLE subjects, suggesting that quantifying resting-state FC across this network may be a potential indicator of lateralization of TLE (useful step in the pre-surgical assessments).</p><p>Functional connectivity findings are related with clinical data: McCormick et al. (<xref rid="B118" ref-type="bibr">118</xref>) shows that MTLE patients with respect to controls have reduced connectivity from the posterior DMN to the epileptogenic hippocampus and increased DMN connectivity to the contra-lateral hippocampus. Stronger DMN connectivity to the epileptogenic hippocampus was associated with better pre-surgical memory and with greater postsurgical memory decline, whereas stronger DMN connectivity to the contra-lateral hippocampus was associated with less postsurgical memory decline. Following surgery, DMN connectivity to the remaining hippocampus increased from pre-surgical values and showed enhanced correlation with postsurgical memory function.</p><p>Hippocampi are considered by some authors as nodes of the DMN (<xref rid="B119" ref-type="bibr">119</xref>), but there is not unanimity on this interpretation (<xref rid="B32" ref-type="bibr">32</xref>, <xref rid="B120" ref-type="bibr">120</xref>, <xref rid="B121" ref-type="bibr">121</xref>). It is important to remember that all the regions of the brain can be functionally connected: a region belonging to a specific network (like the mesial temporal network) can belong also to a less specific network encompassing the previous one. This classification depends on how many different physiological RSNs are extracted from specific analyses: for instance, by extracting four physiological RSNs, the probability that the mesial temporal regions will be included in the DMN is higher than if the number of extracted network is higher (<xref rid="B122" ref-type="bibr">122</xref>).</p><p><italic>Mesolimbic network</italic>. Patients with unilateral MTLE have important decreases of FC with the ventromesial limbic prefrontal regions and with the nucleus accumbens (<xref rid="B109" ref-type="bibr">109</xref>). These regions belong to a dopaminergic mesolimbic network, involved in long-term memory for novel events and reward (<xref rid="B123" ref-type="bibr">123</xref>). Hippocampus and amygdala have been often described as part of this network (<xref rid="B124" ref-type="bibr">124</xref>, <xref rid="B125" ref-type="bibr">125</xref>), and several findings suggest that this network is affected in MTLE. The preferential seizure spread from mesial temporal lobes to mesial frontal lobes, especially the orbito-frontal cortex, has been demonstrated by ictal iEEG in patients with MTLE, suggesting that mesial orbito-frontal cortex is strongly affected by mesial temporal activity (<xref rid="B126" ref-type="bibr">126</xref>). On the other hand, dopaminergic alterations have been demonstrated in the pathophysiology of major depression, and dysfunctional activity of the mesolimbic dopaminergic system plays a crucial role in depressive behavior (<xref rid="B127" ref-type="bibr">127</xref>, <xref rid="B128" ref-type="bibr">128</xref>). Structures belonging to mesolimbic network are functionally (<xref rid="B129" ref-type="bibr">129</xref>) and structurally (<xref rid="B130" ref-type="bibr">130</xref>) impaired in MTLE patients who have psychiatric impairments, such as anxiety/depression. A recent study (<xref rid="B131" ref-type="bibr">131</xref>) showed that the subgenual anterior cingulate cortex (mesolimbic network key-node) is affected only in MTLE patients that have primary affective disorders and not in those without such disorders and neither in controls. The same study confirms these FC findings with voxel-based morphometry and diffusion tensor imaging, corroborating the concept that the affective psychopathology often diagnosed in patients with MTLE has a neurobiological correlate. Antidepressant drugs, when effective, could modulate these connectivity impairments.</p><p>The amygdala has often been described as part of mesolimbic network and it is also involved in emotional processing of stimuli. Facial emotion recognition, particularly for “fear,” is impaired in patients with TLE, especially on the right hemisphere (<xref rid="B132" ref-type="bibr">132</xref>–<xref rid="B135" ref-type="bibr">135</xref>). Broicher and colleagues (<xref rid="B136" ref-type="bibr">136</xref>) showed through fearful-face fMRI-paradigm that the altered amygdala FC in TLE patients is strongly related to the poor performance in behavioral tests evaluating the theory of mind abilities (ability of decoding thoughts and behavior of other human beings). Another study with the same paradigm showed that, in right TLE patients, pre-operative right amygdala activation correlates with post-operative change of anxiety and depression scores [i.e., greater increases in anxiety and depression in patients with greater preoperative activation (<xref rid="B137" ref-type="bibr">137</xref>)]. This confirms that pre-surgical study of FC between TLE and other brain structures can help to predict post-surgery neuropsychological consequences.</p><p><italic>Attention network</italic>. Several studies have shown that dorsal and ventral attention networks are functionally altered in MTLE, explaining why patients with this type of epilepsy have often worse performances than healthy controls (HC) in attention tasks. Cataldi et al. (<xref rid="B138" ref-type="bibr">138</xref>) have recently reviewed this topic.</p></sec><sec id="S4-4-1-2"><title>Extra-temporal lobe epilepsy</title><p>Extra-temporal lobe epilepsies are characterized by a wide range of focus localization and etiology. For this reason, group studies with homogeneous clinical phenotype are difficult to achieve. This contrasts with the large body of group studies in MTLE, which take advantage from the frequent uniform pathology of atrophy and cell-loss in amygdala-hippocampus structures. A group-analysis EEG-fMRI study in different types of epilepsy (frontal, temporal, and posterior quadrant) showed that focal spikes are associated with networks of widespread metabolic changes, specific for each type of epilepsy (<xref rid="B10" ref-type="bibr">10</xref>).</p><p>Negishi et al. (<xref rid="B139" ref-type="bibr">139</xref>) revealed higher lateral pre-surgery FC maps in drug-resistant patients with good surgical outcome (seizure-free at 1-year) compared to those with poor outcome. A recent study on 23 patients with frontal lobe epilepsy used the same seed-FC approach (seed at maximal BOLD value of the spike-related activation map) (<xref rid="B140" ref-type="bibr">140</xref>), finding an increased FC in the neighborhood of the seed and a decrease in regions remote to the seed compared to controls. Patient-specific connectivity pattern was not significantly changed when comparing fMRI runs with spikes vs. fMRI without any spike detectable on the simultaneous EEG. Patients with drug-resistant frontal lobe epilepsy (<xref rid="B141" ref-type="bibr">141</xref>) recruit wider areas compared to controls when performing an fMRI memory encoding task paradigm, particularly in the contra-lateral frontal lobe, suggesting the presence of compensatory mechanisms to maintain memory function.</p></sec></sec><sec id="S4-4-2"><title>Generalized epilepsy</title><p>Different theories have been proposed about the patho-physiology of “generalized seizures”. Meeren et al. (<xref rid="B142" ref-type="bibr">142</xref>) reviewed this topic. All these theories consider cortex and thalamus as being involved in the generation of the typical 2.5–4 Hz generalized spike-wave discharges (GSWD) detected on scalp EEG, but it is still unclear, which of them is the “primum movens” (<xref rid="B143" ref-type="bibr">143</xref>). As discussed below, animal studies in genetic models of absence epilepsy are crucial to gain understanding of these conditions because no invasive validation can be contemplated in human beings. These animal studies suggest that GSWDs are triggered in a restricted cortical region (<xref rid="B144" ref-type="bibr">144</xref>–<xref rid="B147" ref-type="bibr">147</xref>).</p><p>Several EEG-fMRI studies showed that during short GSWD (<xref rid="B16" ref-type="bibr">16</xref>, <xref rid="B148" ref-type="bibr">148</xref>–<xref rid="B150" ref-type="bibr">150</xref>) and absence (<xref rid="B151" ref-type="bibr">151</xref>–<xref rid="B154" ref-type="bibr">154</xref>), there is a characteristic pattern of subcortical (medio-dorsal thalamic and striatus) activation and cortical deactivation, especially in areas of the DMN. It has been hypothesized that the DMN deactivation is linked to reduced consciousness (i.e., absences) (<xref rid="B16" ref-type="bibr">16</xref>, <xref rid="B155" ref-type="bibr">155</xref>, <xref rid="B156" ref-type="bibr">156</xref>). A dynamic analysis study on 17 absences from nine patients with absence epilepsy and classical pattern of 3–4 Hz GSWDs (<xref rid="B83" ref-type="bibr">83</xref>) showed that BOLD responses were remarkably consistent in space and time across different absences of one patient, but were different from patient to patient. Furthermore, this study shows early frontal BOLD activations (specific for each patient), supporting the cortical focus theory. Another EEG-fMRI study on 11 children with absence seizures (<xref rid="B157" ref-type="bibr">157</xref>) revealed that the first brain zone showing BOLD increase was the parietal cortex, this activity starting 10 s before the onset of the discharge. Additionally, this study demonstrated the hemodynamic involvement of subcortical structures in GSWD, including the reticular structures of the brainstem. Focal cortical involvement before the onset of GSWD has been demonstrated also by a magnetoencephalography study in human beings (<xref rid="B158" ref-type="bibr">158</xref>) and a near-infrared spectroscopy study applied on the frontal cortex (<xref rid="B159" ref-type="bibr">159</xref>). An exhaustive review on focal abnormalities in idiopathic generalized epilepsy (IGE) has been recently published (<xref rid="B160" ref-type="bibr">160</xref>). All these findings support the conceptual transition from “primarily generalized epilepsy,” (implying that all brain regions simultaneously would generate GSWD) to seizures “originating within and rapidly engaging, bilaterally distributed networks” of cortical and subcortical regions (<xref rid="B15" ref-type="bibr">15</xref>).</p><p>Concerning the role of subcortical structures, in patients with IGE, it has been shown that both the anterior nucleus of thalamus (ANT) and the centromedian/parafascicular (Cm/Pf) nucleus (which provides diffuse inputs to the cortex) are activated during GSWD; the activity of the cortico-reticular Cm/Pf preceded that of the ANT, suggesting that the Cm/Pf is involved in GSWD initiation or early propagation, while the ANT in its maintenance (<xref rid="B161" ref-type="bibr">161</xref>).</p><p>Recent studies have used fMRI to investigate whether resting-state FC between thalamus, basal ganglia, and DMN areas is altered in patients with IGE, even during GSWD-free periods of brain activity (Figure <xref ref-type="fig" rid="F4">4</xref>). Wang et al. (<xref rid="B162" ref-type="bibr">162</xref>) used ICA to map RSNs in 16 patients with IGE and 16 HC. They found that the DMN had simultaneously reduced FC within the anterior cingulate cortex, but increased connectivity in the precuneus. Moreover, they found widespread connectivity reductions in prefrontal, sensorimotor, and even auditory cortices (<xref rid="B162" ref-type="bibr">162</xref>). Reduced resting-state FC between frontal areas and the rest of the DMN was later confirmed (<xref rid="B163" ref-type="bibr">163</xref>). An important question is whether these changes in DMN connectivity are meaningfully related to clinical information, e.g., disease duration or responsiveness to medication. Of note, in both aforementioned studies, there were significant correlations between RS-FC and disease duration: the reduction in connectivity was inversely correlated to disease duration, indicating that long-standing epilepsy leads to progressive disruption of DMN integration. Interestingly, a study of structural and FC in 26 IGE patients and HC, showed that the degree of coupling between functional and structural connectivity networks is decreased, and exhibited a negative correlation with epilepsy duration in patients (<xref rid="B164" ref-type="bibr">164</xref>).</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>This diagram summarizes the functional connectivity (FC) changes in patients with idiopathic generalized epilepsy compared to healthy controls</bold>. The color map shows the default-mode network (z-scores) derived from independent component analysis or RS-fMRI data overlaid on a standard single-subject anatomy (Montreal Neurological Institute space). Widespread FC reductions were found within the DMN (dashed lines), as well as between anterior DMN and the thalamus. Increased FC related to increased disease duration has been observed between posterior DMN regions and the parahippocampal gyrus (solid line). ACC, anterior cingulate cortex, IPL, inferior parietal lobule, PRE, precuneus, PHG, parahippocampal gyrus, TH, thalamus.</p></caption><graphic xlink:href="fneur-05-00218-g004"/></fig><p>Other RSNs can be affected in patients with IGE, reflecting specific cognitive impairment when compared to controls. A seed-based RS-FC study in 14 patients with IGE showed that attention network is impaired even in interictal periods, and that this impairment is related with the disease duration (<xref rid="B165" ref-type="bibr">165</xref>).</p><p>One study in 60 IGE patients specifically addressed the question whether pharmacoresistance was correlated with RS-FC changes in the DMN (<xref rid="B166" ref-type="bibr">166</xref>). DMN connectivity was reduced in the IGE group compared to HC, and the strongest decrease was found in those patients that were resistant to valproate. Finally, a recent study directly addressed RS-FC within the thalamo-cortical system (<xref rid="B167" ref-type="bibr">167</xref>), finding decreased RS-FC between thalamus and anterior DMN. Collectively, these studies suggest that there is a loss of functional integration in the thalamocortical and default-mode system of the brain in IGE, even outside the GSWD. Although small sample size and heterogeneous methodology limit “generalization,” the abnormal RS-FC patterns found in IGE so far could serve as endophenotypes of different IGE syndromes, and thus inform clinical diagnostics. Importantly, the confounding effect of anti-epileptic drug on dysconnectivity needs to be further elucidated.</p><p>The most frequent IGE syndrome is juvenile myoclonic epilepsy (JME), where seizures are characterized by myoclonic jerks of the upper limbs, often triggered by cognitive inputs. Several RS and task-related functional studies have shown an impairment of connectivity among supplementary motor area and the rest of the brain (<xref rid="B168" ref-type="bibr">168</xref>–<xref rid="B170" ref-type="bibr">170</xref>), suggesting that this structure may represent the anatomic basis for triggering motor seizures in JME. JME patients have often personality characteristics suggestive of a frontal lobe dysfunction (e.g., risk-taking behavior, dysexecutive syndrome). A task-related FC study in JME patients (<xref rid="B171" ref-type="bibr">171</xref>) shows that patients with ongoing seizures learn less from previous experiences compared to seizure-free patients and to controls.</p></sec><sec id="S4-4-3"><title>Pediatric syndromes</title><p>Numerous EEG-fMRI studies have been conducted on pediatric syndromes [for review, see in Ref. (<xref rid="B172" ref-type="bibr">172</xref>)]. Several studies in Lennox–Gastaut syndrome (<xref rid="B173" ref-type="bibr">173</xref>–<xref rid="B175" ref-type="bibr">175</xref>) have shown hemodynamic involvement of brainstem, thalamus, and basal ganglia during paroxysmal fast activity and slow spike-and-wave discharges, underlying the importance of cortical–subcortical networks in Lennox–Gastaut syndrome. A group-analysis study in patients with myoclonic-astatic-epilepsy (<xref rid="B176" ref-type="bibr">176</xref>) showed that GSWD are related not only to a thalamo-cortical network (commonly found in IGE), but also to brain areas associated with motor function, suggesting that the involvement of these structures may predispose to the typical myoclonic jerks observed in this syndrome.</p><p>Concerning idiopathic focal epilepsies of childhood, these comprise a broad spectrum of phenotypes showing an overlap with each other, from benign childhood epilepsy with centro-temporal spikes (BECTS) to more severe seizure and cognitive disorders, like atypical benign partial epilepsy (ABPE), continuous spikes and waves during slow sleep (CSWS), and Landau-Kleffner syndrome. In patients with BECTS, EEG-fMRI studies have revealed focal spike-associated BOLD signal changes in the sensorimotor cortex, well corresponding to spikes localization, and typical seizure semiology (<xref rid="B177" ref-type="bibr">177</xref>–<xref rid="B180" ref-type="bibr">180</xref>). In patients with CSWS, a consistent neuronal network including both cortical and subcortical structures was described with positive BOLD signal changes in the perisylvian region, insula, cingulated cortex, and thalamus, and negative BOLD signal changes in the DMN areas and caudate nucleus (<xref rid="B181" ref-type="bibr">181</xref>). Source analysis of the simultaneously recorded EEG in these patients allowed differentiating initiation from propagation of epileptic activity in these common networks. The importance of assessing sleep state when studying networks is given by the report of a patient, whose centro-temporal spikes were recorded during wakefulness and sleep. BOLD response during wakefulness showed a focal activation concordant with the spike topography, whereas BOLD response to the same event during sleep showed the involvement of a thalamic–perisylvian neural network similar to the one previously observed in patients with CSWS, suggesting a common sleep-related network dysfunction even in cases with milder phenotypes (<xref rid="B182" ref-type="bibr">182</xref>).</p><p>A single-subject and group-analysis study (<xref rid="B183" ref-type="bibr">183</xref>) on patients with ABPE demonstrated that this syndrome is characterized by patterns similar to studies in rolandic epilepsy (focal BOLD signal changes in the spike field) as well as patterns observed in CSWS (distant BOLD signal changes in cortical and subcortical structures), thereby corroborating the concept that idiopathic focal epilepsies of childhood form a spectrum of overlapping syndromes.</p><p>An EEG-fMRI study in thirteen patients with ring 20 chromosome syndrome (<xref rid="B184" ref-type="bibr">184</xref>) shows specific networks involved by different interictal and ictal events of interest, suggesting that some hemodynamic networks are the expression of epilepsy-related cognitive and behavioral deficits typical of ring 20 chromosome syndrome, whereas others can be common to other syndromes with neurobehavioral regression.</p></sec></sec></sec><sec id="S5"><title>Intracranial EEG Studies</title><p>The indication for video-iEEG monitoring is the absence of a unique focal hypothesis regarding the source of the patient’s seizures (obtained with non-invasive investigation), or the need for cortical mapping of the epileptogenic cortex vs. eloquent cortex (<xref rid="B7" ref-type="bibr">7</xref>). Therefore, intracranial electrodes often sample from more than one lobe, although their spatial sampling remains limited and needs to be targeted using all available clinical and paraclinical information. Subdural grids allow dense sampling of the cortical convexity while intracerebral depth electrodes are able to reach deeper structures (e.g., medial temporal structures); combinations of both techniques are feasible. Therefore, iEEG studies represent a unique opportunity to investigate seizure networks in human beings with optimal temporal and excellent spatial resolution.</p><p>The concept of the seizure-onset zone as a single, circumscribed brain region implies that, assuming that intracranial electrodes sample this region, ictal iEEG activity should invariably start there. Careful observation of ictal iEEG recordings, however, reveals that this is not always the case. Rather, there are patients in whom clinically indistinguishable seizures seem to start at any of two or more distinct brain areas (<xref rid="B100" ref-type="bibr">100</xref>). Observations such as this were one of the major factors spurring the interest in considering seizure-generating brain regions as distributed networks. Therefore, the seizure-onset zone could be seen as the particular regions with the lowest seizure threshold while other regions could also give rise to independent seizure onsets, which explains the need to consider more than the sole seizure-onset zone for estimating the epileptogenic zone. In an attempt to quantitatively analyze seizure-onset patterns, Bartolomei and colleagues (<xref rid="B185" ref-type="bibr">185</xref>) developed the epileptogenicity index (EI), which takes into account the transition of iEEG activity toward higher frequencies (a general observation of ictal iEEG patterns) together with the delay in which the transition is observed compared to the ictal onset. This approach has revealed that in a significant portion of TLE patients, the medial and lateral temporal lobe display similar EI, implying that both structures could equally subtend seizure generation. Also of interest, some patients with what seemed like TLE before implantation actually displayed the highest values of EI in the fronto-orbital, opercular, or insular cortex rather than the temporal lobe, and these patients had poorer outcomes after resective surgery, suggesting that they harbored more distributed seizure-generating networks not easily amenable to full resection (<xref rid="B186" ref-type="bibr">186</xref>). The number of brain regions with a high EI increases with the duration of epilepsy, suggesting that epilepsy networks may extend over time as a result of plasticity triggered by pathological activity (<xref rid="B185" ref-type="bibr">185</xref>, <xref rid="B186" ref-type="bibr">186</xref>).</p><p>The same authors analyzed the neurophysiological correlates of alterations of consciousness in TLE (<xref rid="B187" ref-type="bibr">187</xref>). They found that alteration of consciousness was associated with increased broadband synchronization in a network of structures, which were mainly extra-temporal, including the thalamus, cingulate cortex, and parieto-temporal association cortex. Consciousness was preserved as long as excessive synchrony was confined to the temporal lobe. Similarly, loss of consciousness in parietal seizures was associated with widespread parietal and frontal synchronization (<xref rid="B188" ref-type="bibr">188</xref>). The authors framed these observations into the context of the global workspace theory, in which the sustained synchronization of neuronal activity in widely distributed modules renders perceptions, memories, and intentions available to consciousness (<xref rid="B189" ref-type="bibr">189</xref>). This work rejoins observations made with single photon emission computed tomography that temporal lobe seizures causing altered consciousness were associated with widespread cortical and subcortical blood-flow alterations (<xref rid="B190" ref-type="bibr">190</xref>). That group later showed increases in the power of delta oscillations in the frontal and parietal association cortices during seizure-related loss of consciousness (<xref rid="B191" ref-type="bibr">191</xref>). Results from studies in a rat model of complex partial seizures suggest that these widespread changes are caused by transient alteration of activity in the subcortical septal nuclei (<xref rid="B192" ref-type="bibr">192</xref>), implying that the widespread effects of temporal lobe seizures on cortical networks could be mediated indirectly via the midline arousal structures (<xref rid="B193" ref-type="bibr">193</xref>).</p><p>Measures of directed connectivity in seizure networks are starting to reveal the internal organization of the individual nodes that make up the network. To date, most iEEG studies use mathematical approaches to determine the direction of connections. For instance, using focal cortical dysplasia as a model of a circumscribed seizure-onset zone and applying partial directed coherence analysis, Varotto et al. (<xref rid="B12" ref-type="bibr">12</xref>) found that the focal dysplasia indeed acted as the initial generator of ictal activity, as evidenced by its greater out-degree both interictally, pre-ictally, and during ictal onsets [the out-degree is a summary measure of the influence of one network node on all the others (<xref rid="B194" ref-type="bibr">194</xref>)]. Cortical-areas remote from the morphological lesion could also be involved in the onset or early propagation of ictal high-frequency activity and could thus represent secondary foci. Wilke et al. (<xref rid="B195" ref-type="bibr">195</xref>) used frequency-band-specific betweenness centrality, a graph theoretical measure of the “importance” of a node in network pathways, to demonstrate a significant overlap between the intracranial electrodes showing the highest betweenness centrality and the seizure-onset zone delineated visually by clinical neurophysiologists, as well as the resected cortical area. That correspondence was present both during ictal and interictal recordings and was highest for gamma-band frequencies. In addition, the analysis also revealed nodes with high betweenness centrality that had not been clinically identified as part of the seizure-onset zone. Van Mierlo et al. (<xref rid="B196" ref-type="bibr">196</xref>) showed that the single intracranial electrode with the highest out-degree during seizure onsets was included in the clinically defined seizure-onset zone as well as the resection area in all of eight patients. These first findings suggest that analyzing epileptic networks in the framework of graph theory, taking into account the direction of connections between nodes in the network, can help clinicians delineate the primary drivers from secondary nodes in seizure nodes [see also in Ref. (<xref rid="B197" ref-type="bibr">197</xref>) for a review]. In the near future, we expect that the tools of graph theory will be applied more generally to iEEG data to describe more fully the spatio-temporal dynamics of seizure networks. Another unique perspective could be offered by the analysis of simultaneous recordings of iEEG and fMRI (<xref rid="B198" ref-type="bibr">198</xref>, <xref rid="B199" ref-type="bibr">199</xref>) to combine the spatial resolution of iEEG with the mapping of whole-brain BOLD changes related to epileptic activity. This could allow bridging the poorly understood gap between increased iEEG connectivity and reduced BOLD connectivity within epileptic networks (<xref rid="B108" ref-type="bibr">108</xref>).</p><sec><title/><sec id="S5-4-4"><title>Micro-electrode studies in human beings</title><p>Another potential breakthrough in the investigation of epileptic networks could stem, in a somewhat paradoxical fashion, from micro-electrode array recordings, which revealed new insights on the pathophysiology of epilepsy. Schevon et al. (<xref rid="B200" ref-type="bibr">200</xref>) inserted arrays comprising of 96 electrodes arranged in a 4-by-4-mm square pattern in the putative seizure-onset zone allowing to record single unit activity in cortical layers 4 and 5 as well as recording the local-field potentials. They showed that there is a sharp delineation (at a sub-millimetric scale) between cortical-areas involved in intense hypersynchronous firing (the hallmark of ictal activity, based on animal studies) and adjacent areas with only mildly increased firing rate and synchrony. Crucially, visual inspection of the iEEG alone did not allow distinguishing between what the authors termed the seizure core and its (presumably) inhibitory penumbra. The same investigators further proposed that ictal high-frequency oscillations phase-locked to the lower-frequency, high-amplitude ictal iEEG recorded by standard intracranial electrodes might represent a signature of increased firing in the seizure core (<xref rid="B201" ref-type="bibr">201</xref>). These new findings open the possibility of investigating neuronal firing in distributed seizure networks using conventional iEEG electrodes, without the need for micro-electrode arrays. Future work building on these exciting findings will likely increase our understanding of the ways in which seizures alter normal neuronal firing across the nodes of the involved networks.</p></sec><sec id="S5-4-5"><title>Direct electrical stimulation studies</title><p>Direct electrical stimulation (DES) in epileptic patients consists of administering electrical currents to the brain tissue in order to transiently influence or perturb its function. A technique almost as old as epilepsy surgery, it has mostly been used to probe the function of the cortex directly underlying or surrounding the stimulation site (<xref rid="B202" ref-type="bibr">202</xref>–<xref rid="B205" ref-type="bibr">205</xref>). In that context, DES is generally delivered at high frequencies (e.g., 50–60 Hz) for a few seconds with the aim of inducing clinical changes in the patient (<xref rid="B206" ref-type="bibr">206</xref>). More recently, DES has also been used to investigate FC; in that case, single stimulation pulses are delivered at low frequencies (e.g., 1 Hz) and the readout consists of time-locked perturbations in the activity of points distant from the stimulation site (cortico-cortical evoked potentials, CCEPs) (<xref rid="B207" ref-type="bibr">207</xref>). An interesting aspect of DES-based FC assessments is that they are directed, i.e., the effect of stimulation at site A on site B is not necessarily symmetrical with the effect of stimulating B on A (Figure <xref ref-type="fig" rid="F5">5</xref>). There is an intuitive appeal to this “hands-on” interventional approach to reveal directional connectivity. Evoked effective connectivity was found to correlate with FC measured by resting-state fMRI (<xref rid="B22" ref-type="bibr">22</xref>) as well as with anatomical connectivity probed by diffusion tensor imaging (<xref rid="B208" ref-type="bibr">208</xref>). It has been pointed out, however, that DES can activate axons in the antidromic as well as the orthodromic direction, and could also stimulate <italic>fibers de passage</italic>, an important caveat to keep in mind when interpreting the directionality information provided by these data (<xref rid="B209" ref-type="bibr">209</xref>). This highlights the importance of aiming at obtaining multimodal functional and structural information to better understand brain connectivity and dynamics.</p><fig id="F5" position="float"><label>Figure 5</label><caption><p><bold>Evoked effective connectivity reveals the directionality of neural connections in the human brain</bold>. In this example, subdural electrodes are represented by circles, and CCEP responses as lines linking bipolar electrode pairs. Missing (e.g., artifacted) data are indicated by light gray lines, sub-threshold (non-significant) responses by black lines, and significant responses by progressively lighter shades of blue. <bold>(A)</bold> Stimulation of the middle frontal gyrus (electrode pair colored in pink) triggers widespread responses in the frontal and temporal lobes, including the middle temporal gyrus (inset: CCEP waveform from 50 ms before to 250 ms after stimulation; the arrow indicates the time of stimulation). By contrast, stimulation of the middle temporal gyrus <bold>(B)</bold> does not evoke any significant response in the frontal lobe, illustrating that effective connectivity between remote brain structures is not necessarily reciprocal.</p></caption><graphic xlink:href="fneur-05-00218-g005"/></fig><p>Evoked effective connectivity has revealed strong intralobar connectivity in the temporal and frontal lobes, as well as connections between the frontal and temporal lobes that are more prominent in the frontal-to-temporal than in the temporal-to-frontal direction (<xref rid="B210" ref-type="bibr">210</xref>, <xref rid="B211" ref-type="bibr">211</xref>). An intriguing aspect of these studies is the observation that, whereas interhemispheric connections between the frontal lobes are relatively common, temporal-temporal connections appear sparse, being observed in <20% of patients (<xref rid="B211" ref-type="bibr">211</xref>). This begs the question of which neuronal pathways are responsible for bitemporal synchronized spiking as well as the propagation of seizures from one temporal lobe to the other one (<xref rid="B212" ref-type="bibr">212</xref>). Recently, David et al. (<xref rid="B213" ref-type="bibr">213</xref>) generalized this approach offering to develop an atlas of evoked effective connectivity that would eventually allow, through data sharing, sampling most of the human brain’s volume.</p><p>Direct electrical stimulation has also been used to specifically evaluate epileptic networks, the general idea being that the responses of remote sites to stimulation of epileptogenic cortex (<xref rid="B214" ref-type="bibr">214</xref>) or the responses of epileptogenic cortex to stimulation of remote sites (<xref rid="B215" ref-type="bibr">215</xref>) differ from those involving only normal brain tissue. Interestingly, the network of brain areas that respond to DES of the seizure-onset zone overlaps partially but not completely with the areas of ictal propagation, suggesting both that seizures propagate sequentially through multiple nodes in the network and that some existing connections between the seizure-onset zone and distant brain areas “shut down” during seizures (<xref rid="B216" ref-type="bibr">216</xref>). Further research combining iEEG and DES, as well as integrating these techniques with fMRI and high-density non-invasive electromagnetic recordings, will improve our understanding of the physiology of seizure networks.</p></sec></sec></sec><sec id="S6"><title>What We Can Learn from Animal Models</title><p>Recording the activity of any node suspected to be determinant in the disease is not feasible in human beings, contrarily to animal research. Moreover, animal-related technologies offer the possibility to desiccate and manipulate cellular and molecular components of such networks, as well as scrutinizing the associated structural and functional alterations. A great perspective in pathological networks study is detecting features associated with the risk of recurrence after a first seizure as well as predicting the evolution toward pharmaco-resistance.</p><p>Animal models allow studying networks connectivity and recording the underlying brain activity with high spatial coverage and resolution (<xref rid="B217" ref-type="bibr">217</xref>), and addressing the process of epileptogenesis and ictogenesis, including their molecular and genetic mechanisms at cellular and subcellular levels (<xref rid="B218" ref-type="bibr">218</xref>–<xref rid="B222" ref-type="bibr">222</xref>). Imbalance between excitation and inhibition might not only occur at the local microscopic level (<xref rid="B223" ref-type="bibr">223</xref>, <xref rid="B224" ref-type="bibr">224</xref>), but could also reflect dysregulation of excitatory and inhibitory neuronal interactions at a larger (network) scale. Recent evidence emphasizes the modifications of the network dynamic, or network configuration that characterizes, and sometimes precedes or even predicts a seizure. Network analysis could be a powerful tool to more precisely define the different epilepsies and develop new treatments that target networks, instead of focal activity (<xref rid="B11" ref-type="bibr">11</xref>, <xref rid="B100" ref-type="bibr">100</xref>).</p><p>In animals and human beings, focal onsets have been identified in generalized epilepsy, and complex large-scale network involvement has been shown in focal epilepsies (<xref rid="B8" ref-type="bibr">8</xref>, <xref rid="B11" ref-type="bibr">11</xref>, <xref rid="B14" ref-type="bibr">14</xref>). Spontaneous epileptic disease occurs in animals, as in the case of the genetic absence epileptic rats of Strasbourg (GAERS) or in the WAG/Rij rats (<xref rid="B225" ref-type="bibr">225</xref>–<xref rid="B227" ref-type="bibr">227</xref>); other models studied are epileptic conditions induced by – mainly – chemical or electrical interventions (<xref rid="B220" ref-type="bibr">220</xref>). A major animal model of TLE is the kainate, or pilocarpine, model of hippocampal sclerosis (HS) (<xref rid="B228" ref-type="bibr">228</xref>–<xref rid="B232" ref-type="bibr">232</xref>). Kainate, a glutamatergic agonist, is injected either in the hippocampus or intraperitoneally. It is suspected that the kainate has a tropism for the hippocampus, which led several authors to consider that the kainate induces specifically a HS. Yet, the mechanisms by which kainate induces an epileptic activity is still debated; the immune system and leakage of the blood–brain barrier have been cited as critical for the expression of the disease (<xref rid="B233" ref-type="bibr">233</xref>, <xref rid="B234" ref-type="bibr">234</xref>). Hence, it is not excluded that systemic kainate may have diffuse effects on the brain.</p><p>Models of focally induced epileptic disorders might avoid this limitation. One of them, electrical kindling, triggers focal epileptic activity using focal electrical stimulation in accordance with standard stimulation parameters, e.g., duration of the stimulation, frequency, and intensity of the stimulus (<xref rid="B220" ref-type="bibr">220</xref>, <xref rid="B235" ref-type="bibr">235</xref>, <xref rid="B236" ref-type="bibr">236</xref>). The emergence of a distant pathological activity can be related to remote effects of the focally induced epilepsy, and not to the direct diffuse effects of the electrical or chemical triggers. Electrical stimulation, in particular of the performant-path, has also been described as a model of induced status epilepticus (<xref rid="B237" ref-type="bibr">237</xref>, <xref rid="B238" ref-type="bibr">238</xref>).</p><sec id="S6-5"><title>Connectivity studies in animals</title><p>Electrophysiology can assess connectivity and RS networks in animal models of epilepsy by recording several brain regions simultaneously. The great advantage is that the signal can be directly linked to neuronal activity. Using intrahippocampal recording in a rat model of induced TLE, Wang et al. (<xref rid="B239" ref-type="bibr">239</xref>) showed that neuronal pairs presented a decreased FC prior to the status epilepticus induced by an intraperitoneal injection of pilocarpine. Using Graph Theory measures in an <italic>in vitro</italic> Mg<sup>2+</sup>-free model of hippocampal epilepsy, Gong et al. (<xref rid="B240" ref-type="bibr">240</xref>) reported the modifications in network configuration that appear in parallel to epileptiform discharges. More interestingly, they revealed that the changes in network configuration appeared before and lasted longer than the epileptiform discharges (<xref rid="B240" ref-type="bibr">240</xref>). These two observations suggest that the classical ictal activity, i.e., the presence of spikes in the EEG, could be the resultant of network reconfiguration, i.e., it could even be an epiphenomenon of a more profound alteration in brain connectivity, indicating that it could be possible to identify certain network alterations as a biomarker of epilepsy. Such studies aimed at identifying markers of an upcoming ictal activity and have mainly looked at the local activity changes (<xref rid="B241" ref-type="bibr">241</xref>). Knowledge on remote involvement is sparse. Recent works (<xref rid="B224" ref-type="bibr">224</xref>) showed structural alterations remote from the focus, but only a few evidence of distant, abnormal neuronal activity exists (<xref rid="B242" ref-type="bibr">242</xref>). Major advancement has been made to record as many neurons or neuronal populations as possible at the same time (<xref rid="B145" ref-type="bibr">145</xref>, <xref rid="B243" ref-type="bibr">243</xref>–<xref rid="B248" ref-type="bibr">248</xref>); this shows the feasibility to investigate large-scale networks in animal models with high temporal and spatial resolution (Figure <xref ref-type="fig" rid="F6">6</xref>). Their combination with effective connectivity measures (<xref rid="B25" ref-type="bibr">25</xref>) will help to better understand the hierarchical organization of epileptic networks. Gong et al. (<xref rid="B240" ref-type="bibr">240</xref>) demonstrated the leading activity of pyramidal cells over granular cells in an <italic>in vitro</italic> model of TLE, illustrating the utility of effective connectivity in the field of epilepsy.</p><fig id="F6" position="float"><label>Figure 6</label><caption><p><bold>Dynamic of somato-sensory network mapped with high-density EEG</bold>. Somato-sensory evoked potential (SEP) from left whisker stimulation. <italic>Top left:</italic> each black dot represents the position of one recording electrode; the most anterior one is ground. <italic>Top right:</italic> 31 electrode traces displaying the SEP with sub-milliseconds resolution. <italic>Bottom left:</italic> the same electrode traces represented over the mouse brain. In the lissencephalic mouse brain, dipoles are estimated to be generated below the recording electrodes. <italic>Bottom right: s</italic>egmentation of the SEP in six stable configurations of potential maps. The technique’s high spatial and temporal resolutions identify the first component, somato-sensory barrel field activity, followed by motor cortex and contra-lateral somato-sensory areas recruitment within a few milliseconds. Adapted from Megevand et al. (<xref rid="B243" ref-type="bibr">243</xref>) with permission.</p></caption><graphic xlink:href="fneur-05-00218-g006"/></fig><p>Research on animal models of epilepsy has been dominated by invasive electro-physiology technique. Recently, the combination of EEG and fMRI has emerged with interesting results, such as those reported by Englot et al. (<xref rid="B192" ref-type="bibr">192</xref>), where they describe how a partial limbic seizure lead to neocortical slow-wave activity; yet, technical issues makes difficult to obtain combined EEG-fMRI in awake animals. As in Englot et al. (<xref rid="B192" ref-type="bibr">192</xref>), fMRI could possibly be a powerful <italic>in vivo</italic> screening method for anatomical regions that could then be more deeply investigated with EEG.</p><p>Using fMRI, Mishra et al. (<xref rid="B249" ref-type="bibr">249</xref>) showed that rats submitted to traumatic brain injury through left parietal fluid percussion presented a decreased correlation coefficient between the left parietal cortex and other brain regions. Dysfunctional activity in the left parietal cortex, as highlighted by the decreased correlation coefficient could have been expected, yet the pattern of resting BOLD-fMRI connectivity showed that only certain regions were specifically affected, namely the left hippocampus and the contra-lateral parietal cortex. This illustrates that BOLD-fMRI can be used to identify secondary dysfunctional brain regions in rodents following a proepileptogenic injury (<xref rid="B249" ref-type="bibr">249</xref>). The same group investigated with fMRI the FC in WAG/Rij rats (<xref rid="B250" ref-type="bibr">250</xref>) and found an increased correlation between brain areas suspected to be involved in seizures when compared to non-epileptic rats; more importantly, this increase was observed outside of the ictal periods. Choi et al. (<xref rid="B251" ref-type="bibr">251</xref>) performed a FC study using the <sup>18</sup>fluorodeoxyglucose positron emission tomography (PET) signal in a rat model of TLE. They revealed the decreased correlation of several pairs of brain structures, most of them included left amygdala and left entorhinal cortex (<xref rid="B251" ref-type="bibr">251</xref>). Hence, despite the systemic injection of pilocarpine, the affected network appeared to be mainly restricted to the left hemisphere (<xref rid="B251" ref-type="bibr">251</xref>). It would have been very interesting to see if the electro-physiological counterpart of such functional deficit was also restricted to one hemisphere, yet no EEG recording was reported. Asymmetry in the central nervous system is well recognized, e.g., asymmetry of the temporal lobes, but the reason why the left hemisphere appears to be more functionally altered in this rat model of TLE is unclear, although electro-physiological experiments suggest that the left hemisphere is indeed more prone to develop epileptic discharges (<xref rid="B252" ref-type="bibr">252</xref>). The authors claimed that the PET images were acquired in the interictal period, but no EEG recording was used (<xref rid="B251" ref-type="bibr">251</xref>); yet, if true, this would suggest that epileptic animals can be identified as such on the basis of the FC of particular networks outside of any ictal activity. These studies (<xref rid="B249" ref-type="bibr">249</xref>–<xref rid="B251" ref-type="bibr">251</xref>) indicate that the pathological process in these rats is ongoing: the epileptic brain is not suffering from epilepsy only during seizures.</p><p>The anatomical basis of FC is largely unclear. Zhou et al. (<xref rid="B253" ref-type="bibr">253</xref>) nicely investigated the anatomical substrate and plasticity of such connections. They observed that after partial posterior callosotomy of wild-type rats, the FC of the auditory and visual cortices decreased at day 7 and returned to baseline at day 28, whereas this decrease was still present in rats submitted to complete callosotomy (<xref rid="B253" ref-type="bibr">253</xref>). The authors concluded that it could be possible to identify the anatomical basis of FC, and that these functional connections were also capable of plasticity. This is an important proof-of-concept: it is possible to identify morphological substrate of functional connections and manipulate them.</p></sec><sec id="S6-6"><title>Differential involvement of specific brain regions in animal models of generalized epilepsy</title><p>Different rat models of generalized absence epilepsy have been studied and all share the presence of the characteristic SWDs (<xref rid="B254" ref-type="bibr">254</xref>, <xref rid="B255" ref-type="bibr">255</xref>). Using combined EEG-fMRI in WAG/Rij rats, Mishra et al. (<xref rid="B250" ref-type="bibr">250</xref>) demonstrated that during SWDs, the associated fluctuations in the BOLD signal were specific to certain brain regions. Indeed, the somato-sensory barrel field showed an increase, whereas the striatum showed a decrease in the BOLD signal and cerebral blood-flow (<xref rid="B225" ref-type="bibr">225</xref>, <xref rid="B250" ref-type="bibr">250</xref>). On the other hand, the local-field potential (LFP) and the multi-unit activity (MUA) were increased in both regions (<xref rid="B225" ref-type="bibr">225</xref>). Vascular steal or dopamine-regulated blood volume could account, at least in part, for this lack of matching between BOLD signal and CBF on the one hand and LFP and MUA on the other hand (<xref rid="B225" ref-type="bibr">225</xref>). An earlier study using surface and deep EEG recordings in the same rat-model showed that these rats shared a similar focus located in the ventrolateral part of the somato-sensory cortex (SC) (<xref rid="B145" ref-type="bibr">145</xref>). More importantly, the authors observed that the ictal activity of the cortical focus preceded the one in the thalamus, suggesting that the cortex was leading the thalamus (<xref rid="B145" ref-type="bibr">145</xref>). Nersesyan et al. (<xref rid="B256" ref-type="bibr">256</xref>) investigated the relation between SWDs and CBF in the same animal model. They showed that regions involved in SWDs, i.e., SC, presented a 1- to 2-s delayed increase in CBF during a SWD, whereas this increase did not appear in regions not involved in the SWDs, such as primary visual cortex (<xref rid="B256" ref-type="bibr">256</xref>). In a parallel work using the same animal model of absence epilepsy, they observed that the BOLD signal was not equally modified across brain regions during a SWD (<xref rid="B257" ref-type="bibr">257</xref>): the somato-sensory and motor cortices, as well as subcortical regions, i.e., thalamus, basal ganglia, and brainstem, showed an increased BOLD signal, whereas other regions such as the occipital cortex did not show such a modulation of the signal (<xref rid="B257" ref-type="bibr">257</xref>). Again, the increase in the BOLD signal appeared with a lag of a few seconds after the electro-physiological SWDs (<xref rid="B257" ref-type="bibr">257</xref>). This finding is in contrast with a work by Desalvo et al. (<xref rid="B258" ref-type="bibr">258</xref>), in which they used a rat model of generalized tonico-clonic seizures induced by injection of iv bicuculline, and observed that BOLD increased significantly in primary and secondary somato-sensory cortices, as well as in primary auditory cortex and thalamus <italic>before</italic> the onset of electro-physiological seizures. The role of the SC in initiating GSWDs was further investigated through inactivation of this cortical region in GAERS animals (<xref rid="B259" ref-type="bibr">259</xref>). The pharmacological inactivation of the SC with the sodium channel blocker tetrodotoxine prevented the spike-and-wave activity; yet unilateral application of the drug did not completely abolish the abnormal contra-lateral oscillations. On the whole, these studies highlight the importance of abnormal focal brain activity as a potential trigger of generalized seizures (<xref rid="B258" ref-type="bibr">258</xref>). The identification of interacting yet independent nodes within a network of suspected generalized epilepsy is a major advance in epilepsy research. Indeed, it will permit to refine the therapeutic intervention toward the manipulation of one particular and decisive node.</p></sec><sec id="S6-7"><title>Short-range and long-range network modulations in animal models of focal epilepsy</title><p>Different animal models of focal epilepsy exist (<xref rid="B220" ref-type="bibr">220</xref>), such as the kainate- or pilocarpine-models of MTLE (<xref rid="B228" ref-type="bibr">228</xref>, <xref rid="B229" ref-type="bibr">229</xref>, <xref rid="B260" ref-type="bibr">260</xref>), posttraumatic epilepsy (<xref rid="B261" ref-type="bibr">261</xref>, <xref rid="B262" ref-type="bibr">262</xref>), or electrical kindling (<xref rid="B227" ref-type="bibr">227</xref>, <xref rid="B263" ref-type="bibr">263</xref>). Despite an initially focal insult, recent evidence (e.g., <xref rid="B242" ref-type="bibr">242</xref>) shows that remote brain areas become affected by the pathological activity of the epileptic focus.</p><p>The induction of a focal epileptic syndrome in a rat model of generalized epilepsy allows better understanding how these two entities interact. Carcak et al. (<xref rid="B227" ref-type="bibr">227</xref>) took advantage of the fact that absence epilepsy may increase the resistance to limbic seizures. They investigated the role of the cortico-thalamo-cortical circuitry, involved in SWDs, in the development of limbic seizures induced by unilateral electrical stimulation of the rat amygdala. Whereas control rats, i.e., those without absence epilepsy, presented all convulsive epileptic seizures following amygdala electrical stimulation, rats suffering from absence epilepsy did not (<xref rid="B227" ref-type="bibr">227</xref>). In order to understand how the circuit involved in absence epilepsy could affect the one of TLE, the authors investigated the spontaneous activity in the reticular thalamic nucleus (RTN), known to be involved in the slow-waves discharges that characterize absence epilepsy (<xref rid="B227" ref-type="bibr">227</xref>). Remarkably, the electrical stimulation of the amygdala had a different effect on the mean firing frequency of neurons of the RTN: in not-stimulated animals, there was no significant difference between epileptic and non-epileptic rats, whereas the increase after stimulation was higher in epileptic rats when compared to non-epileptic rats (<xref rid="B227" ref-type="bibr">227</xref>). This suggests first that the development of an epileptic focus alters the activity of neurons in the RTN and second that this alteration depends on the activity before the induction of the epilepsy. The use of Wistar rats as controls for GAERS rats in that study is commonly accepted, but could still be questioned; yet the conclusion would still remains the same: the effects of an epileptic focus seem to depend on the brain state in which it is being established. It would hence be interesting to investigate how an epileptic focus affects a given network, but also how a particular network configuration can modulate the effects of an epileptic focus. The involvement of the thalamus in propagation of temporal lobe seizures has already been the scope of several studies (<xref rid="B156" ref-type="bibr">156</xref>, <xref rid="B264" ref-type="bibr">264</xref>). If the thalamus has a major role in the generation of SWDs (<xref rid="B227" ref-type="bibr">227</xref>), this could highlight the relevance of studying the interaction between hippocampus and thalamus, in the context of focal epilepsies.</p><p>Hippocampal sclerosis is a frequent lesion that has been deeply investigated, although, most of the works conducted local, intrahippocampal recordings. Yet, recent publications have shown the involvement of remote brain areas. Using 16 bipolar deep electrodes in the pilocarpine rat-model of HS, Toyoda et al. (<xref rid="B247" ref-type="bibr">247</xref>) showed that the initial focus varied from one seizure to another in each individual rat. The subiculum, the dorsal and ventral hippocampus, and the amygdala were the regions where seizure onsets were most often recorded. All regions could be considered as belonging to the same network; indeed, an interesting observation is that most seizures were convulsive, and this did not depend upon where the seizure started (<xref rid="B247" ref-type="bibr">247</xref>). This suggests that the involved network is more determining than the seizure-onset zone for the clinical expression of a seizure. Long-range modifications in the kainate mouse-model of TLE were also observed. It has been shown that non-injected hippocampus presented indeed morphological alterations, notably in the expression of the neuropeptide-Y, which is known to modulate neuronal activity (<xref rid="B265" ref-type="bibr">265</xref>, <xref rid="B266" ref-type="bibr">266</xref>), and electro-physiological changes, such as significant decrease in the power of the theta frequency band (<xref rid="B265" ref-type="bibr">265</xref>). <italic>In vitro</italic>, Khalilov et al. (<xref rid="B267" ref-type="bibr">267</xref>) demonstrated that a mirror focus in the contra-lateral hippocampus appears after 10–15 successive applications of kainate in the ipsilateral hippocampus. These findings are in line with the hypothesis that an epileptic focus leads to permanent electro-physiological and morphological modifications in remote brain areas (<xref rid="B268" ref-type="bibr">268</xref>–<xref rid="B270" ref-type="bibr">270</xref>). Other works have also stressed the possibility that subcortical brain regions, such as the basal ganglia, could influence or even inhibit the progression of an ictal activity originating from the temporal lobe (<xref rid="B271" ref-type="bibr">271</xref>, <xref rid="B272" ref-type="bibr">272</xref>).</p><p>Evidence of distant brain involvement arises also from electrically induced epilepsy. For instance, during hippocampal seizures induced by electrical stimulation in rats, the frontal neocortex presented a parallel modification in spontaneous activity, i.e., fast polyspike activity when the seizure was generalized and slow oscillations when it was partial (<xref rid="B242" ref-type="bibr">242</xref>). This example illustrates that distant brain areas are affected even after a few or only one focal epileptic seizure. It would be extremely interesting to study how this involvement evolves in a chronic disease.</p><p>On the whole, evidence exists that other brain areas are recruited in propagation or in inhibition of the seizure spread. Hence, the epileptic threshold does not seem to depend only on the imbalance between excitation and inhibition within the focus, but could also be determined by the intricate interactions between the components of a given network.</p></sec><sec id="S6-8"><title>Experimental therapeutic interventions on the epileptic network</title><p>Conceiving epilepsy as a network disease has therapeutic consequences. The classical view is to modulate the activity of the so-called epileptic focus, or seizure-onset zone, in order to control the disease. Yet, any node of an epileptic network could possibly be a target. In this sense, open-loop or closed-loop devices, either through electrical or optogenetic stimulation, are promising tools for generalized (<xref rid="B217" ref-type="bibr">217</xref>) as well as for focal epilepsy (<xref rid="B273" ref-type="bibr">273</xref>, <xref rid="B274" ref-type="bibr">274</xref>). Major work has shown that it is possible to identify critical nodes in a given epileptic network: the modification of their activation – mainly inhibition – could help to control, or even stop an ictal activity. Paz et al. (<xref rid="B274" ref-type="bibr">274</xref>) showed in a rat model of cortical epilepsy that the inactivation through optogenetics of the thalamic ventrobasal nucleus could stop an ongoing seizure. In the same line, Langlois et al. (<xref rid="B264" ref-type="bibr">264</xref>) showed in an <italic>in vivo</italic> model of TLE that DBS of the ipsilateral parafascicular nucleus of the thalamus (PF) stopped the ongoing hippocampal paroxysmal discharges (HPD), while higher current intensities were needed to stop the HPD if DBS was applied to neighboring areas (<xref rid="B264" ref-type="bibr">264</xref>), illustrating the specificity of PF in controlling HPDs. The anterior thalamic nucleus (ANT) appears also to be involved in control or spread of epileptic activity of mesial temporal onset (<xref rid="B156" ref-type="bibr">156</xref>). Ablation or electrical stimulation of ANT increases the epileptic threshold (<xref rid="B263" ref-type="bibr">263</xref>, <xref rid="B275" ref-type="bibr">275</xref>–<xref rid="B277" ref-type="bibr">277</xref>); yet, opposite results have also been observed (<xref rid="B278" ref-type="bibr">278</xref>). On the whole, ANT is a recognized target for refractory epilepsy, although mechanisms by which manipulation of the ANT increases epileptic threshold are poorly understood. Use of animal research and the possibility to identify how the activity of ANT may modulate epileptic activity at remote sites, e.g., with the use of effective connectivity measures, is crucial to tailor therapeutic interventions. Such recent evidence shows that the manipulation of the primary epileptic focus does not seem to be the only possibility to achieve the control of an epileptic disease. The thalamus in particular, and other subcortical regions as well (<xref rid="B272" ref-type="bibr">272</xref>) have been identified as major targets for epileptic network modulation culminating in clinical applications in the form of DBS of ANT in focal epilepsies (<xref rid="B279" ref-type="bibr">279</xref>).</p></sec></sec><sec id="S7"><title>Conclusion</title><p>With increasingly complex methodological strategies and an ever-increasing wealth of possible approaches, the study of brain connectivity and its neuroscientific and clinical correlates are very promising. Nevertheless, the application of connectivity techniques for diagnostic or prognostic purposes requires further studies to be firmly grounded by invasive studies and sufficient follow-up investigations before it can be reliably applied to the clinical management of individual patients. Combining functional techniques can lead to the achievement of complementary information for improving each single technique.</p><p>Focal epilepsies, despite focal epileptogenic zone, are diseases affecting the whole brain: altered large-scale FC is reflected in neuropsychological features of individual specific syndrome. Hippocrates (400 years b.c.) considered epilepsy as a systemic disease, centered in the brain, due to an altered “defluxion of cold phlegm” through the body. In more recent times, the concept of epilepsy as “focus disease” has been largely developed (<xref rid="B280" ref-type="bibr">280</xref>–<xref rid="B282" ref-type="bibr">282</xref>), whereas in the last decade it has shifted to a “brain-network disease” (<xref rid="B15" ref-type="bibr">15</xref>). In parallel to the “brain-network” concept of epilepsy, psychiatric and neurological co-morbidities, such as strokes, dementia, and migraine are more and more defined. Interestingly, somatic co-morbidities have also come to light, since several medical conditions, such as cardiac, gastrointestinal, and respiratory disorders, are often associated with epilepsy (<xref rid="B283" ref-type="bibr">283</xref>). These findings may lead to re-consider epilepsy as a “systemic disease,” this time with the diagnostic and therapeutic knowledge obtained recently by ground-breaking work on network analysis.</p><p>Concerning “generalized” epilepsy, neuro-imaging, and especially connectivity studies have allowed considering them as focal brain disorders with fast bilateral discharge propagation. This concept leads to the idea that focal and generalized epilepsies are the two extremes of a single spectrum and to a possible new way of studying mechanisms of AED: do they have an effect on particular nodes of a network where receptors are more expressed? Is it possible to detect an anatomical target to avoid generation/propagation of seizures, using disconnection or stimulation? For all these reasons, translational research in light of network analysis, based on fundamental science through animal experiments and clinical perspectives through human research, opens new opportunities to better understand the complexity of epilepsy and define new and more effective treatments for patients.</p></sec><sec id="S8"><title>Conflict of Interest Statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec>
An exploratory study of the effects of spatial working-memory load on prefrontal activation in low- and high-performing elderly	<p>Older adults show more bilateral prefrontal activation during cognitive performance than younger adults, who typically show unilateral activation. This over-recruitment has been interpreted as compensation for declining structure and function of the brain. Here we examined how the relationship between behavioral performance and prefrontal activation is modulated by different levels of working-memory load. Eighteen healthy older adults (70.8 ± 5.0 years; MMSE 29.3 ± 0.9) performed a spatial working-memory task (n-back). Oxygenated ([O<sub>2</sub>Hb]) and deoxygenated ([HHb]) hemoglobin concentration changes were registered by two functional Near-Infrared Spectroscopy (fNIRS) channels located over the left and right prefrontal cortex. Increased working-memory load resulted in worse performance compared to the control condition. [O<sub>2</sub>Hb] increased with rising working-memory load in both fNIRS channels. Based on the performance in the high working-memory load condition, the group was divided into low and high performers. A significant interaction effect of performance level and hemisphere on [O<sub>2</sub>Hb] increase was found, indicating that high performers were better able to keep the right prefrontal cortex engaged under high cognitive demand. Furthermore, in the low performers group, individuals with a larger decline in task performance from the control to the high working-memory load condition had a larger bilateral increase of [O<sub>2</sub>Hb]. The high performers did not show a correlation between performance decline and working-memory load related prefrontal activation changes. Thus, additional bilateral prefrontal activation in low performers did not necessarily result in better cognitive performance. Our study showed that bilateral prefrontal activation may not always be successfully compensatory. Individual behavioral performance should be taken into account to be able to distinguish successful and unsuccessful compensation or declined neural efficiency.</p>	<contrib contrib-type="author"><name><surname>Vermeij</surname><given-names>Anouk</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><xref ref-type="author-notes" rid="fn001"><sup>*</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/104719"/></contrib><contrib contrib-type="author"><name><surname>van Beek</surname><given-names>Arenda H. E. A.</given-names></name><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/190068"/></contrib><contrib contrib-type="author"><name><surname>Reijs</surname><given-names>Babette L. R.</given-names></name><xref ref-type="aff" rid="aff3"><sup>3</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/187883"/></contrib><contrib contrib-type="author"><name><surname>Claassen</surname><given-names>Jurgen A. H. R.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/125575"/></contrib><contrib contrib-type="author"><name><surname>Kessels</surname><given-names>Roy P. C.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><xref ref-type="aff" rid="aff4"><sup>4</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/140509"/></contrib>	Frontiers in Aging Neuroscience	<sec sec-type="introduction" id="s1"><title>Introduction</title><p>Studies on the cognitive neuroscience of aging have reliably revealed age-related differences in brain activation during cognitive task performance (for reviews, see Spreng et al., <xref rid="B40" ref-type="bibr">2010</xref>; Eyler et al., <xref rid="B14" ref-type="bibr">2011</xref>; Grady, <xref rid="B17" ref-type="bibr">2012</xref>; Turner and Spreng, <xref rid="B45" ref-type="bibr">2012</xref>). Two patterns of age-related differences in brain activation have been consistently reported. The first is an age-related reduction in occipitotemporal activation together with an age-related increase in activation of the prefrontal cortex. This has been called the “posterior-anterior shift in aging” (PASA; Grady et al., <xref rid="B18" ref-type="bibr">1994</xref>; Davis et al., <xref rid="B9" ref-type="bibr">2008</xref>). The second is a more bilateral pattern of prefrontal activation in older adults on tasks for which young adults typically show unilateral activation. This pattern has been referred to as Hemispheric Asymmetry Reduction in OLDer adults (HAROLD; Cabeza, <xref rid="B2" ref-type="bibr">2002</xref>). Age-related over-recruitment of the prefrontal cortex has been observed across several cognitive domains such as perception, attention, memory encoding and retrieval, and executive functioning, but most extensively for working memory and inhibitory control tasks (Spreng et al., <xref rid="B40" ref-type="bibr">2010</xref>).</p><p>Over-recruitment of the prefrontal cortex in older adults has been interpreted as a compensatory mechanism that can aid cognitive performance (Cabeza, <xref rid="B2" ref-type="bibr">2002</xref>). The traditional cognitive aging theories, such as the sensory deficit theory (Baltes and Lindenberger, <xref rid="B1" ref-type="bibr">1997</xref>), resources deficit theory (Craik, <xref rid="B7" ref-type="bibr">1986</xref>), speed deficit theory (Salthouse, <xref rid="B37" ref-type="bibr">1996</xref>), and inhibition deficit theory (Hasher and Zacks, <xref rid="B19" ref-type="bibr">1988</xref>) were developed to explain age-related differences in behavioral performance, but did not always incorporate assumptions regarding age-related differences in brain activation. Dennis and Cabeza (<xref rid="B10" ref-type="bibr">2008</xref>) expanded these traditional theories with additional assumptions regarding brain correlates of relevant cognitive processes and regarding compensatory mechanisms. They concluded that these theories are consistent with the notion of compensation and with evidence from functional neuroimaging studies. Alternatively, age-related prefrontal over-recruitment may reflect less efficient use of neural resources or a less selective recruitment of brain areas, also known as dedifferentiation, which might not necessarily lead to better task performance (Logan et al., <xref rid="B23" ref-type="bibr">2002</xref>). Although there is support for both alternatives, most neuroimaging results are consistent with the compensation account rather than the dedifferentiation account (Spreng et al., <xref rid="B40" ref-type="bibr">2010</xref>; Eyler et al., <xref rid="B14" ref-type="bibr">2011</xref>).</p><p>Meta-analysis of performance-related prefrontal activation across cognitive domains revealed that when performance was equivalent in young and older adults, young adults showed stronger activity in the left ventrolateral prefrontal cortex, whereas older adults showed stronger activity in the left dorsolateral prefrontal cortex. When performance was not equivalent, worse performing older adults showed stronger recruitment of the right dorsolateral prefrontal cortex and right rostrolateral prefrontal cortex (Spreng et al., <xref rid="B40" ref-type="bibr">2010</xref>). The meta-analytic review by Turner and Spreng (<xref rid="B45" ref-type="bibr">2012</xref>) provided evidence that patterns of age-related functional brain change are dissociable for two of the most frequently studied executive processes: inhibition and working memory. During inhibitory control tasks, older adults engaged brain regions commonly recruited in younger adults, but to a larger extent. In contrast, during working-memory performance, older adults showed stronger recruitment of both left and right dorsolateral prefrontal cortex than younger adults. These results were consistent with previous studies on working memory reporting larger and less lateralized recruitment of the dorsolateral prefrontal cortex (Reuter-Lorenz et al., <xref rid="B34" ref-type="bibr">2000</xref>; Cabeza et al., <xref rid="B3" ref-type="bibr">2004</xref>). Due to lack of sufficient statistical power, the relationship between performance differences and prefrontal activation patterns could unfortunately not be examined in the meta-analytic review by Turner and Spreng (<xref rid="B45" ref-type="bibr">2012</xref>).</p><p>An unresolved issue is how over-recruitment of the prefrontal cortex is associated with the variation of cognitive performance levels among older adults. The review by Eyler et al. (<xref rid="B14" ref-type="bibr">2011</xref>) focused on the association between functional response and cognitive performance in healthy young and older adults. Of the 74 reviewed studies that examined the relation between prefrontal activation and cognitive performance, 35% found a positive correlation, 18% found a negative correlation, 16% found mixed results and 31% did not find a significant correlation. Of the 29 studies that were consistent with HAROLD and/or PASA patterns, 34% found a positive correlation between prefrontal activation and cognitive performance, whereas 27% found a negative correlation. Although these results suggest that increased prefrontal activation might be beneficial rather than detrimental at older age, clearly more work is needed to unravel the brain-behavior correlations at older age.</p><p>The aim of the present study is to gain more insight into the role of over-recruitment of the prefrontal in older adults during working-memory performance. Specifically, we examined the relationship between prefrontal activation and behavioral performance by comparing high and low performers. The participants performed a spatial working-memory task with varying levels of cognitive load while their prefrontal activation was measured by functional Near-Infrared Spectroscopy (fNIRS). fNIRS enables monitoring of concentration changes of oxygenated hemoglobin ([O<sub>2</sub>Hb]) and deoxygenated hemoglobin ([HHb]) in the cortex with high temporal resolution. In comparison to fMRI, fNIRS has the advantage that it is less expensive, less invasive, less sensitive to movement artefacts, and that it is portable. The spatial resolution of fNIRS is however limited (Cui et al., <xref rid="B8" ref-type="bibr">2011</xref>; Ferrari and Quaresima, <xref rid="B15" ref-type="bibr">2012</xref>). The majority of neuroimaging studies investigating the brain-behavior relationship in older adults assessed cognitive performance by accuracy, followed by reaction time (Eyler et al., <xref rid="B14" ref-type="bibr">2011</xref>). In the current study, cognitive performance was assessed by a composite score of these measures to take speed/accuracy trade-offs into account and to diminish strategy effects.</p><p>Prefrontal activation is modulated by working-memory load. Previous fMRI studies showed that in young as well as older adults, prefrontal activation increases with working-memory load up to where the working-memory capacity limit is reached, and then levels off or decreases (Mattay et al., <xref rid="B24" ref-type="bibr">2006</xref>; Schneider-Garces et al., <xref rid="B39" ref-type="bibr">2009</xref>). In order to explain contrasting evidence of both age-related under-recruitment as well as age-related over-recruitment of the prefrontal cortex during working-memory performance, Reuter-Lorenz and Cappell (<xref rid="B33" ref-type="bibr">2008</xref>) formulated the Compensation-Related Utilization of Neural Circuits Hypothesis (CRUNCH). CRUNCH proposes that, irrespective of age, neural engagement varies with the level of task demand; activity in cortical regions is upregulated up to a certain level as cognitive load increases. The relationship between cognitive load and brain activation has been described as an S-shaped function. The idea behind CRUNCH is that at low levels of cognitive load, older adults need to recruit more neural resources than young adults in order to maintain task performance, due to less efficient neural processing at older age. At high levels of cognitive load, this compensatory mechanism is no longer effective, leading to reduced or equivalent activation in older adults in comparison to young adults. Hence, in older adults the S-shaped function would be shifted to the left relative to young adults. It has been proposed that a similar effect would be observed when low-performing older adults are compared to high-performing older adults (Grady, <xref rid="B17" ref-type="bibr">2012</xref>), resulting in a leftward shift of the S-shaped curve in low performers relative to high performers. Therefore, in the current study, we expected that high-performing older adults would show increasing prefrontal activation up to a high level of working-memory load. Furthermore, we hypothesized that low performers would reach their working-memory capacity limit sooner than high performers, reflected by reduced prefrontal activation in low performers compared to high performers at a high level of working-memory load.</p></sec><sec sec-type="materials\|methods" id="s2"><title>Materials and methods</title><sec id="s2-1"><title>Participants</title><p>Eighteen healthy older adults participated in this study. Sample characteristics are shown in Table <xref ref-type="table" rid="T1">1</xref>. All participants had completed secondary school or higher. Estimated IQ was based on assessment of the Dutch equivalent of the National Adults Reading Test (Schmand et al., <xref rid="B38" ref-type="bibr">1992</xref>). None of the older adults experienced subjective memory problems, all were living independently at home, and all had unimpaired overall cognitive function as assessed with the Mini Mental State Examination (MMSE; Folstein et al., <xref rid="B16" ref-type="bibr">1975</xref>). All participants were right-handed and had normal or corrected-to-normal vision. None of the participants had a history of neurological or psychiatric disease, or received psychopharmacological drugs or hormone therapy (self report). Six participants used antihypertensive medication. All participants refrained from alcohol, caffeine, and nicotine from at least 3 h before the experimental session. The research proposal for the present study was submitted to the regional medical-ethics committee (CMO Arnhem-Nijmegen, no. 2009/198), but was deemed exempt from formal medical ethical evaluation, because the study does not fall within the remit of the Medical Research Involving Human Subjects Act (WMO). All participants gave written informed consent. The study was performed according to the Helsinki Declaration.</p><table-wrap id="T1" position="float"><label>Table 1</label><caption><p><bold>Sample characteristics</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th rowspan="1" colspan="1"/><th align="center" rowspan="1" colspan="1">Total group</th><th align="center" rowspan="1" colspan="1">Low performers</th><th align="center" rowspan="1" colspan="1">High performers</th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">Participants</td><td align="center" rowspan="1" colspan="1">18 (11 female, 7 male)</td><td align="center" rowspan="1" colspan="1">9 (6 female, 3 male)</td><td align="center" rowspan="1" colspan="1">9 (5 female, 4 male)</td></tr><tr><td align="left" rowspan="1" colspan="1">Age</td><td align="center" rowspan="1" colspan="1">70.8 ± 5.0 (range 64–81)</td><td align="center" rowspan="1" colspan="1">72.7 ± 5.6 (range 64–81)</td><td align="center" rowspan="1" colspan="1">69.0 ± 3.8 (range 65–77)</td></tr><tr><td align="left" rowspan="1" colspan="1">Years of education</td><td align="center" rowspan="1" colspan="1">13.1 ± 3.2 (range 9–18)</td><td align="center" rowspan="1" colspan="1">13.5 ± 3.4 (range 9–18)</td><td align="center" rowspan="1" colspan="1">12.7 ± 3.1 (range 9–18)</td></tr><tr><td align="left" rowspan="1" colspan="1">Estimated IQ</td><td align="center" rowspan="1" colspan="1">108.1 ± 11.0 (range 87–124)</td><td align="center" rowspan="1" colspan="1">109.7 ± 11.3 (range 90–124)</td><td align="center" rowspan="1" colspan="1">106.4 ± 11.0 (range 87–118)</td></tr><tr><td align="left" rowspan="1" colspan="1">MMSE</td><td align="center" rowspan="1" colspan="1">29.3 ± 0.9 (range 27–30)</td><td align="center" rowspan="1" colspan="1">29.7 ± 0.7 (range 28–30)</td><td align="center" rowspan="1" colspan="1">28.9 ± 0.9 (range 27–30)</td></tr></tbody></table></table-wrap></sec><sec id="s2-2"><title>Experimental procedure</title><p>Participants performed three versions of a spatial n-back task (Figure <xref ref-type="fig" rid="F1">1</xref>): 0-back task (control condition), 1-back (low working-memory load), and 2-back task (high working-memory load condition). Prior to all conditions, participants practiced the task for 1 min and received feedback about their performance. The conditions were presented in ascending order and were preceded by a baseline period of 1 min, during which a black fixation cross was displayed at the center of the 15 inch screen. All conditions consisted of 60 trials, 17 of which were target trials. In each trial, a square was presented in black on a light gray background with a presentation time of 500 ms at 1 of 14 pre-specified locations on the display. During the interstimulus interval of 3000 ms, a fixation cross was displayed. During each trial, participants indicated whether the stimulus was a target by pressing the button under the right index finger, or a non-target by pressing the button under the right middle finger (PST Serial Response Box, Psychology Software Tools Inc., PA, USA). Participants were allowed to respond until the next stimulus appeared. In the 0-back condition, a square at one of the four outer corners of the screen was defined as target. In the 1-back condition, the target was any square that appeared at the same location as the square presented one trial before, while squares no longer appeared in the corners. In the 2-back condition, the target was any square that appeared at the same location as the square presented two trials before. In order to prevent verbalization of the locations by the participant, no grid or clock configuration of the squares was chosen. The experimental procedure lasted around 20 min per participant. To minimize effects of fatigue, participants were able to rest a couple of minutes between conditions.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Schematic overview of the spatial n-back task</bold>. At the right upper corner of the figure, all possible positions of the square are shown. ISI = interstimulus interval.</p></caption><graphic xlink:href="fnagi-06-00303-g0001"/></fig></sec><sec id="s2-3"><title>Data acquisition</title><p>We used a continuous-wave NIRS device (Oxymon Mk III, Artinis Medical Systems, Netherlands), using light of three wavelengths (765, 857, 859 nm), to monitor concentration changes in cortical [O<sub>2</sub>Hb] and [HHb] with high temporal resolution. The principle behind fNIRS is that near-infrared light penetrates the skull and brain and is absorbed by the chromophores [O<sub>2</sub>Hb] and [HHb], which have different absorption spectra (Ferrari and Quaresima, <xref rid="B15" ref-type="bibr">2012</xref>). Assuming constant scattering (Sakatani et al., <xref rid="B35" ref-type="bibr">2006</xref>), and by using the modified Lambert-Beer Law, it is possible to calculate the concentration changes of these chromophores in the penetrated brain tissue based on changes in the detected light intensity. Increases in [O<sub>2</sub>Hb] and decreases in [HHb] are indicators of cortical activation.</p><p>In the present study, two pairs of optodes were bilaterally attached to the forehead and were tightly fixed in a customized headband (Spencer technologies, Seattle, WA, USA). The detection optodes were placed 25–30 mm above the midpoint of the eyebrow, at approximately FP1 and FP2 according to the international 10–20 electrode system. The emission optodes were laterally placed at approximately F7 and F8. The emitter-detector spacing was 50 mm to minimize contamination from the extra-cerebral circulation and maximize signal intensity. The differential pathlength factor, which accounts for the increased distance traveled by light due to scattering, is age-dependent (Duncan et al., <xref rid="B12" ref-type="bibr">1996</xref>). At present, however, no data are available on the actual variation of differential pathlength factor in adults aged above 50 years. Therefore, it was set to 6.61, corresponding to age 50 (Duncan et al., <xref rid="B12" ref-type="bibr">1996</xref>; Claassen et al., <xref rid="B5" ref-type="bibr">2006</xref>).</p></sec><sec id="s2-4"><title>Data processing</title><p>Functional Near-Infrared Spectroscopy data were analyzed using commercially available software (Oxysoft, Artinis Medical Systems, Netherlands). Movement artefacts were kept to a minimum by instructing the participants to refrain from talking, frowning or chewing, to avoid head and body movements, and to sit as still as possible during the experiment. A moving average window of 1 s was applied to the [O<sub>2</sub>Hb] and [HHb] signals to filter out high-frequency noise, including noise of the heart beat frequency. The first three trials (all non-targets) of all conditions were excluded from behavioral and fNIRS data analyses to take the delay of the hemodynamic response into account and to obtain a stable hemodynamic state. The fNIRS signals were biased (set to zero) at the start of the fourth trial of each condition, that is 0-back, 1-back and 2-back. Changes of [O<sub>2</sub>Hb] and [HHb] were recalculated for 180 s from this point (Hoshi et al., <xref rid="B22" ref-type="bibr">2003</xref>). Subsequently, mean relative values of [O<sub>2</sub>Hb] and [HHb] were calculated for the whole task period.</p><p>Behavioral performance was assessed by hit rate, correct rejection rate, and reaction time on target trials. Composite scores were calculated as [hits (%)/reaction time (ms) × 100] to take speed/ accuracy trade-offs into account.</p></sec><sec id="s2-5"><title>Statistical analysis</title><p>Statistical analysis was performed using IBM SPSS Statistics for Windows version 20.0 (IBM Corp., Armonk, NY, USA). Alpha was set at 0.05 for all analyses. Data are presented as mean ± SD. Shapiro-Wilk tests indicated that assumptions of normality were met. The effects of working-memory load on the composite scores were established by a repeated measures ANOVA with factor working-memory load (0-,1-,2-back).</p><p>[O<sub>2</sub>Hb] is considered to be a more robust and reproducible fNIRS parameter than [HHb] (Plichta et al., <xref rid="B30" ref-type="bibr">2006</xref>). Studies have demonstrated that the fMRI BOLD response is more strongly correlated with [O<sub>2</sub>Hb] than with [HHb], which may be due to higher signal-to-noise ratio in [O<sub>2</sub>Hb] (Strangman et al., <xref rid="B43" ref-type="bibr">2002</xref>; Cui et al., <xref rid="B8" ref-type="bibr">2011</xref>). Therefore, taking into account the small sub-sample size, only [O<sub>2</sub>Hb] changes were further statistically analyzed, but results on [HHb] are presented in Figures <xref ref-type="fig" rid="F2">2</xref>, <xref ref-type="fig" rid="F3">3</xref>. For [O<sub>2</sub>Hb] changes, a 2 (location: left, right hemisphere) × 3 (load: 0-,1-,2-back) repeated measures ANOVA was performed. Based on the composite score for the high working-memory load condition, the group was divided by median split into nine low and nine high performers. These two groups did not significantly differ with respect to the variables age, years of education, estimated IQ, and MMSE score (see Table <xref ref-type="table" rid="T1">1</xref>). Accordingly, a 2 (group: low, high performers) × 2 (location: left, right hemisphere) × 3 (load: 0-,1-,2-back) repeated measures ANOVA was performed. Due to violations of the sphericity assumption, Greenhouse-Geisser corrections were applied. Significant main and interaction effects were further analyzed by means of planned contrasts. Due to the small sub-sample size, effects sizes (partial eta squared; <inline-formula><mml:math id="M1"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula>) will be reported as well. <inline-formula><mml:math id="M2"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> ranges from 0 to 1 and it indicates the proportion of variance in the dependent variable that is attributable to the independent variable. An effect size of <inline-formula><mml:math id="M3"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.01 is considered to be small, <inline-formula><mml:math id="M4"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.06 medium, and <inline-formula><mml:math id="M5"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.14 large (Cohen, <xref rid="B6" ref-type="bibr">1988</xref>). To establish the relationship between working-memory load related changes of composite scores and [O<sub>2</sub>Hb] changes, Pearson correlation coefficients (2-tailed) were calculated, corrected for age and years of education.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Hemodynamic concentration changes in the total sample of older adults</bold>. Mean (± SEM) changes of [O<sub>2</sub>Hb] and [HHb] in the left and right hemisphere for the spatial 1-back minus 0-back contrast <bold>(A)</bold>, 2-back minus 0-back contrast <bold>(B)</bold>, and 2-back minus 1-back contrast <bold>(C)</bold>.</p></caption><graphic xlink:href="fnagi-06-00303-g0002"/></fig><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Hemodynamic concentration changes in low and high performers</bold>. Mean (± SEM) changes of [O<sub>2</sub>Hb] and [HHb] for the spatial 1-back minus 0-back contrast, 2-back minus 0-back contrast, and 2-back minus 1-back contrast.<bold> (A)</bold>, <bold>(B)</bold> and <bold>(C)</bold> display the results for low performers. <bold>(D)</bold>, <bold>(E)</bold> and <bold>(F)</bold> display the results for high performers.</p></caption><graphic xlink:href="fnagi-06-00303-g0003"/></fig></sec></sec><sec sec-type="results" id="s3"><title>Results</title><sec id="s3-1"><title>Behavioral performance</title><p>Table <xref ref-type="table" rid="T2">2</xref> shows the behavioral results of the older adults during performance of the n-back tasks. Increased working-memory load led to a declined hit rate (<italic>F</italic><sub>(1.48, 25.14)</sub> = 6.91, <italic>p</italic> = 0.008; 0- vs. 1-back <italic>p</italic> = 0.064; 0- vs. 2-back <italic>p</italic> = 0.001; 1- vs. 2-back <italic>p</italic> = 0.088), and a declined correct rejection rate (<italic>F</italic><sub>(1.19, 20.19)</sub> = 19.35, <italic>p</italic> < 0.001; 0- vs. 1-back <italic>p</italic> = 0.033; 0- vs. 2-back <italic>p</italic> < 0.001; 1- vs. 2-back <italic>p</italic> = 0.001). Also, in comparison with the control condition, both low and high working-memory load led to increased reaction times on target and non-target trials (all <italic>p</italic>-values < 0.001). Furthermore, composite scores were negatively affected by load (<italic>F</italic><sub>(2, 34)</sub> = 30.63, <italic>p</italic> < 0.001; 0- vs. 1-back <italic>p</italic> = 0.001; 0- vs. 2-back <italic>p</italic> < 0.001; 1- vs. 2-back <italic>p</italic> = 0.001).</p><table-wrap id="T2" position="float"><label>Table 2</label><caption><p><bold>Accuracy and reaction times (Mean ± SD) for the spatial n-back tasks</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th rowspan="1" colspan="1"/><th rowspan="1" colspan="1"/><th align="center" rowspan="1" colspan="1">Total (<italic>n</italic> = 18)</th><th align="center" rowspan="1" colspan="1">Low performers (<italic>n</italic> = 9)</th><th align="center" rowspan="1" colspan="1">High performers (<italic>n</italic> = 9)</th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">Hits (%)</td><td align="left" rowspan="1" colspan="1">0-back</td><td align="center" rowspan="1" colspan="1">95.8 ± 8.0</td><td align="center" rowspan="1" colspan="1">96.7 ± 6.0</td><td align="center" rowspan="1" colspan="1">94.8 ± 9.9</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1">1-back</td><td align="center" rowspan="1" colspan="1">91.2 ± 12.0</td><td align="center" rowspan="1" colspan="1">89.5 ± 13.4</td><td align="center" rowspan="1" colspan="1">92.8 ± 10.9</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1">2-back</td><td align="center" rowspan="1" colspan="1">84.0 ± 12.9</td><td align="center" rowspan="1" colspan="1">78.4 ± 11.8</td><td align="center" rowspan="1" colspan="1">89.5 ± 12.0</td></tr><tr><td align="left" rowspan="1" colspan="1">Correct rejections (%)</td><td align="left" rowspan="1" colspan="1">0-back</td><td align="center" rowspan="1" colspan="1">99.2 ± 2.0</td><td align="center" rowspan="1" colspan="1">99.2 ± 2.3</td><td align="center" rowspan="1" colspan="1">99.2 ± 1.6</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1">1-back</td><td align="center" rowspan="1" colspan="1">97.5 ± 2.5</td><td align="center" rowspan="1" colspan="1">97.2 ± 2.3</td><td align="center" rowspan="1" colspan="1">97.9 ± 2.7</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1">2-back</td><td align="center" rowspan="1" colspan="1">89.9 ± 9.0</td><td align="center" rowspan="1" colspan="1">88.1 ± 9.3</td><td align="center" rowspan="1" colspan="1">91.7 ± 8.8</td></tr><tr><td align="left" rowspan="1" colspan="1">RT target (ms)</td><td align="left" rowspan="1" colspan="1">0-back</td><td align="center" rowspan="1" colspan="1">665.9 ± 110.0</td><td align="center" rowspan="1" colspan="1">697.7 ± 104.3</td><td align="center" rowspan="1" colspan="1">634.0 ± 112.0</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1">1-back</td><td align="center" rowspan="1" colspan="1">772.7 ± 148.2</td><td align="center" rowspan="1" colspan="1">842.3 ± 138.8</td><td align="center" rowspan="1" colspan="1">703.1 ± 128.4</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1">2-back</td><td align="center" rowspan="1" colspan="1">1048.5 ± 333.8</td><td align="center" rowspan="1" colspan="1">1316.4 ± 229.2</td><td align="center" rowspan="1" colspan="1">780.7 ± 151.0</td></tr><tr><td align="left" rowspan="1" colspan="1">RT non-target (ms)</td><td align="left" rowspan="1" colspan="1">0-back</td><td align="center" rowspan="1" colspan="1">643.8 ± 87.0</td><td align="center" rowspan="1" colspan="1">665.9 ± 73.3</td><td align="center" rowspan="1" colspan="1">621.7 ± 98.0</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1">1-back</td><td align="center" rowspan="1" colspan="1">743.0 ± 124.6</td><td align="center" rowspan="1" colspan="1">761.2 ± 77.1</td><td align="center" rowspan="1" colspan="1">724.8 ± 162.1</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1">2-back</td><td align="center" rowspan="1" colspan="1">958.5 ± 270.7</td><td align="center" rowspan="1" colspan="1">1143.4 ± 221.5</td><td align="center" rowspan="1" colspan="1">773.6 ± 172.3</td></tr><tr><td align="left" rowspan="1" colspan="1">Composite score</td><td align="left" rowspan="1" colspan="1">0-back</td><td align="center" rowspan="1" colspan="1">14.6 ± 2.0</td><td align="center" rowspan="1" colspan="1">14.1 ± 1.9</td><td align="center" rowspan="1" colspan="1">15.2 ± 2.1</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1">1-back</td><td align="center" rowspan="1" colspan="1">12.2 ± 2.7</td><td align="center" rowspan="1" colspan="1">11.0 ± 2.8</td><td align="center" rowspan="1" colspan="1">13.4 ± 2.0</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1">2-back</td><td align="center" rowspan="1" colspan="1">8.9 ± 3.3</td><td align="center" rowspan="1" colspan="1">6.2 ± 1.4</td><td align="center" rowspan="1" colspan="1">11.7 ± 2.0</td></tr></tbody></table><table-wrap-foot><p><italic>p ≤ 0.005, *p < 0.05 low performers vs. high performers</italic>.</p></table-wrap-foot></table-wrap><p>Investigating the high and low performers separately, both groups showed a significantly decreased correct rejection rate, increased reaction times on targets and non-targets, and a decreased composite score with increased working-memory load (all <italic>p</italic>-values < 0.05). Hit rate declined in low performers (<italic>p</italic> = 0.011), but not in high performers (<italic>p</italic> = 0.261). Table <xref ref-type="table" rid="T2">2</xref> shows the statistically significant group differences on the behavioral parameters.</p></sec><sec id="s3-2"><title>fNIRS results—overall group</title><p>Figure <xref ref-type="fig" rid="F2">2</xref> displays the mean [O<sub>2</sub>Hb] changes for the 1-back minus 0-back contrast, 2-back minus 0-back contrast, and 2-back minus 1-back contrast. Whole-group analysis revealed a significant main effect of load (<italic>F</italic><sub>(2,34)</sub> = 7.99, <italic>p</italic> = 0.001, <inline-formula><mml:math id="M6"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.320; 0- vs. 1-back <italic>p</italic> = 0.039, <inline-formula><mml:math id="M7"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.226; 0- vs. 2-back <italic>p</italic> = 0.002, <inline-formula><mml:math id="M8"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.443), with a trend for 1- vs. 2-back (<italic>p</italic> = 0.063, <inline-formula><mml:math id="M9"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.189). No significant effects were found for the factor location, or the location × load interaction, indicating bilateral activation during task performance. Further analysis confirmed a load effect in both the left fNIRS channel (<italic>F</italic><sub>(2,34)</sub> = 6.41, <italic>p</italic> = 0.004, <inline-formula><mml:math id="M10"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.274; 0- vs. 1-back, <italic>p</italic> = 0.050, <inline-formula><mml:math id="M11"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.208; 0- vs. 2-back <italic>p</italic> = 0.007, <inline-formula><mml:math id="M12"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.352) and the right fNIRS channel (<italic>F</italic><sub>(2,34)</sub> = 8.02, <italic>p</italic> = 0.001, <inline-formula><mml:math id="M13"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.321; 0- vs. 1-back <italic>p</italic> = 0.048, <inline-formula><mml:math id="M14"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.212; 0- vs. 2-back <italic>p</italic> = 0.001, <inline-formula><mml:math id="M15"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.482). Trends were found for 1- vs. 2-back (Left: <italic>p</italic> = 0.069, <inline-formula><mml:math id="M16"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.182; Right: <italic>p</italic> = 0.080, <inline-formula><mml:math id="M17"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.170).</p></sec><sec id="s3-3"><title>fNIRS results—low and high performers</title><p>A large effect size was found for the significant interaction of group × location × load (<italic>F</italic><sub>(2,32)</sub> = 3.55, <italic>p</italic> = 0.041, <inline-formula><mml:math id="M18"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.182). The interaction of group × location showed a trend towards significance and a large effect size (<italic>F</italic><sub>(1,16)</sub> = 3.70, <italic>p</italic> = 0.073, <inline-formula><mml:math id="M19"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.188). These results indicate group differences in prefrontal activation that may not be consistent across tasks and hemispheres.</p><p>The interaction of group × location was further analyzed for each individual condition. For the 0-back task and 1-back task, no significant effects of group, location, or group × location were found, indicating bilateral activation in both groups. For the 2-back task, analyses revealed a significant group × location interaction with a large effect size (<italic>F</italic><sub>(1,16)</sub> = 6.27, <italic>p</italic> = 0.023, <inline-formula><mml:math id="M20"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.282). No significant main effects of location or group were found. Further group comparisons however, showed a large effect size for the right fNIRS channel (<italic>F</italic><sub>(1,16)</sub> = 2.72, <italic>p</italic> = 0.119, <inline-formula><mml:math id="M21"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.145), which may suggest stronger activation in the high performers than in the low performers during 2-back performance. Furthermore, the effect of location showed a trend towards significance with a large effect size in low performers, indicating lower activation in the right hemisphere compared to the left hemisphere (<italic>F</italic><sub>(1,8)</sub> = 4.00, <italic>p</italic> = 0.080, <inline-formula><mml:math id="M22"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.333). In high performers, no significant effect of location was found, but the large effect size may suggest stronger activation in the right hemisphere compared to the left hemisphere during performance of the 2-back task (<italic>F</italic><sub>(1,8)</sub> = 2.36, <italic>p</italic> = 0.163, <inline-formula><mml:math id="M23"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.227).</p><p>The effects of load were analyzed for low- and high-performing elderly separately. Figure <xref ref-type="fig" rid="F3">3</xref> shows the mean [O<sub>2</sub>Hb] changes for the 1-back minus 0-back contrast, 2-back minus 0-back contrast, and 2-back minus 1-back contrast for both groups. In high performers, significant load effects with a large effect size were found for the left fNIRS channel (<italic>F</italic><sub>(2,16)</sub> = 3.77, <italic>p</italic> = 0.046, <inline-formula><mml:math id="M24"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.320; 0- vs. 1-back <italic>p</italic> = 0.049, <inline-formula><mml:math id="M25"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.401; 0- vs. 2-back <italic>p</italic> = 0.028, <inline-formula><mml:math id="M26"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.475; 1- vs. 2-back <italic>p</italic> = 0.467, <inline-formula><mml:math id="M27"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.068) and for the right fNIRS channel (<italic>F</italic><sub>(2,16)</sub> = 7.86, <italic>p</italic> = 0.004, <inline-formula><mml:math id="M28"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.495; 0- vs. 1-back <italic>p</italic> = 0.060, <inline-formula><mml:math id="M29"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.374; 0- vs. 2-back <italic>p</italic> = 0.002, <inline-formula><mml:math id="M30"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.723; 1- vs. 2-back <italic>p</italic> = 0.200, <inline-formula><mml:math id="M31"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.196). In low performers, the effects of load were not statistically significant, but large effect sizes were found for the left fNIRS channel (<italic>F</italic><sub>(2,16)</sub> = 3.05, <italic>p</italic> = 0.075, <inline-formula><mml:math id="M32"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.276; 0- vs. 1-back <italic>p</italic> =0.444, <inline-formula><mml:math id="M33"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.075; 0- vs. 2-back <italic>p</italic> = 0.102, <inline-formula><mml:math id="M34"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.299; 1- vs. 2-back <italic>p</italic> = 0.088, <inline-formula><mml:math id="M35"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.320) and for the right fNIRS channel (<italic>F</italic><sub>(2,16)</sub> = 1.73, <italic>p</italic> = 0.209, <inline-formula><mml:math id="M36"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.178; 0- vs. 1-back <italic>p</italic> = 0.517, <inline-formula><mml:math id="M37"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.054; 0- vs. 2-back <italic>p</italic> = 0.138, <inline-formula><mml:math id="M38"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.254; 1- vs. 2-back <italic>p</italic> = 0.273, <inline-formula><mml:math id="M39"><mml:mrow><mml:msubsup><mml:mi>η</mml:mi><mml:mi>p</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow></mml:math></inline-formula> = 0.147).</p></sec><sec id="s3-4"><title>Relation behavioral performance and fNIRS</title><p>Figure <xref ref-type="fig" rid="F4">4</xref> displays scatterplots illustrating the relationships between cognitive load-induced changes in composite score and [O<sub>2</sub>Hb]. Whole-group analysis demonstrated that individuals with a larger decline in composite score from the 0-back to the 2-back condition had a larger increase of [O<sub>2</sub>Hb] in the left fNIRS channel between these conditions (Left: <italic>r</italic> = −0.504, <italic>p</italic> = 0.046; Right: <italic>r</italic> = −0.158, <italic>p</italic> = 0.559). Individuals with a larger decline in composite score from the 1-back to the 2-back condition showed a larger increase of [O<sub>2</sub>Hb] in the left fNIRS channel (Left: <italic>r</italic> = −0.503, <italic>p</italic> = 0.047; Right: <italic>r</italic> = −0.427, <italic>p</italic> = 0.099).</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>Correlation of δComposite score and δ[O<sub>2</sub>Hb] in low and high performers. (A)</bold> and <bold>(B)</bold> show the 2-back minus 0-back contrast. <bold>(C)</bold> and <bold>(D)</bold> show the 2-back minus 1-back contrast.</p></caption><graphic xlink:href="fnagi-06-00303-g0004"/></fig><p>Within the low performers group, individuals with a larger decline in composite score from the 0-back to the 2-back condition had a larger bilateral increase of [O<sub>2</sub>Hb] between these conditions (Left: <italic>r</italic> = −0.803, <italic>p</italic> = 0.030; Right: <italic>r</italic> = −0.872, <italic>p</italic> = 0.010). Furthermore, low performers with a larger decline in composite score from the 1-back to the 2-back condition showed a larger increase of [O<sub>2</sub>Hb] in the left fNIRS channel (Left: <italic>r</italic> = −0.856, <italic>p</italic> = 0.014; Right: <italic>r</italic> = −0.577, <italic>p</italic> = 0.175). For the high performers group, no significant correlations were found between load-related changes in behavioral performance and hemodynamic changes (2- back minus 0-back: Left: <italic>r</italic> = 0.204, <italic>p</italic> = 0.661; Right: <italic>r</italic> = 0.102, <italic>p</italic> = 0.827; 2-back minus 1-back: Left: <italic>r</italic> = 0.074, <italic>p</italic> = 0.875; Right: <italic>r</italic> = −0.110, <italic>p</italic> = 0.814).</p></sec></sec><sec sec-type="discussion" id="s4"><title>Discussion</title><p>In the present study, fNIRS was used to investigate possibly compensatory brain-behavior mechanisms at older age, by assessing prefrontal activation in low- and high-performing older adults during spatial working-memory performance. As expected, increased working-memory load led to increased prefrontal activation and decreased behavioral performance. Results revealed an interaction between performance level and hemispheric activation, which suggests stronger right prefrontal activation in high performers in comparison to low performers under high cognitive demand. Furthermore, in low performers, a larger decline in task performance with increasing working-memory load condition was associated with a larger bilateral upregulation of prefrontal activation. In high performers, no correlation between behavioral performance and prefrontal activation was found. Taken together, these results support the view that prefrontal activation may not only be modulated by working-memory load, but may also be related to performance level.</p><p>Whole-group analysis revealed an upregulation of left and right prefrontal activation with increasing spatial working-memory load. This pattern of bilateral recruitment is in accordance with the HAROLD model (Cabeza, <xref rid="B2" ref-type="bibr">2002</xref>). Previous neuroimaging studies in the visual-spatial working-memory domain (Petrella et al., <xref rid="B28" ref-type="bibr">2005</xref>; Holtzer et al., <xref rid="B21" ref-type="bibr">2009</xref>; Nagel et al., <xref rid="B25" ref-type="bibr">2009</xref>; Toepper et al., <xref rid="B44" ref-type="bibr">2014</xref>), and the verbal working-memory domain (Mattay et al., <xref rid="B24" ref-type="bibr">2006</xref>; Nyberg et al., <xref rid="B27" ref-type="bibr">2009</xref>; Cappell et al., <xref rid="B4" ref-type="bibr">2010</xref>; Prakash et al., <xref rid="B32" ref-type="bibr">2012</xref>; Sala-Llonch et al., <xref rid="B36" ref-type="bibr">2012</xref>; Vermeij et al., <xref rid="B46" ref-type="bibr">2012</xref>) have shown sensitivity of prefrontal activation to task demand in older adults. However, the shape of the dose-response curve varies among studies; in the current study we found an increase of prefrontal activation up to 2-back in our study, while, for example, Mattay et al. (<xref rid="B24" ref-type="bibr">2006</xref>) found a consistent decrease with load in their n-back study, and Heinzel et al. (<xref rid="B20" ref-type="bibr">2014</xref>) found a tendency towards an inverted U-shape. This variation may depend on factors such as task design, population and task difficulty (Stern et al., <xref rid="B42" ref-type="bibr">2012</xref>). These findings emphasize the need to take behavioral performance level into account when interpreting and comparing neuroimaging data.</p><p>In the current study, activation patterns of high and low performers were compared. We found a significant interaction between performance level and hemispheric activation, indicating that high performers more strongly activated the right prefrontal cortex under high working-memory load than low performers did. Direct comparison of left and right hemispheric activation within each group did not result in significant differences, but the large effect sizes may suggest that low performers show decreased activation in the right hemisphere in comparison to the left hemisphere under high cognitive demand, while the opposite pattern was found in high performers. These performance-specific findings may indicate that the commonly observed bilateral prefrontal activation pattern, which we also found in our whole-group analysis, may in fact obscure the heterogeneity in activation patterns in older adults and may lead to invalid generalizations.</p><p>A prior study on the hemodynamic response to a spatial working-memory challenge (Nagel et al., <xref rid="B25" ref-type="bibr">2009</xref>) found that the dose-response curves of high-performing older adults resembled those of young adults in most investigated regions of interest, showing an increase of activation with load. In contrast, low-performing older adults showed a drop in activation at the highest level of working-memory load. The interaction between performance level and load was present in right dorsolateral prefrontal cortex, but not in the left dorsolateral prefrontal cortex. Sala-Llonch et al. (<xref rid="B36" ref-type="bibr">2012</xref>) demonstrated that high-performing older adults showed stronger activation of the right inferior gyrus than low-performing older adults during performance of a verbal 2-back task. In comparison to young adults, high-performing older adults showed increased bilateral frontal activation and increased connectivity in the right frontoparietal task-related network. Moreover, high performers recruited frontal areas involved in the default network, indicating that recruitment of task-unrelated resources might be part of a successful compensatory mechanism of the aging brain. These results in combination with our findings suggest that recruitment of the right prefrontal cortex may be beneficial for working-memory performance in older adults.</p><p>We further explored the brain-behavior relationship and found that low performers who demonstrated a larger load-induced decline in behavioral performance showed a larger load-induced increase in bilateral prefrontal activation. In high performers, we did not find an association between behavioral performance and prefrontal activation. The lack of significant correlations may be due to the limited range of accuracy scores. Although behavioral performance significantly declined with increasing load in this group, performance was on average very high. Alternatively, the sub-sample size may not have been large enough to detect significant associations, which is a limitation of this study.</p><p>Previous neuroimaging studies on working-memory performance in older adults showed mixed results on brain-behavior correlations. Nagel et al. (<xref rid="B26" ref-type="bibr">2011</xref>) found that BOLD signal changes, induced by increasing verbal working-memory load, were positively correlated with accuracy scores during 3-back performance in the left and right premotor cortex and right posterior parietal cortex, and, at trend level, in the left and right dorsolateral prefrontal cortex. Podell et al. (<xref rid="B31" ref-type="bibr">2012</xref>) reported that caudate activation was associated with improved accuracy on a working-memory task, and that ventrolateral prefrontal activation was associated with shorter reaction times. Nagel et al. (<xref rid="B25" ref-type="bibr">2009</xref>) reported a positive correlation between BOLD signal changes in the left premotor cortex and accuracy scores on a spatial working-memory load task. Furthermore, Toepper et al. (<xref rid="B44" ref-type="bibr">2014</xref>) showed that activation in the dorsolateral prefrontal cortex was positively correlated with the numbers of errors at a low level of spatial working-memory load. No significant correlations were found for higher levels of working-memory load. Several other studies however, failed to find working-memory-related brain-behavior correlations in older adults, possibly in part due to near-ceiling accuracy levels (Emery et al., <xref rid="B13" ref-type="bibr">2008</xref>; Holtzer et al., <xref rid="B21" ref-type="bibr">2009</xref>; Cappell et al., <xref rid="B4" ref-type="bibr">2010</xref>; Piefke et al., <xref rid="B29" ref-type="bibr">2012</xref>). Since we were able to establish significant correlations in a relatively small sub-sample, we argue that the fNIRS [O<sub>2</sub>Hb] signal might be a more sensitive parameter for detecting brain-behavior associations than the fMRI BOLD signal.</p><p>Since our aim was to gain insight into the possibly compensatory brain-behavior mechanisms at older age, we did not include a group of young adults to examine age-related changes in prefrontal activation. However, our results may be congruent with the CRUNCH hypothesis that prefrontal over-recruitment may reflect an age-invariant compensatory mechanism (Reuter-Lorenz and Cappell, <xref rid="B33" ref-type="bibr">2008</xref>). A significant interaction of performance level and hemispheric activation indicated that high performers were better able than the low performers to keep the right prefrontal cortex engaged at high working-memory load. Low performers may have reached the limit of available neural resources, while high performers may have been able to recruit more neural resources. Hence, recruitment of the right prefrontal cortex might contribute to successful working-memory performance in older adults (Sala-Llonch et al., <xref rid="B36" ref-type="bibr">2012</xref>). In contrast, the negative correlation between load-induced changes in activation and performance that was observed in low performers may point towards declined neural efficiency or unsuccessful compensation rather than successful compensation (Cabeza and Dennis, <xref rid="B11" ref-type="bibr">2012</xref>).</p><p>The term “compensation” has been under debate. According to some researchers, compensation reflects the recruitment by older adults of the same brain regions that are recruited by young adults in response to increasing task demand. Older adults may need to recruit these resources at lower levels of task demand, but the cognitive operations that contribute to task performance are age invariant (Reuter-Lorenz and Cappell, <xref rid="B33" ref-type="bibr">2008</xref>; Cappell et al., <xref rid="B4" ref-type="bibr">2010</xref>). According to other researchers, the term “compensation” should only be used in case older adults show recruitment of brain regions that are not recruited by younger adults. Moreover, engagement of these regions should be directly correlated to a better performance in older adults, but not be related to the performance in younger adults (Stern, <xref rid="B41" ref-type="bibr">2002</xref>). Since we only measured activation in the prefrontal cortex, with limited spatial resolution, we are not be able to evaluate whether or not the older adults showed a reorganization of neurocognitive networks. However, in previous studies we observed cognitive load-dependent activation in the same brain region in a group of younger adults (Vermeij et al., <xref rid="B46" ref-type="bibr">2012</xref>, <xref rid="B47" ref-type="bibr">2014</xref>). This would support the age-invariant view of compensatory recruitment.</p><p>Another limitation in this study was that the order of conditions was ascending (0-back, 1-back, 2-back), as is common in neuropsychological assessment of working-memory span, instead of counterbalanced. Given the finding that prefrontal activation increased with load, we consider it unlikely that the order of presentation may have confounded the results.</p><p>Examination of the relationship between possibly compensatory brain activation and behavioral outcomes may provide starting points for the development and evaluation of cognitive training programs. Heinzel et al. (<xref rid="B20" ref-type="bibr">2014</xref>) reported that the BOLD response pattern and accuracy during n-back performance were predictive of behavioral training gain in older adults. Future research should aim to establish the plasticity of prefrontal compensatory mechanisms, by studying healthy older adults as well as those who are in a preclinical stage of dementia. Cognitive training programs are attractive, especially to those who suffer from cognitive problems, but to date it is unknown how interindividual neurocognitive differences contribute to training success.</p><p>To conclude, we observed performance-related differences in prefrontal activation in older adults during working-memory performance. Additional recruitment of the right prefrontal cortex may be beneficial for performance when task demands are high. However, an increase in bilateral prefrontal activation with cognitive load may not always be compensatory. Therefore, for the interpretation of neuroimaging data, individual behavioral performance should be taken into account to be able to distinguish successful and unsuccessful compensation or declined neural efficiency.</p></sec><sec id="s5"><title>Conflict of interest statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec>
Segmentation of abdomen MR images using kernel graph cuts with shape priors	<sec><title>Background</title><p>Abdominal organs segmentation of magnetic resonance (MR) images is an important but challenging task in medical image processing. Especially for abdominal tissues or organs, such as liver and kidney, MR imaging is a very difficult task due to the fact that MR images are affected by intensity inhomogeneity, weak boundary, noise and the presence of similar objects close to each other.</p></sec><sec><title>Method</title><p>In this study, a novel method for tissue or organ segmentation in abdomen MR imaging is proposed; this method combines kernel graph cuts (KGC) with shape priors. First, the region growing algorithm and morphology operations are used to obtain the initial contour. Second, shape priors are obtained by training the shape templates, which were collected from different human subjects with kernel principle component analysis (KPCA) after the registration between all the shape templates and the initial contour. Finally, a new model is constructed by integrating the shape priors into the kernel graph cuts energy function. The entire process aims to obtain an accurate image segmentation.</p></sec><sec><title>Results</title><p>The proposed segmentation method has been applied to abdominal organs MR images. The results showed that a satisfying segmentation without boundary leakage and segmentation incorrect can be obtained also in presence of similar tissues. Quantitative experiments were conducted for comparing the proposed segmentation with other three methods: DRLSE, initial erosion contour and KGC without shape priors. The comparison is based on two quantitative performance measurements: the probabilistic rand index (PRI) and the variation of information (VoI). The proposed method has the highest PRI value (0.9912, 0.9983 and 0.9980 for liver, right kidney and left kidney respectively) and the lowest VoI values (1.6193, 0.3205 and 0.3217 for liver, right kidney and left kidney respectively).</p></sec><sec><title>Conclusion</title><p>The proposed method can overcome boundary leakage. Moreover it can segment liver and kidneys in abdominal MR images without segmentation errors due to the presence of similar tissues. The shape priors based on KPCA was integrated into fully automatic graph cuts algorithm (KGC) to make the segmentation algorithm become more robust and accurate. Furthermore, if a shelter is placed onto the target boundary, the proposed method can still obtain satisfying segmentation results.</p></sec>	<contrib contrib-type="author" equal-contrib="yes" id="A1"><name><surname>Luo</surname><given-names>Qing</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>qing.luo@siat.ac.cn</email></contrib><contrib contrib-type="author" equal-contrib="yes" id="A2"><name><surname>Qin</surname><given-names>Wenjian</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>wj.qin@siat.ac.cn</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Wen</surname><given-names>Tiexiang</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I3">3</xref><email>tx.wen@siat.ac.cn</email></contrib><contrib contrib-type="author" corresp="yes" id="A4"><name><surname>Gu</surname><given-names>Jia</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>jia.gu@siat.ac.cn</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Gaio</surname><given-names>Nikolas</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>nikolas@siat.ac.cn</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Chen</surname><given-names>Shifu</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I3">3</xref><email>sf.chen@siat.ac.cn</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Li</surname><given-names>Ling</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>ling.li@siat.ac.cn</email></contrib><contrib contrib-type="author" id="A8"><name><surname>Xie</surname><given-names>Yaoqin</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>yq.xie@siat.ac.cn</email></contrib>	BioMedical Engineering OnLine	<sec sec-type="intro"><title>Introduction</title><p>The recent development of open magnetic resonance imaging has provided new opportunities for next generation image-guided surgical and interventional applications. Image-guided surgery is a standard surgical procedure for abdominal disorders that can reduce surgical trauma and open surgery burdens [<xref ref-type="bibr" rid="B1">1</xref>]. However, during surgical planning and surgical navigation based on MR images, there are two problems have to be faced: the shape deformation of the organs and the similarity among abdominal organs. For these reasons, an effective and robust algorithm for abdominal organs segmentation is helpful and very important in image-guided surgery and surgical navigation system [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B3">3</xref>].</p><p>Usually, basic image information, such as intensity and gradient, are used for segmentation; Ostu, k-means clustering, region growing, etc. are the most widely used algorithms for segmentation. However, they are not suitable for MR segmentation. Because of its weak boundary, intensity inhomogeneity and noise, the segmentation of MR images is considered a complex procedure [<xref ref-type="bibr" rid="B4">4</xref>]. For this reason, developing different advanced and intelligent algorithms for MR image segmentation has become a research hotspot over the last few years.</p><p>Abdomen MR image segmentation is a challenging task, because majority of tissues in abdomen are soft tissues, the intensities of abdominal tissues are very similar and shape change in a complex way due to respiratory movements [<xref ref-type="bibr" rid="B5">5</xref>]. To provide more information about the tissues and organs in abdomen to the doctors, more effective and robust algorithm to segment abdomen MR images are developed. For example, Hassan et al. [<xref ref-type="bibr" rid="B6">6</xref>] proposed a novel method to segment liver MR images automatically. This algorithm utilizes artificial neural networks and watershed algorithm. Moreover, Sheng et al. [<xref ref-type="bibr" rid="B7">7</xref>] applied a wavelet-based k-means clustering method to segment the human kidney from MRI data set.</p><p>Since the level set method can be performed on a fix Cartesian grid with no need to parameterize these objects, in the past decade it was used more and more frequently in image segmentation [<xref ref-type="bibr" rid="B8">8</xref>-<xref ref-type="bibr" rid="B11">11</xref>]. Moreover, the level set method can represent contours with complex topology and change their topology in a natural way. However, the algorithms converge to a local minimum easily, so the results are easily affected by the initial values. Additionally, the execution time may be very long in some applications, especially with large images and multi-object segmentations [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B13">13</xref>].</p><p>Graph cuts techniques have received considerable attention for their global energy optimal advantages. It is used in more and more image segmentation applications for different medical images, such as MR images [<xref ref-type="bibr" rid="B14">14</xref>]. However, it requires to choose the object and background seeds interactively, implying a time-consuming procedure. Kernel graph cuts is a fully automatic algorithm based on graph cuts proposed by Salah et al. [<xref ref-type="bibr" rid="B15">15</xref>]. It consists of a multi-region image segmentation algorithm based on graph cuts via kernel mapping of the image data. This algorithm is not successful for abdominal organs segmentation due to the weak boundary and surrounding objects with similar intensities. The graph cuts based on active contours method was proposed by Xu et al. [<xref ref-type="bibr" rid="B16">16</xref>]. The method identifies an initial contour around the target and then forms the narrow banded area containing the target via dilation and erosion. The segmentation focuses on image data in the narrow banded area, and separates the abdominal organs with similar intensities. However it cannot solve the weak boundary problem that leads to edge leakage. Integrating the prior information into the segmentation is a popular solution for the weak boundary; the prior information can lead to a more accurate segmentation result according to [<xref ref-type="bibr" rid="B17">17</xref>-<xref ref-type="bibr" rid="B20">20</xref>]. Asem et al. [<xref ref-type="bibr" rid="B21">21</xref>] proposed graphs cuts integrated shape priors, and applied this method to kidney segmentation in abdomen MR images. The result of the segmentation shows that it can overcome the edge leakage, however, it uses the probabilistic model, which increases its complexity; moreover, it also needs interactive operations. Chen et al. [<xref ref-type="bibr" rid="B21">21</xref>] integrate shape information generated from Active Appearance Model (AAM) into graph cuts for abdominal 3D organ segmentation. However, the long computation time and the dependence of initial location limit its applications. Comparing to the linear PCA (principle component analysis), the nonlinear PCA model performs better on problems with nonlinear deformation [<xref ref-type="bibr" rid="B22">22</xref>]. Malcolm et al. [<xref ref-type="bibr" rid="B23">23</xref>] proposed a segmentation model that integrates the KPCA-based shape priors into graph cuts, but the model executes iteratively so it is time-consuming.</p><p>In this paper, we propose a new method, which combines kernel graph cuts with KPCA to segment abdominal organs. First, a seed point is chosen inside the target organ, and then the region growing algorithm is used to obtain approximately segmentation. Second, image morphology operations (dilation and erosion) are used to form an initial contour close to the target organ. Third, KPCA is used to obtain the shape priors by training the relevant shape templates set. At last, the shape priors are integrated into kernel graph cuts to make a better segmentation. The main contribution of this paper is the proposal of novel methods that combine kernel graph cuts algorithm with shape priors for abdominal organs extraction. The shape priors based on KPCA help our algorithm to increase segmentation accuracy while intensity is not sufficient to obtain accurate segmentation results.</p><p>This paper is organized as follows. Section Methodology describes the methodology which gives a description of kernel graph cuts and shape priors based on KPCA and detailed introduction of our proposed method. Section Experiments presents the experiments results obtained using the proposed novel method. Finally, the discussion about the algorithms is presented and conclusions are drawn in Section Discussion and Section Conclusions, respectively.</p></sec><sec sec-type="methods"><title>Methodology</title><p>This section starts by briefly describing the KGC and KPCA.</p><sec><title>Kernel graph cuts</title><p>Graph cuts algorithm was introduced by Boykov et al. [<xref ref-type="bibr" rid="B14">14</xref>] for binary image segmentation application. The purpose is to segment an object from a given image using a set of seeds (object and background) placed by user.</p><p>The graph cuts algorithm aims to cast the energy-based image segmentation problem into a graph structure global min-cut problem. The energy function of graph cuts contains two terms: a region-based term <italic>R(A)</italic>and a boundary term <italic>B(A)</italic>, where <italic>A</italic> stands for an object or background pixel assignment. The region-based term evaluates the penalty for assigning a particular pixel to a given region. The boundary term evaluates the penalty for assigning two neighboring pixels to different regions. These two terms often weight by 0 ≤ <italic>ϵ</italic> ≤ 1 for relative influence, and the energy function is expressed as follows:</p><p><disp-formula id="bmcM1"><label>(1)</label><mml:math id="M1" name="1475-925X-12-124-i1" overflow="scroll"><mml:mrow><mml:mi>E</mml:mi><mml:mfenced open="(" close=")"><mml:mi>A</mml:mi></mml:mfenced><mml:mo>=</mml:mo><mml:mi mathvariant="italic">ϵ</mml:mi><mml:mo>·</mml:mo><mml:mi>R</mml:mi><mml:mfenced open="(" close=")"><mml:mi>A</mml:mi></mml:mfenced><mml:mo>+</mml:mo><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi mathvariant="italic">ϵ</mml:mi></mml:mrow></mml:mfenced><mml:mo>·</mml:mo><mml:mi>B</mml:mi><mml:mfenced open="(" close=")"><mml:mi>A</mml:mi></mml:mfenced></mml:mrow></mml:math></disp-formula></p><p>KGC was proposed by Salah et al. [<xref ref-type="bibr" rid="B15">15</xref>] for automatic segmentation by mapping image data into high dimension through kernel function. Graph cuts method is a supervised algorithm which requires user intervention for choosing seeds (object & background). The proposed energy function contains two terms: an original kernel-induced data term which evaluates the deviation of the mapped image data and a regularization term expressed as a function of the region indices. The energy function is expressed as follows:</p><p><disp-formula id="bmcM2"><label>(2)</label><mml:math id="M2" name="1475-925X-12-124-i2" overflow="scroll"><mml:mrow><mml:mi>E</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:mfenced open="{" close="}"><mml:msub><mml:mi>μ</mml:mi><mml:mi>l</mml:mi></mml:msub></mml:mfenced><mml:mo>,</mml:mo><mml:mi>δ</mml:mi></mml:mrow></mml:mfenced><mml:mo>=</mml:mo><mml:msup><mml:mrow><mml:mstyle displaystyle="true"><mml:munder><mml:mi mathsize="big">∑</mml:mi><mml:mrow><mml:mi>l</mml:mi><mml:mo>∈</mml:mo><mml:mi>L</mml:mi></mml:mrow></mml:munder><mml:mrow><mml:mstyle displaystyle="true"><mml:munder><mml:mi mathsize="big">∑</mml:mi><mml:mrow><mml:mi>p</mml:mi><mml:mo>∈</mml:mo><mml:msub><mml:mi>R</mml:mi><mml:mi>l</mml:mi></mml:msub></mml:mrow></mml:munder><mml:mrow><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>ϕ</mml:mi><mml:mfenced open="(" close=")"><mml:msub><mml:mi>μ</mml:mi><mml:mi>l</mml:mi></mml:msub></mml:mfenced><mml:mo>−</mml:mo><mml:mi>ϕ</mml:mi><mml:mfenced open="(" close=")"><mml:msub><mml:mi>I</mml:mi><mml:mi>p</mml:mi></mml:msub></mml:mfenced></mml:mrow></mml:mfenced></mml:mrow></mml:mstyle></mml:mrow></mml:mstyle></mml:mrow><mml:mrow><mml:mn>2</mml:mn></mml:mrow></mml:msup><mml:mo>+</mml:mo><mml:mi>α</mml:mi><mml:mstyle displaystyle="true"><mml:munder><mml:mi mathsize="big">∑</mml:mi><mml:mrow><mml:mfenced open="{" close="}"><mml:mrow><mml:mi>p</mml:mi><mml:mo>,</mml:mo><mml:mi>q</mml:mi></mml:mrow></mml:mfenced><mml:mo>∈</mml:mo><mml:mi>D</mml:mi></mml:mrow></mml:munder><mml:mrow><mml:mi>r</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>δ</mml:mi><mml:mfenced open="(" close=")"><mml:mi>p</mml:mi></mml:mfenced><mml:mo>,</mml:mo><mml:mi>δ</mml:mi><mml:mfenced open="(" close=")"><mml:mi>q</mml:mi></mml:mfenced></mml:mrow></mml:mfenced></mml:mrow></mml:mstyle></mml:mrow></mml:math></disp-formula></p><p>Where <italic>E</italic>({μ<sub>1</sub>},<italic>δ</italic>) measures kernel-induced non Euclidean distances between the observations and the regions parameters <italic>μ</italic><sub><italic>1</italic></sub>. <italic>φ</italic> is a nonlinear mapping from the observation space <italic>I</italic> to a higher dimensional mapped space <italic>J</italic>, and the radial basis function (gauss function) kernel is used as usual. <italic>a</italic> is a positive factor. <italic>δ</italic> is an indexing function which assigns each point of the image to a region. <italic>l</italic> ∈ <italic>L</italic> Is a pixel label in some finite set of labels <italic>L. P∈R</italic><sub><italic>1</italic></sub> is a pixel in each region which is characterized by one label <italic>l</italic>. R(<italic>δ((p),δ(q,</italic>)) is a smooth regularization function and <italic>D</italic> is a neighborhood set containing all pairs of neighboring pixels <italic>{p,q}</italic>.</p><p>According to the Mercer’s theorem [<xref ref-type="bibr" rid="B24">24</xref>], which states that kernel function can be expressed as a dot product in a high-dimensional space, explicitly the mapping <italic>φ</italic> is not available. Instead, the kernel function is as follow:</p><p><disp-formula id="bmcM3"><label>(3)</label><mml:math id="M3" name="1475-925X-12-124-i3" overflow="scroll"><mml:mrow><mml:mi>K</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>y</mml:mi><mml:mo>,</mml:mo><mml:mi>z</mml:mi></mml:mrow></mml:mfenced><mml:mo>=</mml:mo><mml:mi>ϕ</mml:mi><mml:msup><mml:mfenced open="(" close=")"><mml:mi>y</mml:mi></mml:mfenced><mml:mi>T</mml:mi></mml:msup><mml:mo>·</mml:mo><mml:mi>ϕ</mml:mi><mml:mfenced open="(" close=")"><mml:mi>z</mml:mi></mml:mfenced><mml:mspace width="1.2em"/><mml:mtext/><mml:mo>,</mml:mo><mml:mspace width="1.2em"/><mml:mo>∀</mml:mo><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>y</mml:mi><mml:mo>,</mml:mo><mml:mi>z</mml:mi></mml:mrow></mml:mfenced><mml:mo>∈</mml:mo><mml:msup><mml:mi>I</mml:mi><mml:mn>2</mml:mn></mml:msup></mml:mrow></mml:math></disp-formula></p><p>Substitution of the kernel function gives:</p><p><disp-formula id="bmcM4"><label>(4)</label><mml:math id="M4" name="1475-925X-12-124-i4" overflow="scroll"><mml:mtable columnalign="left"><mml:mtr><mml:mtd><mml:mspace width=".5em"/><mml:msub><mml:mi>J</mml:mi><mml:mi>K</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi>I</mml:mi><mml:mi>p</mml:mi></mml:msub><mml:mo>,</mml:mo><mml:mi>μ</mml:mi></mml:mrow></mml:mfenced></mml:mtd><mml:mtd><mml:mo>=</mml:mo><mml:msup><mml:mfenced open="∥" close="∥"><mml:mrow><mml:mi>ϕ</mml:mi><mml:mfenced open="(" close=")"><mml:msub><mml:mi>I</mml:mi><mml:mi>p</mml:mi></mml:msub></mml:mfenced><mml:mo>−</mml:mo><mml:mi>ϕ</mml:mi><mml:mfenced open="(" close=")"><mml:mi>μ</mml:mi></mml:mfenced></mml:mrow></mml:mfenced><mml:mn>2</mml:mn></mml:msup></mml:mtd></mml:mtr><mml:mtr><mml:mtd/><mml:mtd><mml:mo>=</mml:mo><mml:msup><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>ϕ</mml:mi><mml:mfenced open="(" close=")"><mml:msub><mml:mi>I</mml:mi><mml:mi>p</mml:mi></mml:msub></mml:mfenced><mml:mo>−</mml:mo><mml:mi>ϕ</mml:mi><mml:mfenced open="(" close=")"><mml:mi>μ</mml:mi></mml:mfenced></mml:mrow></mml:mfenced><mml:mi>T</mml:mi></mml:msup><mml:mo>·</mml:mo><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>ϕ</mml:mi><mml:mfenced open="(" close=")"><mml:msub><mml:mi>I</mml:mi><mml:mi>p</mml:mi></mml:msub></mml:mfenced><mml:mo>−</mml:mo><mml:mi>ϕ</mml:mi><mml:mfenced open="(" close=")"><mml:mi>μ</mml:mi></mml:mfenced></mml:mrow></mml:mfenced></mml:mtd></mml:mtr><mml:mtr><mml:mtd/><mml:mtd><mml:mo>=</mml:mo><mml:mi>ϕ</mml:mi><mml:msup><mml:mfenced open="(" close=")"><mml:msub><mml:mi>I</mml:mi><mml:mi>p</mml:mi></mml:msub></mml:mfenced><mml:mi>T</mml:mi></mml:msup><mml:mi>ϕ</mml:mi><mml:mfenced open="(" close=")"><mml:msub><mml:mi>I</mml:mi><mml:mi>p</mml:mi></mml:msub></mml:mfenced><mml:mo>−</mml:mo><mml:mi>ϕ</mml:mi><mml:msup><mml:mfenced open="(" close=")"><mml:mi>μ</mml:mi></mml:mfenced><mml:mi>T</mml:mi></mml:msup><mml:mi>ϕ</mml:mi><mml:mfenced open="(" close=")"><mml:msub><mml:mi>I</mml:mi><mml:mi>p</mml:mi></mml:msub></mml:mfenced><mml:mo>−</mml:mo><mml:mi>ϕ</mml:mi><mml:msup><mml:mfenced open="(" close=")"><mml:msub><mml:mi>I</mml:mi><mml:mi>p</mml:mi></mml:msub></mml:mfenced><mml:mi>T</mml:mi></mml:msup><mml:mi>ϕ</mml:mi><mml:mfenced open="(" close=")"><mml:mi>μ</mml:mi></mml:mfenced><mml:mo>+</mml:mo><mml:mi>ϕ</mml:mi><mml:msup><mml:mfenced open="(" close=")"><mml:mi>μ</mml:mi></mml:mfenced><mml:mi>T</mml:mi></mml:msup><mml:mi>ϕ</mml:mi><mml:mfenced open="(" close=")"><mml:mi>μ</mml:mi></mml:mfenced></mml:mtd></mml:mtr><mml:mtr><mml:mtd/><mml:mtd><mml:mo>=</mml:mo><mml:mi>K</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi>I</mml:mi><mml:mi>p</mml:mi></mml:msub><mml:mo>,</mml:mo><mml:msub><mml:mi>I</mml:mi><mml:mi>p</mml:mi></mml:msub></mml:mrow></mml:mfenced><mml:mo>+</mml:mo><mml:mi>K</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>μ</mml:mi><mml:mo>,</mml:mo><mml:mi>μ</mml:mi></mml:mrow></mml:mfenced><mml:mo>−</mml:mo><mml:mn>2</mml:mn><mml:mi>K</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi>I</mml:mi><mml:mi>p</mml:mi></mml:msub><mml:mo>,</mml:mo><mml:mi>μ</mml:mi></mml:mrow></mml:mfenced><mml:mspace width="2em"/><mml:mo>,</mml:mo><mml:mspace width="3em"/><mml:mi>μ</mml:mi><mml:mo>∈</mml:mo><mml:mfenced open="{" close="}"><mml:msub><mml:mi>μ</mml:mi><mml:mi>l</mml:mi></mml:msub></mml:mfenced></mml:mtd></mml:mtr></mml:mtable></mml:math></disp-formula></p><p>Equation (4) is solved for∥<italic>ϕ</italic>(<italic>I</italic><sub><italic>p</italic></sub>) − <italic>ϕ</italic>(<italic>μ</italic>)∥<sup>2</sup> and substituted in (2). Thus, the kernel-induced energy function is given by:</p><p><disp-formula id="bmcM5"><label>(5)</label><mml:math id="M5" name="1475-925X-12-124-i5" overflow="scroll"><mml:mrow><mml:mi>E</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:mfenced open="{" close="}"><mml:msub><mml:mi>μ</mml:mi><mml:mi>l</mml:mi></mml:msub></mml:mfenced><mml:mo>,</mml:mo><mml:mi>δ</mml:mi></mml:mrow></mml:mfenced><mml:mo>=</mml:mo><mml:mstyle displaystyle="true"><mml:munder><mml:mi mathsize="big">∑</mml:mi><mml:mrow><mml:mi>l</mml:mi><mml:mo>∈</mml:mo><mml:mi>L</mml:mi></mml:mrow></mml:munder><mml:mrow><mml:mstyle displaystyle="true"><mml:munder><mml:mi mathsize="big">∑</mml:mi><mml:mrow><mml:mi>p</mml:mi><mml:mo>∈</mml:mo><mml:msub><mml:mi>R</mml:mi><mml:mi>l</mml:mi></mml:msub></mml:mrow></mml:munder><mml:mrow><mml:msub><mml:mi>J</mml:mi><mml:mi>K</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi>I</mml:mi><mml:mi>p</mml:mi></mml:msub><mml:mo>,</mml:mo><mml:msub><mml:mi>μ</mml:mi><mml:mi>l</mml:mi></mml:msub></mml:mrow></mml:mfenced></mml:mrow></mml:mstyle></mml:mrow></mml:mstyle><mml:mo>+</mml:mo><mml:mi>α</mml:mi><mml:mstyle displaystyle="true"><mml:munder><mml:mi mathsize="big">∑</mml:mi><mml:mrow><mml:mfenced open="{" close="}"><mml:mrow><mml:mi>p</mml:mi><mml:mo>,</mml:mo><mml:mi>q</mml:mi></mml:mrow></mml:mfenced><mml:mo>∈</mml:mo><mml:mi>D</mml:mi></mml:mrow></mml:munder><mml:mrow><mml:mi>r</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>δ</mml:mi><mml:mfenced open="(" close=")"><mml:mi>p</mml:mi></mml:mfenced><mml:mo>,</mml:mo><mml:mi>δ</mml:mi><mml:mfenced open="(" close=")"><mml:mi>q</mml:mi></mml:mfenced></mml:mrow></mml:mfenced></mml:mrow></mml:mstyle></mml:mrow></mml:math></disp-formula></p><p>Now, based on the kernel-induced energy function, the graph cuts algorithm can be executed to segment images more efficiently.</p></sec><sec><title>Kernel principle component analysis</title><p>KPCA is a nonlinear feature extractor performed in the feature space <italic>F</italic>. The basic idea of the method is to map the data from the input space <italic>S</italic> to a feature space <italic>F</italic> via nonlinear map <italic>ϕ</italic>:<italic>S</italic>→<italic>F</italic>. Because KPCA is able to capture nonlinear features in the data comparing to linear PCA, it can be used more effectively if a pre-image of the projection in the feature space is available. Rathi et al. [<xref ref-type="bibr" rid="B22">22</xref>] proposed a novel method to reconstruct a unique approximate pre-image of a feature vector and applied it for statistical shape analysis.</p><p>To form the statistical model of shape space <italic>S</italic>, the pre-image of the projection (in the KPCA space) of a test point <italic>x∈S</italic> should be found, as shown in Figure <xref ref-type="fig" rid="F1">1</xref> from [<xref ref-type="bibr" rid="B25">25</xref>]. Let {<italic>x</italic><sub><italic>1</italic></sub><italic>, X</italic><sub><italic>n</italic></sub>}⊂<italic>S</italic> be a set of aligned training shapes represented by binary mask where 1 is object and 0 is background and spread as vectors.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p>The pre-image problem in KPCA.</p></caption><graphic xlink:href="1475-925X-12-124-1"/></fig><p>First, the <italic>N</italic>×<italic>N</italic> kernel matrix <italic>K</italic> with the radial basis function (gauss function) has to be computed:</p><p><disp-formula id="bmcM6"><label>(6)</label><mml:math id="M6" name="1475-925X-12-124-i6" overflow="scroll"><mml:mrow><mml:msub><mml:mi>K</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mspace width="0.2em"/><mml:mi>j</mml:mi></mml:mrow></mml:msub><mml:mo>=</mml:mo><mml:mi>k</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi>x</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>,</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>j</mml:mi></mml:msub></mml:mrow></mml:mfenced><mml:mo>=</mml:mo><mml:mo>exp</mml:mo><mml:mfenced open="(" close=")"><mml:mfrac><mml:mrow><mml:mo>−</mml:mo><mml:msup><mml:mfenced open="∥" close="∥"><mml:mrow><mml:msub><mml:mi>x</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>j</mml:mi></mml:msub></mml:mrow></mml:mfenced><mml:mn>2</mml:mn></mml:msup></mml:mrow><mml:mrow><mml:mn>2</mml:mn><mml:msup><mml:mi>σ</mml:mi><mml:mn>2</mml:mn></mml:msup></mml:mrow></mml:mfrac></mml:mfenced></mml:mrow></mml:math></disp-formula></p><p>Second, following Eigen decomposition has to be considered :</p><p><disp-formula id="bmcM7"><label>(7)</label><mml:math id="M7" name="1475-925X-12-124-i7" overflow="scroll"><mml:mrow><mml:mi mathvariant="italic">HKH</mml:mi><mml:mo>=</mml:mo><mml:mi mathvariant="italic">U</mml:mi><mml:mi mathsize="big">∑</mml:mi><mml:msup><mml:mi>U</mml:mi><mml:mi>T</mml:mi></mml:msup></mml:mrow></mml:math></disp-formula></p><p><italic>H</italic> is the centering matrix given by <inline-formula><mml:math id="M8" name="1475-925X-12-124-i8" overflow="scroll"><mml:mrow><mml:mi>H</mml:mi><mml:mo>=</mml:mo><mml:mi>C</mml:mi><mml:mo>−</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mi>N</mml:mi></mml:mfrac><mml:mi>c</mml:mi><mml:msup><mml:mi>c</mml:mi><mml:mi>T</mml:mi></mml:msup></mml:mrow></mml:math></inline-formula>,<italic>C</italic> is the <italic>N</italic>×<italic>N</italic> identity matrix, <italic>c</italic> = [11 … 1]<sup><italic>T</italic></sup> is an <italic>N</italic>×1 vector, <italic>U</italic> = [∂<sub>1</sub>, ⋯ ∂<sub><italic>N</italic></sub>]<sup><italic>T</italic></sup> with ∂<sub><italic>i</italic></sub> = [<italic>a</italic><sub><italic>i</italic>1</sub>, …, <italic>a</italic><sub><italic>iN</italic></sub>]<sup><italic>T</italic></sup> is the matrix containing the eigenvectors and Σ = <italic>diag</italic>(<italic>λ</italic><sub>1</sub>, …, <italic>λ</italic><sub><italic>N</italic></sub>) contains the corresponding eigen values.</p><p>Third, given a point <italic>x∈S</italic> one can compute its projection <italic>Pϕ(x)∈F</italic> and a subspace is spanned by the first n eigenvectors given by:</p><p><disp-formula id="bmcM8"><label>(8)</label><mml:math id="M9" name="1475-925X-12-124-i9" overflow="scroll"><mml:mrow><mml:mi mathvariant="italic">Pφ</mml:mi><mml:mfenced open="(" close=")"><mml:mi>x</mml:mi></mml:mfenced><mml:mo>=</mml:mo><mml:mstyle displaystyle="true"><mml:munderover><mml:mi mathsize="big">∑</mml:mi><mml:mrow><mml:mi>k</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>n</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi>β</mml:mi><mml:mi>k</mml:mi></mml:msub><mml:msub><mml:mi>V</mml:mi><mml:mi>k</mml:mi></mml:msub><mml:mo>+</mml:mo><mml:mover accent="true"><mml:mi>φ</mml:mi><mml:mo stretchy="true">¯</mml:mo></mml:mover></mml:mrow></mml:mstyle></mml:mrow></mml:math></disp-formula></p><p>Where <inline-formula><mml:math id="M10" name="1475-925X-12-124-i10" overflow="scroll"><mml:mrow><mml:mover accent="true"><mml:mi>φ</mml:mi><mml:mo stretchy="true">¯</mml:mo></mml:mover><mml:mo>=</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mi>N</mml:mi></mml:mfrac><mml:mstyle displaystyle="true"><mml:msubsup><mml:mi mathsize="big">∑</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>N</mml:mi></mml:msubsup><mml:mrow><mml:mi>φ</mml:mi></mml:mrow></mml:mstyle><mml:mfenced open="(" close=")"><mml:msub><mml:mi>x</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mfenced></mml:mrow></mml:math></inline-formula>; <inline-formula><mml:math id="M11" name="1475-925X-12-124-i11" overflow="scroll"><mml:mover accent="true"><mml:mi>φ</mml:mi><mml:mo stretchy="true">˜</mml:mo></mml:mover></mml:math></inline-formula> is the map centralized by <inline-formula><mml:math id="M12" name="1475-925X-12-124-i12" overflow="scroll"><mml:mrow><mml:mover accent="true"><mml:mi>φ</mml:mi><mml:mo stretchy="true">˜</mml:mo></mml:mover><mml:mfenced open="(" close=")"><mml:mi>x</mml:mi></mml:mfenced><mml:mo>=</mml:mo><mml:mi>φ</mml:mi><mml:mfenced open="(" close=")"><mml:mi>x</mml:mi></mml:mfenced><mml:mo>−</mml:mo><mml:mover accent="true"><mml:mi>φ</mml:mi><mml:mo stretchy="true">¯</mml:mo></mml:mover></mml:mrow></mml:math></inline-formula>. <inline-formula><mml:math id="M13" name="1475-925X-12-124-i13" overflow="scroll"><mml:mrow><mml:msub><mml:mi>V</mml:mi><mml:mi>k</mml:mi></mml:msub><mml:mo>=</mml:mo><mml:mstyle displaystyle="true"><mml:munderover><mml:mi mathsize="big">∑</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>N</mml:mi></mml:munderover><mml:mrow><mml:mfrac><mml:msub><mml:mi>a</mml:mi><mml:mrow><mml:mi>k</mml:mi><mml:mspace width="0.1em"/><mml:mi>i</mml:mi></mml:mrow></mml:msub><mml:msqrt><mml:msub><mml:mi>λ</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:msqrt></mml:mfrac></mml:mrow></mml:mstyle><mml:mover accent="true"><mml:mi>φ</mml:mi><mml:mo stretchy="true">˜</mml:mo></mml:mover><mml:mfenced open="(" close=")"><mml:msub><mml:mi>x</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mfenced></mml:mrow></mml:math></inline-formula> is the <italic>k</italic>th orthogonal eigenvector of the covariance matrix in <italic>F</italic>. The projection of test point <italic>x</italic> in <italic>F</italic> project onto the <italic>k</italic>th component by <italic>B</italic><sub><italic>k</italic></sub>. Then <inline-formula><mml:math id="M14" name="1475-925X-12-124-i14" overflow="scroll"><mml:mrow><mml:msub><mml:mi>β</mml:mi><mml:mi>k</mml:mi></mml:msub><mml:mo>=</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:msqrt><mml:msub><mml:mi>λ</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:msqrt></mml:mfrac><mml:mstyle displaystyle="true"><mml:munderover><mml:mi mathsize="big">∑</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>N</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi>a</mml:mi><mml:mrow><mml:mi>k</mml:mi><mml:mspace width="0.1em"/><mml:mi>i</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mstyle><mml:mover accent="true"><mml:mi>k</mml:mi><mml:mo stretchy="true">˜</mml:mo></mml:mover><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>x</mml:mi><mml:mo>,</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow></mml:mfenced></mml:mrow></mml:math></inline-formula></p><p><disp-formula><mml:math id="M15" name="1475-925X-12-124-i15" overflow="scroll"><mml:mrow><mml:mi>where</mml:mi><mml:mspace width="0.4em"/><mml:mover accent="true"><mml:mi>k</mml:mi><mml:mo stretchy="true">˜</mml:mo></mml:mover><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>x</mml:mi><mml:mo>,</mml:mo><mml:mi>y</mml:mi></mml:mrow></mml:mfenced><mml:mo>=</mml:mo><mml:mo><</mml:mo><mml:mover accent="true"><mml:mi>φ</mml:mi><mml:mo stretchy="true">˜</mml:mo></mml:mover><mml:mfenced open="(" close=")"><mml:mi>x</mml:mi></mml:mfenced><mml:mo>,</mml:mo><mml:mover accent="true"><mml:mi>φ</mml:mi><mml:mo stretchy="true">˜</mml:mo></mml:mover><mml:mfenced open="(" close=")"><mml:mi>y</mml:mi></mml:mfenced><mml:mo>></mml:mo><mml:mo>=</mml:mo><mml:mi>k</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>x</mml:mi><mml:mo>,</mml:mo><mml:mi>y</mml:mi></mml:mrow></mml:mfenced><mml:mo>−</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mi>N</mml:mi></mml:mfrac><mml:msup><mml:mi>c</mml:mi><mml:mi>T</mml:mi></mml:msup><mml:msub><mml:mi>k</mml:mi><mml:mi>x</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mi>N</mml:mi></mml:mfrac><mml:msup><mml:mi>c</mml:mi><mml:mi>T</mml:mi></mml:msup><mml:msub><mml:mi>k</mml:mi><mml:mi>y</mml:mi></mml:msub><mml:mo>+</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:msup><mml:mi>N</mml:mi><mml:mn>2</mml:mn></mml:msup></mml:mfrac><mml:msup><mml:mi>c</mml:mi><mml:mi>T</mml:mi></mml:msup><mml:mi mathvariant="italic">Kc</mml:mi></mml:mrow></mml:math></disp-formula></p><p><disp-formula><mml:math id="M16" name="1475-925X-12-124-i16" overflow="scroll"><mml:mrow><mml:mi>with</mml:mi><mml:mspace width="0.4em"/><mml:msub><mml:mi>k</mml:mi><mml:mi>x</mml:mi></mml:msub><mml:mo>=</mml:mo><mml:msup><mml:mfenced open="[" close="]"><mml:mrow><mml:mi>k</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>x</mml:mi><mml:mo>,</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mn>1</mml:mn></mml:msub></mml:mrow></mml:mfenced><mml:mo>,</mml:mo><mml:mo>…</mml:mo><mml:mo>,</mml:mo><mml:mi>k</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>x</mml:mi><mml:mo>,</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>N</mml:mi></mml:msub></mml:mrow></mml:mfenced></mml:mrow></mml:mfenced><mml:mi>T</mml:mi></mml:msup></mml:mrow></mml:math></disp-formula></p><p>Finally, the method in [<xref ref-type="bibr" rid="B22">22</xref>] is used to compute the approximate pre-image <inline-formula><mml:math id="M17" name="1475-925X-12-124-i17" overflow="scroll"><mml:mover accent="true"><mml:mi>x</mml:mi><mml:mo stretchy="true">^</mml:mo></mml:mover></mml:math></inline-formula>:</p><p><disp-formula id="bmcM9"><label>(9)</label><mml:math id="M18" name="1475-925X-12-124-i18" overflow="scroll"><mml:mrow><mml:mover accent="true"><mml:mi>x</mml:mi><mml:mo stretchy="true">^</mml:mo></mml:mover><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mstyle displaystyle="true"><mml:msubsup><mml:mi mathsize="big">∑</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>N</mml:mi></mml:msubsup><mml:mrow><mml:msub><mml:mover accent="true"><mml:mi>γ</mml:mi><mml:mo stretchy="true">˜</mml:mo></mml:mover><mml:mi>i</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:mstyle scriptlevel="+1"><mml:mfrac><mml:mn>1</mml:mn><mml:mn>2</mml:mn></mml:mfrac></mml:mstyle><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>2</mml:mn><mml:mo>−</mml:mo><mml:msup><mml:mfenced open="∥" close="∥"><mml:mrow><mml:mi>φ</mml:mi><mml:mfenced open="(" close=")"><mml:msub><mml:mi>x</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mfenced><mml:mo>−</mml:mo><mml:mi mathvariant="italic">Pφ</mml:mi><mml:mfenced open="(" close=")"><mml:mi>x</mml:mi></mml:mfenced></mml:mrow></mml:mfenced><mml:mn>2</mml:mn></mml:msup></mml:mrow></mml:mfenced></mml:mrow></mml:mfenced></mml:mrow></mml:mstyle><mml:msub><mml:mi>x</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:mstyle displaystyle="true"><mml:msubsup><mml:mi mathsize="big">∑</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>N</mml:mi></mml:msubsup><mml:mrow><mml:msub><mml:mover accent="true"><mml:mi>γ</mml:mi><mml:mo stretchy="true">˜</mml:mo></mml:mover><mml:mi>i</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:mstyle scriptlevel="+1"><mml:mfrac><mml:mn>1</mml:mn><mml:mn>2</mml:mn></mml:mfrac></mml:mstyle><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>2</mml:mn><mml:mo>−</mml:mo><mml:msup><mml:mfenced open="∥" close="∥"><mml:mrow><mml:mi>φ</mml:mi><mml:mfenced open="(" close=")"><mml:msub><mml:mi>x</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mfenced><mml:mo>−</mml:mo><mml:mi mathvariant="italic">Pφ</mml:mi><mml:mfenced open="(" close=")"><mml:mi>x</mml:mi></mml:mfenced></mml:mrow></mml:mfenced><mml:mn>2</mml:mn></mml:msup></mml:mrow></mml:mfenced></mml:mrow></mml:mfenced></mml:mrow></mml:mstyle></mml:mrow></mml:mfrac></mml:mrow></mml:math></disp-formula></p><p>And the distance used in feature space in equation (9) is defined as follows:</p><p><disp-formula id="bmcM10"><label>(10)</label><mml:math id="M19" name="1475-925X-12-124-i19" overflow="scroll"><mml:mrow><mml:msup><mml:mfenced open="∥" close="∥"><mml:mrow><mml:mi>φ</mml:mi><mml:mfenced open="(" close=")"><mml:msub><mml:mi>x</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mfenced><mml:mo>−</mml:mo><mml:mi mathvariant="italic">Pφ</mml:mi><mml:mfenced open="(" close=")"><mml:mi>x</mml:mi></mml:mfenced></mml:mrow></mml:mfenced><mml:mn>2</mml:mn></mml:msup><mml:mo>=</mml:mo><mml:msup><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mover accent="true"><mml:mi>k</mml:mi><mml:mo stretchy="true">˜</mml:mo></mml:mover><mml:mi>x</mml:mi></mml:msub><mml:mo>+</mml:mo><mml:mn>2</mml:mn><mml:mi>H</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:mstyle scriptlevel="+1"><mml:mfrac><mml:mn>1</mml:mn><mml:mi>N</mml:mi></mml:mfrac></mml:mstyle><mml:mi>K</mml:mi><mml:mspace width="0.12em"/><mml:mi>c</mml:mi><mml:mo>−</mml:mo><mml:msub><mml:mi>k</mml:mi><mml:mrow><mml:mi>x</mml:mi><mml:mspace width="0.12em"/><mml:mi>i</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mfenced></mml:mrow></mml:mfenced><mml:mi>T</mml:mi></mml:msup><mml:mi>M</mml:mi><mml:msub><mml:mover accent="true"><mml:mi>k</mml:mi><mml:mo stretchy="true">˜</mml:mo></mml:mover><mml:mi>x</mml:mi></mml:msub><mml:mo>+</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:msup><mml:mi>N</mml:mi><mml:mn>2</mml:mn></mml:msup></mml:mfrac><mml:msup><mml:mi>c</mml:mi><mml:mi>T</mml:mi></mml:msup><mml:mi>K</mml:mi><mml:mspace width="0.12em"/><mml:mi>c</mml:mi><mml:mo>+</mml:mo><mml:msub><mml:mi>K</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mspace width="0.12em"/><mml:mi>i</mml:mi></mml:mrow></mml:msub><mml:mo>−</mml:mo><mml:mfrac><mml:mn>2</mml:mn><mml:mi>N</mml:mi></mml:mfrac><mml:msup><mml:mi>c</mml:mi><mml:mi>T</mml:mi></mml:msup><mml:msub><mml:mi>k</mml:mi><mml:mrow><mml:mi>x</mml:mi><mml:mspace width="0.12em"/><mml:mi>i</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:math></disp-formula></p><p>where <inline-formula><mml:math id="M20" name="1475-925X-12-124-i20" overflow="scroll"><mml:mrow><mml:mi>M</mml:mi><mml:mo>=</mml:mo><mml:mstyle displaystyle="true"><mml:munderover><mml:mi mathsize="big">∑</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>N</mml:mi></mml:munderover><mml:mrow><mml:mfrac><mml:mn>1</mml:mn><mml:msub><mml:mi>λ</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mfrac></mml:mrow></mml:mstyle><mml:msub><mml:mo>∂</mml:mo><mml:mi>k</mml:mi></mml:msub><mml:msup><mml:msub><mml:mo>∂</mml:mo><mml:mi>k</mml:mi></mml:msub><mml:mi>T</mml:mi></mml:msup></mml:mrow></mml:math></inline-formula>, <italic>γ</italic> = [∂<sub>1</sub>, …, ∂<sub><italic>n</italic></sub>]<italic>β</italic>, <inline-formula><mml:math id="M21" name="1475-925X-12-124-i21" overflow="scroll"><mml:mrow><mml:mover accent="true"><mml:mi>γ</mml:mi><mml:mo stretchy="true">˜</mml:mo></mml:mover><mml:mo>=</mml:mo><mml:mi>γ</mml:mi><mml:mo>+</mml:mo><mml:mfrac><mml:mn>1</mml:mn><mml:mi>N</mml:mi></mml:mfrac><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:msup><mml:mi>c</mml:mi><mml:mi>T</mml:mi></mml:msup><mml:mi>γ</mml:mi></mml:mrow></mml:mfenced></mml:mrow></mml:math></inline-formula>.</p><p>The pre-image <inline-formula><mml:math id="M22" name="1475-925X-12-124-i22" overflow="scroll"><mml:mover accent="true"><mml:mi>x</mml:mi><mml:mo stretchy="true">^</mml:mo></mml:mover></mml:math></inline-formula> contains the information that is used to form the shape priors. In following section, the proposed method is presented and summarized.</p></sec><sec><title>Proposed method</title><p>To mitigate the effect of weak boundary in MR images and to segment abdominal organs from surrounding objects with similar intensities, a novel segmentation method is proposed. The method incorporates KPCA with KGC enlightened by [<xref ref-type="bibr" rid="B23">23</xref>]. KPCA is used to form the shape priors based on the statistical model proposed by Rathi et al. [<xref ref-type="bibr" rid="B22">22</xref>]. Enlightened by the GCBAC (graph cuts based Active Contours) proposed by Xu et al. [<xref ref-type="bibr" rid="B16">16</xref>], the initial dilation and erosion contour idea is introduced into the method to locate the position of shape priors. The framework of the segmentation method is shown in Figure <xref ref-type="fig" rid="F2">2</xref>.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p>The proposed segmentation framework.</p></caption><graphic xlink:href="1475-925X-12-124-2"/></fig><p>The proposed segmentation framework can be described in two phases.</p><p>Phase I is the pre-segmentation phase. It includes two procedures to form the image data.</p><p>First, to form the initial contour, a seed point is chosen inside the target region manually, and then the region growing algorithm is used to segment the target region approximately. Based on the result of region growing algorithm, a lot of isolated small regions in the target region are not segmented correctly. For this reason, the morphological dilate and erode operation are executed over the pre-segmented contour using the region growing algorithms. The dilate operation aims to eliminate those isolated small regions and to form a continuous contour around the target region; the erode operation aims to draw the contour near the real target region after the dilate operation. Same morphological structuring element is used for the operation of erode and dilate. As a result, the initial contour is obtained. However, because of the weak boundary in the MR images, a further processing is needed to overcome this problem to reach a more accurate segmentation.</p><p>Second step includes obtaining the shape priors. Because of the differences varying from person to person, and different parameters of MR imaging, the shapes of abdominal organs to be segmented differ from each other. Moreover, the respiratory movements make the deformation more complex. Since the deformation is nonlinear, the KPCA is used to train the shape templates determined by experts. Before training the data set, all the shape templates and the initial contour should be aligned. So the image registration is needed here, and only translating, scaling and rotating transforms are taken into consideration during the registration process. Suppose that <italic>X</italic><sub><italic>i</italic></sub> is the vector of one shape template, one of the shape templates <italic>Xj</italic> is to be chosen, the fittest parameters can be obtained through translating <italic>tj</italic>, scaling and rotating transform <italic>M(s</italic><sub><italic>j</italic></sub><italic>,θ</italic><sub><italic>j</italic></sub><italic>)</italic>. The target function is obtained by minimizing the error measure <italic>Ej</italic>:</p><p><disp-formula id="bmcM11"><label>(11)</label><mml:math id="M23" name="1475-925X-12-124-i23" overflow="scroll"><mml:mrow><mml:msub><mml:mi>E</mml:mi><mml:mi>j</mml:mi></mml:msub><mml:mo>=</mml:mo><mml:msup><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi>X</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:mi>M</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi>s</mml:mi><mml:mi>j</mml:mi></mml:msub><mml:mo>,</mml:mo><mml:msub><mml:mi>θ</mml:mi><mml:mi>j</mml:mi></mml:msub></mml:mrow></mml:mfenced><mml:mfenced open="[" close="]"><mml:msub><mml:mi>X</mml:mi><mml:mi>j</mml:mi></mml:msub></mml:mfenced><mml:mo>−</mml:mo><mml:msub><mml:mi>t</mml:mi><mml:mi>j</mml:mi></mml:msub></mml:mrow></mml:mfenced><mml:mi>T</mml:mi></mml:msup><mml:mi>W</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi>X</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>−</mml:mo><mml:mi>M</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi>s</mml:mi><mml:mi>j</mml:mi></mml:msub><mml:mo>,</mml:mo><mml:msub><mml:mi>θ</mml:mi><mml:mi>j</mml:mi></mml:msub></mml:mrow></mml:mfenced><mml:mfenced open="[" close="]"><mml:msub><mml:mi>X</mml:mi><mml:mi>j</mml:mi></mml:msub></mml:mfenced><mml:mo>−</mml:mo><mml:msub><mml:mi>t</mml:mi><mml:mi>j</mml:mi></mml:msub></mml:mrow></mml:mfenced></mml:mrow></mml:math></disp-formula></p><p><italic>W</italic> is the weight matrix, and the least square method is usually used to solve it.</p><p>Commonly, most applications, such as feature extraction and pattern classification, only need the new features generated by KPCA. However, for some other applications, reconstructing the pre-image from the KPCA features is needed. In this case KPCA feature is not necessary to describe the deformation patterns; on the contrary, it is required to reconstruct the shapes from the KPCA features. For a Gaussian kernel, the pre-image <inline-formula><mml:math id="M24" name="1475-925X-12-124-i24" overflow="scroll"><mml:mover accent="true"><mml:mi>x</mml:mi><mml:mo stretchy="true">^</mml:mo></mml:mover></mml:math></inline-formula> can be obtained using Equations (6-10).</p><p>Phase II is the segmentation phase. Since the registration is composed of translating, scaling and rotating transform, and the abdominal organs change from patient to patient, the result of registration contour does not represent the real boundary of MR images, and a more accurate segmentation procedure need to be followed. The image data constructs the graph using energy function (5) in KGC, introducing the shape priors <inline-formula><mml:math id="M25" name="1475-925X-12-124-i25" overflow="scroll"><mml:mover accent="true"><mml:mi>x</mml:mi><mml:mo stretchy="true">^</mml:mo></mml:mover></mml:math></inline-formula> into the data term to overcome the weak boundary. This work aims to use the data term for representing the penalty of pixel attribute to the object or background. Thus, it is assumed that non-uniform shape priors <italic>Pp</italic>(<italic>O</italic>) and P<sub>p</sub>(B) represent penalty of the shape priors attribute to the object or background at a pixel <italic>P</italic>. A parameter <italic>η</italic>(0≤<italic>η</italic>≤1) is also introduced to represent the weight of relative influence between kernel-induced data term <italic>J</italic><sub><italic>K</italic></sub> and shape priors, so the new data terms can be written as follows:</p><p><disp-formula id="bmcM12"><label>(12)</label><mml:math id="M26" name="1475-925X-12-124-i26" overflow="scroll"><mml:mrow><mml:msub><mml:mi>R</mml:mi><mml:mi>p</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mi>O</mml:mi></mml:mfenced><mml:mo>=</mml:mo><mml:mi>η</mml:mi><mml:mo>·</mml:mo><mml:msub><mml:mi>J</mml:mi><mml:mi>K</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi>I</mml:mi><mml:mi>p</mml:mi></mml:msub><mml:mo>,</mml:mo><mml:msub><mml:mi>μ</mml:mi><mml:mi>O</mml:mi></mml:msub></mml:mrow></mml:mfenced><mml:mo>+</mml:mo><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>η</mml:mi></mml:mrow></mml:mfenced><mml:mo>·</mml:mo><mml:msub><mml:mi>P</mml:mi><mml:mi>p</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mi>O</mml:mi></mml:mfenced></mml:mrow></mml:math></disp-formula></p><p><disp-formula id="bmcM13"><label>(13)</label><mml:math id="M27" name="1475-925X-12-124-i27" overflow="scroll"><mml:mrow><mml:msub><mml:mi>R</mml:mi><mml:mi>p</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mi>B</mml:mi></mml:mfenced><mml:mo>=</mml:mo><mml:mi>η</mml:mi><mml:mo>·</mml:mo><mml:msub><mml:mi>J</mml:mi><mml:mi>K</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi>I</mml:mi><mml:mi>p</mml:mi></mml:msub><mml:mo>,</mml:mo><mml:msub><mml:mi>μ</mml:mi><mml:mi>B</mml:mi></mml:msub></mml:mrow></mml:mfenced><mml:mo>+</mml:mo><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>η</mml:mi></mml:mrow></mml:mfenced><mml:mo>·</mml:mo><mml:msub><mml:mi>P</mml:mi><mml:mi>p</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mi>B</mml:mi></mml:mfenced></mml:mrow></mml:math></disp-formula></p><p>Since the pre-image has value between 0 and 1, <italic>P</italic><sub><italic>p</italic></sub><italic>(O)</italic> is directly used to represent <inline-formula><mml:math id="M28" name="1475-925X-12-124-i28" overflow="scroll"><mml:mover accent="true"><mml:mi>x</mml:mi><mml:mo stretchy="true">^</mml:mo></mml:mover></mml:math></inline-formula> and set <italic>Pp = (</italic><bold>1-</bold><italic>P</italic><sub><italic>p</italic></sub><italic>(O))</italic>. The smooth term used the original term. At last, based on Equation (5), the new energy function is given by:</p><p><disp-formula id="bmcM14"><label>(14)</label><mml:math id="M29" name="1475-925X-12-124-i29" overflow="scroll"><mml:mrow><mml:mi>E</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:mfenced open="{" close="}"><mml:msub><mml:mi>μ</mml:mi><mml:mi>l</mml:mi></mml:msub></mml:mfenced><mml:mo>,</mml:mo><mml:mi>δ</mml:mi></mml:mrow></mml:mfenced><mml:mo>=</mml:mo><mml:mstyle displaystyle="true"><mml:munder><mml:mi mathsize="big">∑</mml:mi><mml:mrow><mml:mi>l</mml:mi><mml:mo>∈</mml:mo><mml:mi>L</mml:mi></mml:mrow></mml:munder><mml:mrow><mml:mstyle displaystyle="true"><mml:munder><mml:mi mathsize="big">∑</mml:mi><mml:mrow><mml:mi>p</mml:mi><mml:mo>∈</mml:mo><mml:msub><mml:mi>R</mml:mi><mml:mi>l</mml:mi></mml:msub></mml:mrow></mml:munder><mml:mrow><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>η</mml:mi><mml:mo>·</mml:mo><mml:msub><mml:mi>J</mml:mi><mml:mi>K</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi>I</mml:mi><mml:mi>p</mml:mi></mml:msub><mml:mo>,</mml:mo><mml:msub><mml:mi>μ</mml:mi><mml:mi>l</mml:mi></mml:msub></mml:mrow></mml:mfenced><mml:mo>+</mml:mo><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>1</mml:mn><mml:mo>−</mml:mo><mml:mi>η</mml:mi></mml:mrow></mml:mfenced><mml:mo>·</mml:mo><mml:msub><mml:mi>R</mml:mi><mml:mi>p</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mi>l</mml:mi></mml:mfenced></mml:mrow></mml:mfenced></mml:mrow></mml:mstyle><mml:mo>+</mml:mo><mml:mi>α</mml:mi><mml:mstyle displaystyle="true"><mml:munder><mml:mi mathsize="big">∑</mml:mi><mml:mrow><mml:mfenced open="{" close="}"><mml:mrow><mml:mi>p</mml:mi><mml:mo>,</mml:mo><mml:mi>q</mml:mi></mml:mrow></mml:mfenced><mml:mo>∈</mml:mo><mml:mi>D</mml:mi></mml:mrow></mml:munder><mml:mrow><mml:mi>r</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>δ</mml:mi><mml:mfenced open="(" close=")"><mml:mi>p</mml:mi></mml:mfenced><mml:mo>,</mml:mo><mml:mi>δ</mml:mi><mml:mfenced open="(" close=")"><mml:mi>q</mml:mi></mml:mfenced></mml:mrow></mml:mfenced></mml:mrow></mml:mstyle></mml:mrow></mml:mstyle></mml:mrow></mml:math></disp-formula></p><p>Where <italic>l</italic> must be <italic>O</italic> or <italic>B</italic>, which stands for object or background. Thus, the multi-region segmentation of KGC turns to be the binary segmentation of object or background. The new energy function is used to construct graph, compute min-cut and get the segmentation.</p></sec></sec><sec><title>Experiments</title><p>Segmentation is performed on abdominal organs in the abdomen MR images of T1 sequence, and the proposed novel segmentation method is validated using the MATLAB 7.11 program on Windows XP with Lenovo PC with Intel (R) Core (TM) 2 Duo CPU, E7500, and tested using the liver and kidney dataset in abdomen MR images with size of 462 × 310 pixels. All MR images are obtained by Siemens 3.0T MR imaging equipment and all the shape templates are segmented manually by different experts. The size of train set is 30.</p><sec><title>Validation</title><p>First, the original MR images have to be segmented as shown in Figure <xref ref-type="fig" rid="F3">3</xref>. Figure <xref ref-type="fig" rid="F3">3</xref> (a) is used to segment liver in abdominal organs, and Figure <xref ref-type="fig" rid="F3">3</xref> (b) is used to segment both of the two kidneys in abdominal organs.</p><fig id="F3" position="float"><label>Figure 3</label><caption><p>Original image, (a)the abdomen MR image of liver; (b)the abdomen MR image of kidney.</p></caption><graphic xlink:href="1475-925X-12-124-3"/></fig><p>Second, a seed point is placed inside the liver and kidney region in the MR images, and then the region growing algorithm is used to segment the image approximately. The result of region growing algorithm is shown in Figure <xref ref-type="fig" rid="F4">4</xref>. From the Figure <xref ref-type="fig" rid="F4">4</xref>, it can be seen that many isolated small regions are not segmented, and boundary leakage and incorrect segmentation are observed. Many blood vessels exist inside the organ tissues and the noise caused by MR equipment lead to lots of isolated small regions; meanwhile the soft tissues are very similar and the overlaps between different soft tissues lead to weak boundary. So, more work is needed to solve this problem.</p><fig id="F4" position="float"><label>Figure 4</label><caption><p>Region growing algorithm, (a) liver; (b) right kidney; (c) left kidney.</p></caption><graphic xlink:href="1475-925X-12-124-4"/></fig><p>Third, the morphological dilate and erode operation are implemented to fix the problem caused by region growing algorithm. The result contour of region growing is dilated and then eroded as demonstrated in Figure <xref ref-type="fig" rid="F5">5</xref>, the yellow contour is the result of dilation and the blue contour is the result of erosion. The type of morphological structuring element that is used in dilation and erosion is a disk whose radius can be adjusted depending on the result of region growing algorithm. In this experiment, the size of radius is 15 pixels in liver segmentation, and 10 pixels in kidney segmentation. Same morphological structuring element is used in the same MR image during the dilation and the erosion operation. The blue contour is close to the real boundary of abdominal organs as can be seen in Figure <xref ref-type="fig" rid="F5">5</xref>, but boundary leakage is still observable.</p><fig id="F5" position="float"><label>Figure 5</label><caption><p>Dilation (yellow contour) and erosion (blue contour), (a) liver(radius = 15); (b) right kidney(radius = 10); (c) left kidney(radius = 10).</p></caption><graphic xlink:href="1475-925X-12-124-5"/></fig><p>Finally, the shape priors are added into the segmentation to guide the contour more accurate to the real boundary. The size of the training shape template set is 30, and a registration is performed between initial erosion contour (blue contour) and all shape templates (green contour). The result is shown in Figure <xref ref-type="fig" rid="F6">6</xref>. Figure <xref ref-type="fig" rid="F7">7</xref> demonstrates the shape priors which is obtained by training the shape template set using KPCA method. Finally, KGC is combined with shape priors to segment the region inside initial dilation contour (yellow contour) based on function (14). The result of segmentation is shown in Figure <xref ref-type="fig" rid="F8">8</xref>, from which we can find the boundary leakage is eliminated and the segmentation becomes more correct and accurate.</p><fig id="F6" position="float"><label>Figure 6</label><caption><p>Shape template set and initial contour registration, (a) liver; (b) right kidney; (c) left kidney.</p></caption><graphic xlink:href="1475-925X-12-124-6"/></fig><fig id="F7" position="float"><label>Figure 7</label><caption><p>Shape priors, (a) liver; (b) right kidney; (c) left kidney.</p></caption><graphic xlink:href="1475-925X-12-124-7"/></fig><fig id="F8" position="float"><label>Figure 8</label><caption><p><bold>Segmentation results of proposed novel method, (a) liver (</bold><bold> <italic>η</italic> </bold><bold>=0.65); (b) right kidney (</bold><bold> <italic>η</italic> </bold><bold>=0.72); (c) left kidney (</bold><bold> <italic>η</italic> </bold><bold>=0.75).</bold></p></caption><graphic xlink:href="1475-925X-12-124-8"/></fig></sec><sec><title>Quantitative verification</title><p>In order to verify the proposed segmentation method, quantitatively experiments are performed to compare our method and other three methods: DRLSE [<xref ref-type="bibr" rid="B11">11</xref>], initial erosion contour, KGC in the initial dilation contour without shape priors. The comparison is based on two quantitative performance measures: the probabilistic rand index (PRI) and the variation of information (VoI) [<xref ref-type="bibr" rid="B26">26</xref>-<xref ref-type="bibr" rid="B29">29</xref>].</p><p>The PRI counts the fraction of pairs of pixels whose labels are consistent between the computed segmentation and the ground truth. The VoI metric defines the distance between two segmentations like the average conditional entropy. Since the result segmentation boundaries of the proposed method are very close to the ground truth, visible difference between them cannot be identified, for this reason the PRI and VoI are used to quantify the segmentation results.</p><p>For each segmentation method, a higher value of PRI and a lower value of VoI imply that the segmentation results are closer to the expert manual segmentation. The statistic data are illustrated in Table <xref ref-type="table" rid="T1">1</xref>, and the results of different segmentation algorithms are shown in Figure <xref ref-type="fig" rid="F9">9</xref>. As can be seen in the Table <xref ref-type="table" rid="T1">1</xref>, no matter whether the abdominal organ is liver, right kidney or left kidney, the proposed method has the highest PRI values and the lowest VoI values. As shown in Figure <xref ref-type="fig" rid="F9">9</xref>, the segmentation with the proposed method has better performance than the other methods. The KGC with shape priors based on KPCA can overcome the boundary leakage and segment every abdominal organ independently without incorrect segmentation of the similar tissues. It indicates that both for liver and for kidney segmentation, the proposed method is better than the other methods.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>The PRI and VI of different methods in liver and kidneys</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th rowspan="2" align="center" valign="top"><bold>Measures</bold></th><th colspan="3" align="center" valign="bottom"><bold>PRI</bold><hr/></th><th colspan="3" align="center" valign="bottom"><bold>VoI</bold><hr/></th></tr><tr><th align="center"><bold>Liver</bold></th><th align="center"><bold>Right kidney</bold></th><th align="center"><bold>Left kidney</bold></th><th align="center"><bold>Liver</bold></th><th align="center"><bold>Right kidney</bold></th><th align="center"><bold>Left kidney</bold></th></tr></thead><tbody valign="top"><tr><td align="center" valign="bottom">DRLSE<hr/></td><td align="center" valign="bottom">0.9808<hr/></td><td align="center" valign="bottom">0.9911<hr/></td><td align="center" valign="bottom">0.9930<hr/></td><td align="center" valign="bottom">1.6648<hr/></td><td align="center" valign="bottom">0.3823<hr/></td><td align="center" valign="bottom">0.3595<hr/></td></tr><tr><td align="center" valign="bottom">Initial erosion contour<hr/></td><td align="center" valign="bottom">0.9155<hr/></td><td align="center" valign="bottom">0.9906<hr/></td><td align="center" valign="bottom">0.9917<hr/></td><td align="center" valign="bottom">1.8772<hr/></td><td align="center" valign="bottom">0.3485<hr/></td><td align="center" valign="bottom">0.3311<hr/></td></tr><tr><td align="center" valign="bottom">KGC in the initial dilation contour without shape priors<hr/></td><td align="center" valign="bottom">0.8903<hr/></td><td align="center" valign="bottom">0.9808<hr/></td><td align="center" valign="bottom">0.9913<hr/></td><td align="center" valign="bottom">2.1212<hr/></td><td align="center" valign="bottom">0.5351<hr/></td><td align="center" valign="bottom">0.3965<hr/></td></tr><tr><td align="center">The proposed method</td><td align="center">0.9912</td><td align="center">0.9983</td><td align="center">0.9980</td><td align="center">1.6193</td><td align="center">0.3205</td><td align="center">0.3217</td></tr></tbody></table></table-wrap><fig id="F9" position="float"><label>Figure 9</label><caption><p><bold>The results of different algorithms on liver, right kidney and left kidney from left to right of each row in order, (a-c) DRLSE; (d-f) initial erosion contour; (g-i) KGC in the initial dilation contour without shape priors; (j-l)the proposed method (</bold><bold> <italic>η</italic> </bold><bold>=0.65, 0.72, 0.75).</bold></p></caption><graphic xlink:href="1475-925X-12-124-9"/></fig></sec><sec><title>Parameter adjustment</title><p>The value of parameter represents the relative influence of KGC data term and shape priors term. In Figure <xref ref-type="fig" rid="F8">8</xref>, the value of <italic>η</italic> increases gradually from liver to right kidney. Since the incorrect segmentation in the Figure <xref ref-type="fig" rid="F9">9</xref>(g-i) decreases gradually from liver to right kidney, the KGC data term weighted in the energy function become high. If <italic>η</italic> has a high value, the weight of KGC data term is high too and the weight of shape priors term is small. For different organs different organs, the value of <italic>η</italic> can be adjusted to optimize segmentation performance.</p><p>As shown in Figure <xref ref-type="fig" rid="F10">10</xref>, when =0.65, 0.72 and 0.75, the segmentation of liver, right kidney and left kidney reach the best accuracy. From the third row of Table <xref ref-type="table" rid="T1">1</xref>, the value of PRI is 0.8903, 0.9808 and 0.9913, and the value of VoI is 2.1212, 0.5351 and 0.3965 from liver to right kidney when using KGC inside initial dilation contour without shape priors. In another words, the segmentation error decreases from liver to right kidney, as also shown in Figure <xref ref-type="fig" rid="F9">9</xref> (g-i).</p><fig id="F10" position="float"><label>Figure 10</label><caption><p><bold>The segmentation of different </bold><bold> <italic>η </italic> </bold><bold>using the proposed method, from left to right in each row, (a)liver:</bold><bold> <italic>η</italic> </bold><bold>= 0.8, 0.7, 0.65, 0.6; (b) right kidney: </bold><bold> <italic>η</italic> </bold><bold>= 0.8, 0.75, 0.72, 0.7; (c) left kidney: </bold><bold> <italic>η</italic> </bold><bold>= 0.9, 0. 8, 0.75, 0.7.</bold></p></caption><graphic xlink:href="1475-925X-12-124-10"/></fig><p>As mentioned above, by adjusting the value of parameter <italic>η</italic>, the segmentation can be optimized. If the incorrect segmentation proportion of KGC inside initial dilation contour without shape priors is small, the weight of KGC data term is high, so that the value of <italic>η</italic> become bigger. On the contrary, if the incorrect segmentation proportion of KGC inside initial dilation contour without shape priors is high, the value of <italic>η</italic> should become smaller. In this way, a satisfying segmentation result can be obtained finally.</p></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>Because of the noise, weak boundary, intensity inhomogeneity and similar intensities among different abdominal organs, the segmentation of abdominal organs is normally considered a challenging task. Furthermore, the deformation caused by individual difference and respiratory movement makes the segmentation task even more difficult.</p><p>KGC is a fully automatic segmentation algorithm based on graph cuts. If the KGC algorithm is used alone to segment abdomen MR image, the segmentation result is not satisfying as shown in the Figure <xref ref-type="fig" rid="F11">11</xref> (a). The blood vessels in organ and the overlaps among abdominal organs affect the segmentation, thus, a fully complete boundary of liver or other organs cannot be obtained. To make the segmentation procedure focus on a given organ, the initial contour is produced by region growing algorithm and morphology operation. But this is considered not enough. The segmentation result can be seen from Figure <xref ref-type="fig" rid="F11">11</xref>(b). For this reason some shape priors has been used to make the segmentation algorithm more robust and accurate. Since KPCA can handle nonlinear deformable information, the shape priors based on KPCA is integrated into KGC. Moreover, if a shelter is placed onto the target’s boundary, a satisfying segmentation result can be also obtained. This is validated by experiments as shown in Figure <xref ref-type="fig" rid="F11">11</xref> (c).</p><fig id="F11" position="float"><label>Figure 11</label><caption><p>Comparison of different constraint with KGC on liver, (a) KGC’s segmentation for liver without initial contour, (b) KGC in the initial dilation contour without shape priors, (c) the proposed method’s segmentation with shelter.</p></caption><graphic xlink:href="1475-925X-12-124-11"/></fig><p>Additionally, to develop reliable prior knowledge, we should choose patients with similar age and weight. In this way their shape prior would guide to the correct segmentation.</p><p>On the clinical side, the work assumed that the geometric inaccuracy, such as distortion, does not exist. Because of this, the MR imaging technology has experienced a rapid development to overcome the magnetic field inhomogeneity. Nowadays, the image spatial resolution can be very high and geometric distortion of MR images can be ignored. However, if a more accurate segmentation is needed, geometric distortion should be corrected at first. This will be investigated in future works.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>In this paper, a novel method is proposed to segment abdominal organs integrating kernel graph cuts with KPCA shape priors after a series of pre-processing on the abdomen MR images. The morphology operation can eliminate the isolated small region after region growing algorithm. The kernel graph cuts is a fully automatic segmentation algorithm, and it also has global minimization and polynomial time complexity characteristics. The shape priors, generated by pre-image projection via KPCA, can handle nonlinear deformation. Experiments on liver and kidney segmentation of abdomen MR image showed that the novel method can obtain satisfying results. The kernel function used in this method is gauss function, and other kernel functions have not been tested yet. Currently, the seed point is obtained manually, but automation will be considered in future works. For the time performance, the algorithm can be parallelized on Graphic processors to achieve higher performance.</p></sec><sec><title>Abbreviations</title><p>MR: Magnetic resonance; KGC: Kernel graph cuts; KPCA: Kernel principle component analysis; PRI: Probabilistic rand index; VoI: Variation of information; DRLSE: Distance regularized level set evolution; AAM: Active appearance model; PCA: Principle component analysis; GCBAC: Graph cuts based active contours.</p></sec><sec><title>Competing interests</title><p>The authors declare they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>QL suggested the algorithm for images analyzing and processing, implemented it and analyzed the images. WJ gave the suggestion on algorithm analysis, experiment discussion and manuscript modification. TW, NG, SF and LL performed the acquisition of the abdominal MR images and manuscript discussion. JG and YQ expressed opinions on the evaluation metric of the segmentation results. All authors have read and approved the final manuscript.</p></sec>
Leveraging Online Learning Resources to Teach Core Research Skills to Undergraduates at a Diverse Research University	<p>Today’s students have unique learning needs and lack knowledge of core research skills. In this program report, we describe an online approach that we developed to teach core research skills to freshman and sophomore undergraduates. Specifically, we used two undergraduate kinesiology (KIN) courses designed to target students throughout campus (KIN1304: Public Health Issues in Physical Activity and Obesity) and specifically kinesiology majors (KIN1252: Foundations of Kinesiology). Our program was developed and validated at the 2<sup>nd</sup> largest ethnically diverse research university in the United States, thus we believe that it would be effective in a variety of student populations.</p>	<contrib contrib-type="author"><name><surname>McFARLIN</surname><given-names>BRIAN K.</given-names></name></contrib><contrib contrib-type="author"><name><surname>BRESLIN</surname><given-names>WHITNEY L.</given-names></name></contrib><contrib contrib-type="author"><name><surname>CARPENTER</surname><given-names>KATIE C.</given-names></name></contrib><contrib contrib-type="author"><name><surname>STROHACKER</surname><given-names>KELLEY</given-names></name></contrib><contrib contrib-type="author"><name><surname>WEINTRAUB</surname><given-names>RANDI J.</given-names></name></contrib><aff id="af1-ijes_03_02_49">Department of Health and Human Performance, University of Houston, Houston, Texas, USA</aff>	International Journal of Exercise Science	<sec sec-type="intro"><title>INTRODUCTION</title><p>It is well documented that today’s college students lack a fundamental knowledge of the research process and the skills necessary to complete effective and impactful research (<xref rid="b5-ijes_03_02_49" ref-type="bibr">5</xref>, <xref rid="b9-ijes_03_02_49" ref-type="bibr">9</xref>). Such skills are of critical importance to undergraduates who are interested in pursuing a graduate degree or medical professional degree (i.e. M.D., D.V.M., etc.). The concept of teaching undergraduates research skills has recently received renewed emphasis, as there is a critical shortage of students pursuing degrees in Science, Technology, Engineering, and Mathematics (STEM). It has been postulated that involving undergraduates in research from the beginning of their degree may represent one way to engage interest in STEM related degrees.</p><p>Despite the obvious need to expand undergraduate student exposure to research concepts, this is not always easy to accomplish due to the unique learning needs of today’s student. Individuals born between 1980 and 2000 are generally referred to as Generation Y or the Net generation and have very different learning needs than Generation X students (<xref rid="b2-ijes_03_02_49" ref-type="bibr">2</xref>–<xref rid="b4-ijes_03_02_49" ref-type="bibr">4</xref>). For example students in the Net Generation are used to having instant access to information that they can customize to their expectations. As such the traditional linear, lecture-based approach to instruction, often fails to meet learning needs. One alterative to traditional classroom instruction is the use of online instruction. When properly designed and implemented, we have demonstrated online learning modules are more effective than a traditional instructional approach at long-term maintenance of course content (<xref rid="b6-ijes_03_02_49" ref-type="bibr">6</xref>, <xref rid="b7-ijes_03_02_49" ref-type="bibr">7</xref>). From these efforts, we have established a best-practice approach to the design and use of online instruction.</p><p>In this case report we demonstrate our approach to using online instruction to teach freshman and sophomore undergraduates about the key concepts of the research process. Students also had an opportunity to test their knowledge by completing a simulated research experiment. Our learning activities were developed to address the University of Houston’s Quality Enhancement Plan (QEP) aimed at promoting research experiences among undergraduates on the UH campus. The ultimate goal of the QEP plan is to graduate students who are better prepared to pursue careers in STEM-related fields.</p></sec><sec><title>COURSE DESIGN TEAM</title><p>The learning activities described in this report were developed by a team lead by Dr. McFarlin. Dr. McFarlin has extensive experience in the design of engaging, effective online learning experiences (<xref rid="b6-ijes_03_02_49" ref-type="bibr">6</xref>, <xref rid="b7-ijes_03_02_49" ref-type="bibr">7</xref>). His efforts were supported by HHP’s Program Director (Weintraub) that assisted in the development and evaluation of learning module assessments. Breslin, Carpenter, and Strohacker are graduate students working under Dr. McFarlin’s direction and assisted with the development and pilot testing of the online activities described in this report. We believe that another advantage of our approach is that the graduate students working on this project will eventually be able to use their newly developed instructional skills after they graduate and become professors themselves. Thus the reach of our approach is well beyond the UH campus.</p></sec><sec><title>COURSES TARGETED</title><p>The learning units described in this report were incorporated into two kinesiology (KIN) courses: KIN1304 (Public Health Issues in Physical Activity and Obesity) and KIN1252 (Foundations of Kinesiology). KIN1304 is a core social science course, which is available to students from across the campus and has an average annual enrollment >1,500 students. KIN1252 targets individuals who are specifically kinesiology majors and has an average annual enrollment >750 students. We selected these courses because they were primarily taken by freshman and sophomore students, had a large annual enrollment, and were already offered in a fully online format.</p></sec><sec><title>LEARNING MANAGEMENT SYSTEM</title><p>The research learning activities were administered using Blackboard Vista as a learning management system (LMS). Blackboard is the most common LMS used on college campuses nationwide, making it the ideal platform for demonstrating our approach. The LMS learning unit for this project included (<xref ref-type="fig" rid="f1-ijes_03_02_49">Figure 1</xref>): 1) narrated, automated slide show, 2) downloadable text presentation, and 3) downloadable audio presentation. Student learning was demonstrated using self-test questions (<xref ref-type="fig" rid="f2-ijes_03_02_49">Figure 2</xref>) that were embedded in the automated slide show and a formal, graded quiz that students completed at the end of the module.</p></sec><sec><title>ONLINE LEARNING UNIT: BASIC RESEARCH SKILLS</title><p>The first learning unit we developed was designed to educate students about basic concepts associated with the research process and included the following elements: 1) developing a background, 2) research aim/hypothesis formation, 3) developing a research plan, 4) collecting data, 5) interpreting results, 6) writing a discussion, and 7) presenting/publishing the analysis. These concepts were demonstrated via a narrated, automated slide show that incorporated a series of self-test questions every 5–6 slides. Presentations were initially developed in PowerPoint 2007 (<xref ref-type="fig" rid="f3-ijes_03_02_49">Figure 3A</xref>), narrated using Articulate Studio (<xref ref-type="fig" rid="f3-ijes_03_02_49">Figure 3B</xref>), and published to a flash-media format (<xref ref-type="fig" rid="f3-ijes_03_02_49">Figure 3C</xref>). The flash-media format was ideal because it produces a compact file that can easily be uploaded into Blackboard Vista via the SCORM-compliant feature. A formal, graded quiz (administered in Blackboard) was used as one outcome measure to track student learning of basic research concepts.</p></sec><sec><title>ONLINE LEARNING UNIT: SIMULATED RESEARCH PROJECT</title><p>Once students scored at least 70% correct on the quiz over Basic Research Skills, access was granted to a simulated research experience. We selected 70% as a benchmark because it meant that the students understood the majority of the Basic Research Skills material. The simulated research experience was also developed using PowerPoint and Articulate Presenter; however, rather than a linear approach, students were allowed to progress through the simulation by answering a series of interactive questions. This approach is often referred to as decision-based learning (<xref rid="b1-ijes_03_02_49" ref-type="bibr">1</xref>, <xref rid="b8-ijes_03_02_49" ref-type="bibr">8</xref>). Slide linking was accomplished using the hyperlink feature in PowerPoint 2007. <xref ref-type="fig" rid="f4-ijes_03_02_49">Figure 4</xref> demonstrates the key elements of a question and answer sequence for our simulated research experience. It is important to note that while only 15 slides are shown in <xref ref-type="fig" rid="f4-ijes_03_02_49">Figure 4</xref>, it actually took the creation of 35 slides to ensure proper function of the simulation. Based on student feedback it is clear that our students view the simulated research experience as a valuable opportunity to apply the research skills that they have learned in a low-pressure environment.</p></sec><sec><title>EVALUATION</title><p>We used pre- and post-course evaluations to assess the effectiveness of our approach. We also collected monthly, anonymous feedback from the enrolled students to ensure that our approach is meeting their individual learning needs. Interpretation of this qualitative data indicated that students felt that our approach increased their knowledge of the research process and the ability to apply their knowledge in a practical simulation. In the future, we will complete a more comprehensive analysis of our approach.</p></sec><sec><title>FUTURE DIRECTIONS</title><p>Now that we have validated that our online learning modules are an effective way to teach undergraduates about the research process, the next logical step is to determine if our approach would be as effective at other undergraduate institutions. We are in the process of developing collaborations with other universities to implement and test our learning units. It is our ultimate goal to establish a best practice model for using online instruction to teach undergraduates about the research process.</p></sec>
Microglial intracellular Ca<sup>2+</sup> signaling as a target of antipsychotic actions for the treatment of schizophrenia	<p>Microglia are resident innate immune cells which release many factors including proinflammatory cytokines, nitric oxide (NO) and neurotrophic factors when they are activated in response to immunological stimuli. Recent reports show that pathophysiology of schizophrenia is related to the inflammatory responses mediated by microglia. Intracellular Ca<sup>2+</sup> signaling, which is mainly controlled by the endoplasmic reticulum (ER), is important for microglial functions such as release of NO and cytokines, migration, ramification and deramification. In addition, alteration of intracellular Ca<sup>2+</sup> signaling underlies the pathophysiology of schizophrenia, while it remains unclear how typical or atypical antipsychotics affect intracellular Ca<sup>2+</sup> mobilization in microglial cells. This mini-review article summarizes recent findings on cellular mechanisms underlying the characteristic differences in the actions of antipsychotics on microglial intracellular Ca<sup>2+</sup> signaling and reinforces the importance of the ER of microglial cells as a target of antipsychotics for the treatment of schizophrenia.</p>	<contrib contrib-type="author"><name><surname>Mizoguchi</surname><given-names>Yoshito</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="author-notes" rid="fn001"><sup>*</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/138661"/></contrib><contrib contrib-type="author"><name><surname>Kato</surname><given-names>Takahiro A.</given-names></name><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><xref ref-type="aff" rid="aff3"><sup>3</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/66748"/></contrib><contrib contrib-type="author"><name><surname>Horikawa</surname><given-names>Hideki</given-names></name><xref ref-type="aff" rid="aff2"><sup>2</sup></xref></contrib><contrib contrib-type="author"><name><surname>Monji</surname><given-names>Akira</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref></contrib>	Frontiers in Cellular Neuroscience	<sec sec-type="introduction" id="s1"><title>Introduction</title><p>Microglia are immune cells which are derived from progenitors that have migrated from the periphery and are from mesodermal/mesenchymal origin (Kettenmann et al., <xref rid="B16" ref-type="bibr">2011</xref>). After invading the brain parenchyma, microglia transform into the “resting” ramified phenotype and are distributed in the whole brain. However, microglia revert to an ameboid appearance when they are activated in the disturbances including infection, trauma, ischemia, neurodegenerative diseases or any loss of brain homeostasis (Aguzzi et al., <xref rid="B1" ref-type="bibr">2013</xref>; Cunningham, <xref rid="B7" ref-type="bibr">2013</xref>). Recent <italic>in vivo</italic> imaging has shown that microglial cells actively scan their environment with motile protrusions even in their resting state and are ready to transform to “activated” state in responses to injury, ischemia or autoimmune challenges in the brain (Wake et al., <xref rid="B46" ref-type="bibr">2013</xref>). Microglia can release many factors including proinflammatory cytokines (such as TNFα, IL-6), nitric oxide (NO) and neurotrophic factors (such as BDNF) when they are activated in response to immunological stimuli (Kettenmann et al., <xref rid="B16" ref-type="bibr">2011</xref>; Smith and Dragunow, <xref rid="B39" ref-type="bibr">2014</xref>). In addition, microglia are shown to be involved in the development of neural circuits or synaptic plasticity thereby maintaining the brain homeostasis (Schwartz et al., <xref rid="B37" ref-type="bibr">2013</xref>).</p><p>There is increasing evidence suggesting that pathophysiology of schizophrenia is related to the inflammatory responses mediated by microglia (Müller and Schwarz, <xref rid="B27" ref-type="bibr">2007</xref>; Kato et al., <xref rid="B14" ref-type="bibr">2011</xref>; Monji et al., <xref rid="B26" ref-type="bibr">2013</xref>; Myint and Kim, <xref rid="B28" ref-type="bibr">2014</xref>). A recent meta-analysis of associations between schizophrenia and dysfunction of immune systems including aberrant circulating cytokine levels showed that IL-1β, IL-6 and transforming growth factor-β (TGF-β) appeared to be state markers, as they were elevated in acutely relapsed inpatients or in first-episode psychosis and then normalized with antipsychotic medications. In contrast, IL-12, interferon-γ (IFNγ) and tumor necrosis factor α (TNFα) appeared to be trait markers, as they remained elevated in acute exacerbations of psychotic symptoms and even after the antipsychotic treatment (Miller et al., <xref rid="B21" ref-type="bibr">2011</xref>). Microglial activation can be estimated by positron emission tomography (PET) using radiopharmaceuticals. For example, a quantitative (R)-[(11)C]PK11195 PET scan showed that activated microglia were present in the gray matter of patients suffered from schizophrenia within the first 5 years of disease onset (van Berckel et al., <xref rid="B45" ref-type="bibr">2008</xref>). Another PET study using [11C]DAA1106 showed a positive correlation between cortical [11C]DAA1106 binding and positive symptom scores obtained from patients with schizophrenia (Takano et al., <xref rid="B42" ref-type="bibr">2010</xref>). In addition, we and others have reported that pretreatment with antipsychotics significantly inhibits the release of proinflammatory cytokines and/or NO from activated microglial cells (Hou et al., <xref rid="B10" ref-type="bibr">2006</xref>; Kato et al., <xref rid="B15" ref-type="bibr">2013</xref>). Interestingly, pretreatment with haloperidol or risperidone significantly suppressed the release of proinflammatory cytokines and NO from activated microglial cells, although the inhibitory effects of risperidone were much stronger than those of haloperidol (Kato et al., <xref rid="B13" ref-type="bibr">2007</xref>). In addition, we have previously shown that pretreatment with aripiprazole suppressed the elevation of intracellular Ca<sup>2+</sup> concentration ([Ca<sup>2+</sup>]i) induced by IFNγ in microglial cells, suggesting the importance of microglial intracellular Ca<sup>2+</sup> signaling as a target of antipsychotics for the treatment of schizophrenia (Kato et al., <xref rid="B12" ref-type="bibr">2008</xref>; Mizoguchi et al., <xref rid="B23" ref-type="bibr">2011</xref>), because elevation of intracellular Ca<sup>2+</sup> is important in activation of microglial cell functions, including proliferation, release of NO and cytokines, migration, ramification and deramification (Färber and Kettenmann, <xref rid="B8" ref-type="bibr">2006</xref>). Here, we briefly review our current understanding of the cellular mechanisms underlying the characteristic differences in the actions of antipsychotics on neuronal or microglial intracellular Ca<sup>2+</sup> signaling and reinforces the importance of the endoplasmic reticulum (ER) of microglial cells as a target of antipsychotics for the treatment of schizophrenia.</p></sec><sec id="s2"><title>Schizophrenia and intracellular Ca<sup>2+</sup> signaling</title><p>The electrical activity of neurons (i.e., excitable cells) depends on a number of different types of voltage- or ligand-gated ion channels that are permeable to inorganic ions such as sodium, potassium, chloride and calcium. While the former three ions predominantly support the electrogenic role, Ca<sup>2+</sup> are different in that they can not only alter the membrane potential but also serve as important intracellular signaling entities by themselves. In the CNS, intracellular Ca<sup>2+</sup> signaling regulates many different neuronal functions, such as cell proliferation, gene transcription and exocytosis at synapses (Berridge et al., <xref rid="B5" ref-type="bibr">2003</xref>). In neurons, because the prolonged elevation of [Ca<sup>2+</sup>]i is cytotoxic, [Ca<sup>2+</sup>]i is tightly regulated by intrinsic gating processes mediated by voltage-gated calcium channels and NMDA receptors (NMDARs; Simms and Zamponi, <xref rid="B38" ref-type="bibr">2014</xref>). In addition, dysregulation of neuronal Ca<sup>2+</sup> signaling have been linked to various neuropsychiatric disorders including schizophrenia (Lidow, <xref rid="B19" ref-type="bibr">2003</xref>). A possible involvement of intracellular Ca<sup>2+</sup> signaling in schizophrenia was originally presented by Jimerson et al. (<xref rid="B11" ref-type="bibr">1979</xref>), based on their finding that remission from acute psychotic symptoms of schizophrenia was accompanied by elevation of the Ca<sup>2+</sup> concentration in the cerebrospinal fluid. Thereafter, the interaction of neuronal dopaminergic transmission and intracellular Ca<sup>2+</sup> signaling was documented. Dopamine D2 receptors were shown to be regulated by intracellular Ca<sup>2+</sup> through the activation of CaMKII or neuronal Ca<sup>2+</sup> sensor 1 (NCS-1). Both CaMKII and NCS-1 have also been reported to be involved in the pathophysiology of schizophrenia (Bai et al., <xref rid="B3" ref-type="bibr">2004</xref>; Luo et al., <xref rid="B20" ref-type="bibr">2014</xref>). Another topic of hypothesis underlying the pathophysiology of schizophrenia is the involvement of intracellular Ca<sup>2+</sup>signaling within the fast spiking GABAergic inhibitory neurons in the hypofunction of NMDARs which leads to the dysfunction of GABAergic inhibitory circuits (Lewis et al., <xref rid="B18" ref-type="bibr">2005</xref>; Berridge, <xref rid="B4" ref-type="bibr">2013</xref>). The sustained and synchronous firing of dorsolateral prefrontal cortical neurons in the gamma frequency range of approximately 40 Hz (gamma rhythms) depends on excitatory pyramidal neurons which release glutamate to activate the inhibitory GABAergic interneurons. The hypofunction of NMDARs results in the reduction of intracellular Ca<sup>2+</sup> signaling, suppression of the induction of transcription factor CREB and reduction in the expression of the glutamic acid decarboxylase 67 (GAD67), which leads to the change of gamma rhythms and the impairment of cognitive functions observed in patients suffered from schizophrenia. In addition, dysregulation of the redox signaling pathway might provide an explanation for the developmental origins of schizophrenia because there appears to be a link between maternal viral infections during gestation and the incidence of schizophrenia. During viral infections, the increase of the IL-6 release and the resultant activation of redox signaling pathway promote the hypofunction of NMDARs in the GABAergic interneurons (Berridge, <xref rid="B4" ref-type="bibr">2013</xref>).</p><p>Recently, there are many reports that have shown that possible involvement of single-nucleotide polymorphisms (SNPs) within two L-type voltage-gated calcium channel subunits, CACNA1C and CACNB2, and neuropsychiatric disorders including schizophrenia, suggesting that dysfunction of L-type voltage-gated calcium channels occurs in patients with schizophrenia (Ripke et al., <xref rid="B35" ref-type="bibr">2013</xref>; Smoller et al., <xref rid="B40" ref-type="bibr">2013</xref>). However, the activation of voltage-gated calcium channels are well known to be suppressed by the treatment of various antipsychotics (Santi et al., <xref rid="B36" ref-type="bibr">2002</xref>; Choi and Rhim, <xref rid="B6" ref-type="bibr">2010</xref>). For example, in cultured HEK cells, haloperidol acutely blocks T-type voltage-gated calcium channels in a dose-dependent manner (Santi et al., <xref rid="B36" ref-type="bibr">2002</xref>), while it remains unclear whether antipsychotics also affect voltage-gated calcium channels in neurons. Solís-Chagoyán et al. (<xref rid="B41" ref-type="bibr">2013</xref>) recently reported that Ca<sup>2+</sup> currents mediated by L-type voltage-gated calcium channels recorded in olfactory neuroepithelial cells obtained from patients with schizophrenia were 50% smaller than those from healthy subjects. Because these patients with schizophrenia were taking antipsychotics, the finding does not simply support the genetic studies suggesting that dysfunction of L-type voltage-gated calcium channels occurs in patients with schizophrenia.</p></sec><sec id="s3"><title>Antipsychotics and the ER-mediated microglial intracellular Ca<sup>2+</sup> mobilization</title><p>Elevation of intracellular Ca<sup>2+</sup> is also important for the activation of microglia, including proliferation, migration, ramification, deramification and release of NO, proinflammatory cytokines and BDNF (Kettenmann et al., <xref rid="B16" ref-type="bibr">2011</xref>). However, in microglial cells, an application of high [K<sup>+</sup>]out or glutamate does not elevate [Ca<sup>2+</sup>]i. This observation is supported by the fact that both voltage-gated Ca<sup>2+</sup> channels and NMDARs are not expressed in microglia (Kettenmann et al., <xref rid="B16" ref-type="bibr">2011</xref>). For electrically non-excitable cells including microglia, the primary source of intracellular Ca<sup>2+</sup> is the release from intracellular Ca<sup>2+</sup> stores and the entry through the ligand-gated and/or store operated Ca<sup>2+</sup> channels (Möller, <xref rid="B25" ref-type="bibr">2002</xref>). Microglia contain at least two types of intracellular Ca<sup>2+</sup> stores: the ER and mitochondria. The main route for the generation of intracellular Ca<sup>2+</sup> signaling is associated with inositol 1,4,5-trisphosphate (InsP3) receptors on the ER membrane. Stimulation of G protein-coupled metabotropic receptors results in the activation of the phospholipase C (PLC), production of two second messengers including the diacylglycerol (DAG) and the InsP3 and the release of Ca<sup>2+</sup> from the ER. Importantly, the depletion of ER activates the store-operated Ca<sup>2+</sup> entry (SOCE), known as a capacitative Ca<sup>2+</sup> influx, mediated by plasmalemmal channels such as calcium release-activated Ca<sup>2+</sup> (CRAC) channels and/or transient receptor potential (TRP) channels (Parekh and Putney, <xref rid="B32" ref-type="bibr">2005</xref>). In addition, STIM1, one of ER membrane proteins, senses the filling state of ER Ca<sup>2+</sup> and delivers the ER to the plasma membrane where it directly activates Orai1/CRAC channels, thereby facilitating the re-uptake of Ca<sup>2+</sup> to ER through the sarco(endo)plasmic reticulum Ca<sup>2+</sup>-ATPases (SERCA). The concentration of Ca<sup>2+</sup> in the ER is precisely controlled by SERCA. The influx of Ca<sup>2+</sup> through the TRP channels plays an important role in many inflammatory processes including the activation of microglia (Nilius et al., <xref rid="B30" ref-type="bibr">2007</xref>; Mizoguchi et al., <xref rid="B22" ref-type="bibr">2014</xref>). Because there is increasing evidence suggesting that pathophysiology of schizophrenia is related to the inflammatory responses mediated by microglia (Müller and Schwarz, <xref rid="B27" ref-type="bibr">2007</xref>; Monji et al., <xref rid="B26" ref-type="bibr">2013</xref>), it could be important to examine the effects of antipsychotics on the ER function of microglial cells for the treatment of schizophrenia.</p><p>In some electrically non-excitable cells such as macrophages, adipocytes, β-cells and oligodendrocytes, perturbation of the calcium homeostasis in the ER results in the accumulation of unfolded proteins, the induction of the ER stress response, the promotion of the inflammatory processes and the initiation of apoptosis (Zhang and Kaufman, <xref rid="B47" ref-type="bibr">2008</xref>). Experimentally, the ER stress response is frequently induced by selectively inhibiting SERCA using agents such as thapsigargin (TG) which passively deplete the ER (Thastrup et al., <xref rid="B44" ref-type="bibr">1990</xref>). It remains unclear how typical or atypical antipsychotics affect the ER-mediated intracellular Ca<sup>2+</sup> mobilization in microglia. Thus, we examined how pretreatment with typical (haloperidol) or atypical (risperidone) antipsychotics affects TG-induced intracellular Ca<sup>2+</sup> mobilization, which represents a cellular stress response. In rodent microglial cells, we observed that opposite effects of haloperidol and risperidone on the TG-induced intracellular Ca<sup>2+</sup> mobilization (Mizoguchi et al., unpublished observations). There are two other reports showing opposite effects of haloperidol and risperidone on intracellular Ca<sup>2+</sup> mobilization. In cultured astrocytes derived from rat cortex and striatum, intracellular Ca<sup>2+</sup> imaging showed that pretreatment with risperidone but not haloperidol suppressed the dopamine-induced increase in [Ca<sup>2+</sup>]i (Reuss and Unsicker, <xref rid="B34" ref-type="bibr">2001</xref>). In another study obtained from rat PC12 cells, pretreatment with haloperidol potentiated the rotenone-induced neurotoxicity, while risperidone suppressed it. Likewise, pretreatment with haloperidol potentiated the rotenone-induced increase in [Ca<sup>2+</sup>]i, while risperidone completely suppressed it, suggesting that opposite effects of haloperidol and risperidone on rotenone-induced neurotoxicity could be mediated by their differential effects on intracellular Ca<sup>2+</sup> mobilization (Tan et al., <xref rid="B43" ref-type="bibr">2007</xref>). In addition, Kurosawa et al. (<xref rid="B17" ref-type="bibr">2007</xref>) reported that pretreatment with risperidone but not with haloperidol suppressed the death of rat cultured cortical neurons induced by treatment with TG for 72 h. Disruption of intracellular Ca<sup>2+</sup> signaling triggers the activation of cell death programs (Orrenius et al., <xref rid="B31" ref-type="bibr">2003</xref>). Treatment of primary cultured microglial cells by TG or ionomycin induced cellular apoptosis and this pathway was suppressed by the pretreatment with BAPTA-AM (Nagano et al., <xref rid="B29" ref-type="bibr">2006</xref>). Thus, these suggest that typical and atypical antipsychotics have different effects on the ER-mediated intracellular Ca<sup>2+</sup> mobilization, which might lead to the differences in the actions of typical and atypical antipsychotics on the induction of the ER stress response, promotion of the inflammatory responses and/or initiation of apoptosis in microglia (Figure <xref ref-type="fig" rid="F1">1</xref>).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Schematic illustration representing the microglial intracellular Ca<sup>2+</sup> signaling, especially the ER function, as major targets of antipsychotics for the treatment of schizophrenia</bold>. SERCA, sarco(endo)plasmic reticulum Ca<sup>2+</sup>-ATPases; TG, thapsigargin; SOC, store-operated calcium channel.</p></caption><graphic xlink:href="fncel-08-00370-g0001"/></fig><p>Brain-derived neurotrophic factor is also well known for its involvement in the pathophysiology of neuropsychiatric disorders including schizophrenia (Autry and Monteggia, <xref rid="B2" ref-type="bibr">2012</xref>). A recent meta-analysis of studies showed that blood levels of BDNF are reduced in both medicated and drug-naïve patients with schizophrenia (Green et al., <xref rid="B9" ref-type="bibr">2011</xref>). In addition, expression of BDNF in rodent microglia is important for the spine elimination/formation and motor-learning processes (Parkhurst et al., <xref rid="B33" ref-type="bibr">2013</xref>). We have recently reported that BDNF induces sustained [Ca<sup>2+</sup>]i elevation, which was mediated by an initial PLC/InsP3-driven Ca<sup>2+</sup> release from the ER that followed by a long-lasting activation of the SOCE via the up-regulation of cell-surface TRPC3 channels in rodent microglial cells (Mizoguchi et al., <xref rid="B24" ref-type="bibr">2009</xref>, <xref rid="B22" ref-type="bibr">2014</xref>). In addition, incubation with BDNF decreased release of NO from the activated microglia, suggesting that BDNF might have an anti-inflammatory effect through the inhibition of microglial activation and could be useful for the treatment of neuropsychiatric disorders including schizophrenia. It remains unclear how typical or atypical antipsychotics affect the BDNF-mediated intracellular Ca<sup>2+</sup> mobilization in microglia.</p><p>There is increasing evidence suggesting that pathophysiology of schizophrenia is related to the inflammatory responses mediated by microglia (Müller and Schwarz, <xref rid="B27" ref-type="bibr">2007</xref>; Kato et al., <xref rid="B14" ref-type="bibr">2011</xref>; Monji et al., <xref rid="B26" ref-type="bibr">2013</xref>; Myint and Kim, <xref rid="B28" ref-type="bibr">2014</xref>). In addition, we have reported that pretreatment with antipsychotics significantly inhibits the release of proinflammatory cytokines and/or NO from activated microglial cells, possibly through the suppression of [Ca<sup>2+</sup>]i elevation in microglial cells (Kato et al., <xref rid="B12" ref-type="bibr">2008</xref>, <xref rid="B15" ref-type="bibr">2013</xref>; Mizoguchi et al., <xref rid="B23" ref-type="bibr">2011</xref>). For electrically non-excitable cells such as microglia, the primary source of intracellular Ca<sup>2+</sup> is the ER. suggesting the importance of the ER as a therapeutic target of antipsychotics for the treatment of schizophrenia.</p></sec><sec sec-type="conclusion" id="s4"><title>Conclusion</title><p>Microglia can release many factors including proinflammatory cytokines, NO and BDNF when they are activated in response to immunological stimuli. There is increasing evidence suggesting that pathophysiology of schizophrenia is related to the inflammatory responses mediated by microglia. In addition, we have previously reported that pretreatment with antipsychotics significantly inhibits the release of proinflammatory cytokines and/or NO from activated microglial cells, possibly through the suppression of the elevation of [Ca<sup>2+</sup>]i, suggesting the importance of microglial intracellular Ca<sup>2+</sup> signaling as a target of antipsychotics for the treatment of schizophrenia. Although the electrical activity of neurons mainly depends on voltage-gated calcium channels and NMDARs, the generation of intracellular Ca<sup>2+</sup> signaling in non-excitable cells such as microglia is mainly regulated by the ER. These suggest the importance of the ER as a therapeutic target of antipsychotics for the treatment of schizophrenia.</p></sec><sec id="s5"><title>Conflict of interest statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec>
γδ T cells as early sensors of tissue damage and mediators of secondary neurodegeneration	<p>Spontaneous or medically induced reperfusion occurs in up to 70% of patients within 24 h after cerebral ischemia. Reperfusion of ischemic brain tissue can augment the inflammatory response that causes additional injury. Recently, T cells have been shown to be an essential part of the post-ischemic tissue damage, and especially IL-17 secreting T cells have been implicated in the pathogenesis of a variety of inflammatory reactions in the brain. After stroke, it seems that the innate γδ T cells are the main IL-17 producing cells and that the γδ T cell activation constitutes an early and mainly damaging immune response in stroke. Effector mechanism of γδ T cell derived IL-17 in the ischemic brain include the induction of metalloproteinases, proinflammatory cytokines and neutrophil attracting chemokines, leading to a further amplification of the detrimental inflammatory response. In this review, we will give an overview on the concepts of γδ T cells and IL-17 in stroke pathophysiology and on their potential importance for human disease conditions.</p>	<contrib contrib-type="author"><name><surname>Gelderblom</surname><given-names>Mathias</given-names></name><xref ref-type="aff" rid="aff1"/><xref ref-type="author-notes" rid="fn001"><sup></sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/175570"/></contrib><contrib contrib-type="author"><name><surname>Arunachalam</surname><given-names>Priyadharshini</given-names></name><xref ref-type="aff" rid="aff1"/><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/185273"/></contrib><contrib contrib-type="author"><name><surname>Magnus</surname><given-names>Tim</given-names></name><xref ref-type="aff" rid="aff1"/><xref ref-type="author-notes" rid="fn001"><sup></sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/176326"/></contrib>	Frontiers in Cellular Neuroscience	<sec sec-type="intro" id="s1"><title>Introduction</title><p>Ischemic stroke is the primary reason for sustained disability and the third leading cause of death in the western world. In 85% of these patients, occlusion of an artery in the brain is the cause of stroke. Early restoration of blood flow (reperfusion) remains the treatment of choice for limiting brain injury following stroke. The reperfusion, which enhances the oxygen and glucose content in the tissue also increases an inflammatory response (Iadecola and Anrather, <xref rid="B19" ref-type="bibr">2011</xref>). The idea that inflammation causes further brain injury is supported by a large number of reports that describe a reduction in infarct size and brain edema in animal models of stroke that receive blocking antibodies against specific cell adhesion molecules that mediate leukocyte recruitment (Yilmaz and Granger, <xref rid="B49" ref-type="bibr">2008</xref>), anti-inflammatory treatment (Sharkey and Butcher, <xref rid="B38" ref-type="bibr">1994</xref>), and immune deficient animals (Yilmaz et al., <xref rid="B48" ref-type="bibr">2006</xref>; Hurn et al., <xref rid="B18" ref-type="bibr">2007</xref>; Kleinschnitz et al., <xref rid="B23" ref-type="bibr">2010</xref>; Gelderblom et al., <xref rid="B14" ref-type="bibr">2012</xref>).</p></sec><sec id="s2"><title>αδ T cells and regulatory T cells in stroke</title><p>Compared to resident microglia, infiltrating macrophages and neutrophils, lymphocytes and NK cells infiltrate the ischemic hemisphere in small numbers. Nevertheless, T cells have a great impact on stroke outcome. The initial observation by Yilmaz et al. that lymphocyte deficient rag1<sup>−/−</sup> mice are protected from stroke (Yilmaz et al., <xref rid="B48" ref-type="bibr">2006</xref>) could be extended to mice with severe combined immunodeficiency lacking T cells and B cells (Hurn et al., <xref rid="B18" ref-type="bibr">2007</xref>) and to CD4<sup>+</sup> and CD8<sup>+</sup> T cell-deficient animals (Yilmaz et al., <xref rid="B48" ref-type="bibr">2006</xref>). Direct detrimental mechanisms elicited by αβ T cell in stroke pathophysiology include CD8<sup>+</sup> T cell derived perforin mediated cytotoxicity (Liesz et al., <xref rid="B29" ref-type="bibr">2011</xref>) and IL-21 secreted by CD4<sup>+</sup> T cells (Clarkson et al., <xref rid="B9" ref-type="bibr">2014</xref>).</p><p>The classical activation of αβ T cells requires several coincident signals: (1) engagement of the antigen receptor; (2) co-stimulatory receptors; (3) cytokine receptors such IL-2 receptor; a process requiring at least 3–5 d (Jensen et al., <xref rid="B20" ref-type="bibr">2008</xref>). Multiple studies using antigen specific mucosal tolerization protocols against myelin antigens suggest the involvement of adaptive mechanism in stroke pathophysiology. Already in 1997 the group from Hallenbeck demonstrated that rodents tolerized with myelin peptides are protected from ischemic stroke (Becker et al., <xref rid="B3" ref-type="bibr">1997</xref>). Mechanistically the protective effects could be attributed to IL-10 producing T cells (Frenkel et al., <xref rid="B12" ref-type="bibr">2005</xref>) and transforming growth factor-β1 (Becker et al., <xref rid="B2" ref-type="bibr">2003</xref>).</p><p>These classical concepts of T cell activation are challenged by the observation that detrimental T cell dependent effects following cerebral ischemia can be observed already 24 h post stroke, in an antigen independent fashion (Kleinschnitz et al., <xref rid="B23" ref-type="bibr">2010</xref>). Similarly controversial is the role of regulatory T<sub>regs</sub> and B cells in stroke. Liesz and colleagues showed that endogenous T<sub>regs</sub> are protective in later stages following stroke when the lesions were small (Liesz et al., <xref rid="B28" ref-type="bibr">2009</xref>) and that their beneficial functions depend on IL-10 (Liesz et al., <xref rid="B30" ref-type="bibr">2013</xref>). However, a lot of the observed effects of T<sub>regs</sub> cannot be attributed to concepts of adaptive immunity. For example, an early direct inhibitory effect of T<sub>regs</sub> on the MMP9 production from neutrophils was a recently suggested mechanism (Li et al., <xref rid="B25" ref-type="bibr">2013</xref>). In this model, transfer of regulatory T<sub>regs</sub> conferred protective effects on the outcome already on day one after stroke even before T<sub>regs</sub> infiltrated the ischemic brain. Protective effects could be attributed to program death-1 ligand 1 (PD-L1) dependent inhibition on MMP9 production in neutrophils in the peripheral circulation which then led to a consecutive protection of the blood brain barrier (Li et al., <xref rid="B26" ref-type="bibr">2014</xref>). Further studies even challenged the overall concept of T<sub>regs</sub> as endogenous protective immune cell population in stroke (Ren et al., <xref rid="B36" ref-type="bibr">2011</xref>) and a recent study suggests that T<sub>regs</sub> have an early detrimental role, by inducing dysfunction of the cerebral microcirculation (Kleinschnitz et al., <xref rid="B22" ref-type="bibr">2013</xref>). While the data on T cell effects in particular T<sub>regs</sub> in stroke is still controversial, it is clear that most of the important immunological effects are not following classical concepts of adaptive immunity, suggesting an innate like behavior of lymphocytes. In this line, atypical T cells such as γδ T cell and NK cells are likely to participate in the early orchestration of the inflammatory reaction. For NK cells it has been shown that neuronal cell death is mediated by IFN-γ- and Perforin-dependent pathways as early as 3 h post reperfusion (Gan et al., <xref rid="B13" ref-type="bibr">2014</xref>). A lot more data exist on γδ T cell, which we will focus on in the following section.</p></sec><sec id="s3"><title>Biology of γδ T cells subpopulations</title><p>Like αβ T cells, γδ T cells develop in the thymus using the recombinase activated gene product (RAG) for the somatic rearrangement of V (variable), D (Diversity and J (joining) gene segments of the γ and δ chains of their T cell receptor (TcR) (reviewed in Raulet, <xref rid="B35" ref-type="bibr">1989</xref>). Compared to αβ TcR, the sets of TcR detected on γδ T cells are limited. Many γδ subsets, primarily the ones populating certain tissues such as the epidermis, dermis, intestine, lungs and uterus are displaying an even higher limitation of their TcR diversity. These tissue-specific γδ T cell subsets show a biased use of certain TcR V gene segments. Since some of them express “invariant” TcRs with identical (canonical) junctional sequences, they are also named canonical γδ T cells. As reviewed by Vantourout and Hayday, the limited TcR diversity implies that these cells recognize either pathogen encoded antigens, that are likely to be encountered in specific tissues such as the epidermis, or self-encoded molecules that reflect a dysregulated state of that tissue (Vantourout and Hayday, <xref rid="B47" ref-type="bibr">2013</xref>). Since these γδ T cell subsets can be rapidly activated without the requirement of prior clonal expansion they are also called “innate like” T cells. In contrast to canonical γδ T cells so-called non-canonical γδ T cells, which are characterized by an expression of more diverse γδ TcRs, are homing into secondary lymphoid tissues. Here they make up a minor fraction of rodent and human T cells after birth (in mice 1–4% of all T cells). In the context of immune responses non-canonical γδ T cells are capable to participate distant from their original site of residence, by trafficking to the site of inflammation in solid organs (reviewed by Korn and Petermann, <xref rid="B24" ref-type="bibr">2012</xref>). Similar to αβ T cells, γδ T cells can be divided by their cytokine profile. Mouse γδ T cells which are developing from fetal liver progenitors undergo functional pre-programming, which leads to a subpopulation of IL-17 producing Scart-2<sup>+</sup> and CCR6<sup>+</sup> γδ T cells on one side and IFN-γ producing NK.1.1<sup>+</sup> and CD27<sup>+</sup> γδ T cells on the other side. Both subpopulations have an innate like phenotype, since they can be rapidly activated without prior clonal expansion (Vantourout and Hayday, <xref rid="B47" ref-type="bibr">2013</xref>). γδ T cells fulfill important sentinel functions in the immune system. The ability of γδ T cells to recognize molecules that are rapidly displayed after stress without requiring extensive clonal expansion permits γδ T cells to participate in early stages of immune responses. In such scenarios γδ T cells act in parallel with cells of the innate immune system as sensors of dysregulation. γδ T cells may respond to classical signals of the adaptive immune system or to cytokine signals and either Toll-like receptor (TLR) or dectin stimuli in the absence of TcR ligation. Activation of the γδ TcR can occur through major histocompatibility complex (MHC)-related and unrelated TcR ligands, which are including foreign- and self-antigens. This allows γδ T cells to respond to infection and sterile tissue dysregulation such as ischemia. Beside TcR dependent mechanisms γδ T cell activation can be mediated through engagement of the activating natural killer receptors (NKRs) such as NK group 2 member D (NKD2D), by patter recognition receptors including TLRs (reviewed by Bonneville et al., <xref rid="B4" ref-type="bibr">2010</xref>) and through cytokines such as IL-1β and/or IL-23 (Sutton et al., <xref rid="B43" ref-type="bibr">2009</xref>). The constitutive expression of IL-23 and IL-1β receptors by γδ T cells assures this rapid responsiveness. Within hours upon activation and without prior expansion systemic γδ T cells can express high levels of effector cytokines, such as IFN-γ, IL-17, TNF-α and granzymes. In addition, γδ T cells are capable of producing numerous chemokines and regulatory factors including IL-13 and insulin-like-growth factor 1 (IGF-1), allowing them to interact with other immune cells, such as B cells and αβ T cells in the afferent phase of the immune response. Regarding the cellular interplay between γδ T cells and innate immune cells neutrophils play a central role. Once activated, γδ T cells can stimulate the release of potent chemoattractants for neutrophils. In this respect, γδ T cells were recently shown to be the primary sources of the neutrophil-attracting IL-17 in mouse models of infection (Shibata et al., <xref rid="B40" ref-type="bibr">2007</xref>), hypersensitivity (Simonian et al., <xref rid="B42" ref-type="bibr">2009</xref>) and autoimmunity (Roark et al., <xref rid="B37" ref-type="bibr">2007</xref>). Often the activation of the innate immune system results in a feed back loop that increasingly stimulates γδ T cells.</p></sec><sec id="s4"><title>γδ T cells as sensors of tissue damage in stroke</title><p>Stroke resembles classical features of a “sterile inflammation”, which is characterized by a inflammation in response to tissue disruption without the involvement of pathogenic microorganisms (See Figure <xref ref-type="fig" rid="F1">1</xref>; Chen and Nuñez, <xref rid="B6" ref-type="bibr">2010</xref>). Sterile inflammation shares similar mechanisms with inflammation during infection. Receptors essential for sensing microorganisms are collectively called pattern recognition receptors (PPRs). PRRs sense conserved structural moieties that are found in microorganisms and are often called pathogen-associated molecular patterns (PAMPs) (for review see Chen and Nuñez, <xref rid="B6" ref-type="bibr">2010</xref>). Following ligand recognition these receptors activate downstream signaling pathways, such as the nuclear factor-κb (NF-κb), mitogen-activated protein kinase (MAPK) and type I interferon pathways, which result in the upregulation of pro-inflammatory cytokines and chemokines that are important in inflammatory responses. In non-infectious conditions immune cells can be activated via recognition of endogenous material by PPRs. These endogenous molecules have been named danger-associated molecular patterns (DAMPs). Under physiological conditions these DAMPs are localized intracellularly. Under conditions of apoptotic cells death, cells are cleared immunologically silent without significant release of DAMPs into the extracellular environment. In contrast, necrosis following ischemia leads to loss of cell integrity and release of the cell content into the extracellular space. DAMPs derived from necrotic cells include the chromatin-associated protein high-mobility group box 1 (HMGB1), heat shock proteins (HSPs), mitochondrial peptides and purine metabolites, such as adenosine triphosphate (ATP) and uric acid (reviewed by Chen and Nuñez, <xref rid="B6" ref-type="bibr">2010</xref> and Shen et al., <xref rid="B39" ref-type="bibr">2013</xref>). Consecutively, activated receptors and signaling pathways include TLR2/4/9, CD24, CD44, NLRP3, formyl peptide receptor 1, RAGE and IL-1 receptor. In the context of stroke DAMPs are massively released into the extracellular compartment. In stroke several pathways have been described, including TLR2/4, CD38, P2X7 and RAGE, which are associated with an worsened outcome (Liu et al., <xref rid="B31" ref-type="bibr">2007</xref>; Tang et al., <xref rid="B45" ref-type="bibr">2008</xref>; Choe et al., <xref rid="B7" ref-type="bibr">2011</xref>; Arbeloa et al., <xref rid="B1" ref-type="bibr">2012</xref>). As we discussed above γδ T cells can be activated directly by DAMPs via TLR1/2 and dectin receptors and cytokines, such as IL-1β and IL-23 (Martin et al., <xref rid="B32" ref-type="bibr">2009</xref>; Sutton et al., <xref rid="B44" ref-type="bibr">2012</xref>). Following stroke, there is clear evidence that IL-23 activates IL-17 production in γδ T cells (Shichita et al., <xref rid="B41" ref-type="bibr">2009</xref>). Even though it is likely that further signals via TcR and TLR/dectin receptors are necessary to fully activate γδ T cells, the actual experimental data is outstanding.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Sequential events leading to neutrophil infiltration</bold>. First, release from DAMPs from injured cells activates resident microglia via PPRs to release proinflammatory factors including TNF-α. Second, IL-23 activates γδ T cells to rapidly secrete IL-17 in the ischemic tissue. Third, neutrophil infiltration is initiated via IL-17 and TNF-α synergistically induced expression of CXCL-1 in astrocytes.</p></caption><graphic xlink:href="fncel-08-00368-g0001"/></fig></sec><sec id="s5"><title>Effector mechanisms of γδ T cells in stroke</title><p>Several papers have shown a significant contribution of γδ T cells and IL-17 in stroke and other conditions of central nervous system inflammation (Kebir et al., <xref rid="B21" ref-type="bibr">2007</xref>; Shichita et al., <xref rid="B41" ref-type="bibr">2009</xref>; Caccamo et al., <xref rid="B5" ref-type="bibr">2011</xref>; Gelderblom et al., <xref rid="B14" ref-type="bibr">2012</xref>). In ischemia reperfusion injury of the brain we and others have observed a pathogenic role of γδ T cells, which can be detected in ischemic brain tissue as early as 6 h post ischemia (Shichita et al., <xref rid="B41" ref-type="bibr">2009</xref>; Gelderblom et al., <xref rid="B14" ref-type="bibr">2012</xref>). Effector mechanisms of γδ T cells in stroke primarily depend on their IL-17 production. In stroke, synergistic stimulation of astrocytes by IL-17 and TNF-α induces a massive induction of neutrophil attracting chemokines including CXCL-1 (Gelderblom et al., <xref rid="B14" ref-type="bibr">2012</xref>), resulting in a subsequent neutrophil infiltration, which is leading to an increased tissue damage. Activated macrophages and microglia are secreting high amounts of TNF-α in the ischemic tissue. In the presence of the TNF-α rich milieu the additional IL-17 signal leads to the rapid increase of CXCL-1 via a stabilizing effect on the CXCL-1 RNA in astrocytes. Blocking either signal, IL-17 or the CXCL-1/CXCR2-axis, results in a robust reduction in infarct size and a significant improved neurological outcome. Even if an anti-IL-17 antibody is administered 6 h after stroke, neutrophil invasion can be blocked (Gelderblom et al., <xref rid="B14" ref-type="bibr">2012</xref>). Neutrophil independent effects of IL-17 secreted by γδ T cells in stroke include the induction of MMP3 and MMP9 which are associated with blood brain barrier breakdown (Shichita et al., <xref rid="B41" ref-type="bibr">2009</xref>; Gelderblom et al., <xref rid="B14" ref-type="bibr">2012</xref>). Other potential effector functions of γδ T are engagement of death inducing receptors such as CD95 or TNF-related apoptosis-inducing ligand receptors (TRAILR), and the release of cytotoxic effector molecules, such as perforin and granzymes (See Figure <xref ref-type="fig" rid="F1">1</xref>). Molecular signals directing γδ T cell into the ischemic brain is another unresolved issue. γδ T cell subpopulations can be divided by Scart-2 and CCR6 vs. NK.1.1 and CD27 expression into IL-17 vs. IFN-γ producing T cells, respectively. The functional relevance of the CCR6 expression on IL-17 producing γδ T cells is supported by experimental data, showing that the migration of γδ T cells into the inflamed liver depends on the CCL20/CCR6 axis (Hammerich et al., <xref rid="B15" ref-type="bibr">2014</xref>). Nevertheless, in stroke it is so far unclear which chemokines/chemokine receptors are essential for the entry of γδ T cells into the ischemic brain and which γδ T cell subpopulations are migrating into the ischemic brain.</p></sec><sec id="s6"><title>Role of γδ T cells in human stroke pathophysiology</title><p>Most of the data on inflammation in stroke is derived from studies in rodent models. These models have several drawback, including differences between the immune system of rodents and humans. Further, the vast majority of stroke patients are older that 65 and are characterized by co-morbidities, which are not reflected in rodent models (Heuschmann et al., <xref rid="B16" ref-type="bibr">2010</xref>). Despite these discrepancies, results from post-mortem and imaging studies in human stroke demonstrate that a rapid activation of the resident and systemic immune system are hallmarks of human stroke pathophysiology (Mena et al., <xref rid="B33" ref-type="bibr">2004</xref>; Price et al., <xref rid="B34" ref-type="bibr">2004</xref>; Thiel and Heiss, <xref rid="B46" ref-type="bibr">2011</xref>). Similar to experimental stroke, neutrophils are recruited into the ischemic brain within 24 h after symptom onset (Chuaqui and Tapia, <xref rid="B8" ref-type="bibr">1993</xref>; Price et al., <xref rid="B34" ref-type="bibr">2004</xref>) and microglia undergo rapid activation in the infarct core but also remote areas such as fiber tracts or relay nuclei (Thiel and Heiss, <xref rid="B46" ref-type="bibr">2011</xref>). These findings let to several clinical trials targeting neutrophils in human stroke. Studies employing inhibitors of the neutrophil—endothelial cell interaction including CD18 and ICAM-1 were conducted, none of them showing favorable results on the clinical outcome parameters (del Zoppo, <xref rid="B10" ref-type="bibr">2010</xref>). Nevertheless, the immunological understanding of the post ischemic inflammatory response was limited when these human trials were designed. Regarding our current understanding of the stroke induced inflammation IL-17 seems to be promising target. Infiltration by γδ T cells and secretion of IL-17 have been demonstrated in ischemic pathological human brain tissue (Li et al., <xref rid="B27" ref-type="bibr">2005</xref>; Gelderblom et al., <xref rid="B14" ref-type="bibr">2012</xref>). Similarly, IL-17 induced downstream pathways can be found. The IL-17 presence in the ischemic brain is early and short-lived and has most likely only pro-inflammatory effects. Therefore a short anti-IL-17 intervention could be beneficial without producing side effects, for example enhancing the systemic immune suppression. Recent data from human clinical trials with humanized neutralizing IL-17A antibodies in patients with autoimmune disease showed that treatment is well tolerized and effective (Hueber et al., <xref rid="B17" ref-type="bibr">2010</xref>).</p></sec><sec id="s7"><title>Summary</title><p>Inflammation can enhance ischemic damage and lymphocytes seem to be important component of this process. Interestingly, the classical concepts of adaptive immune responses do not explain all observed effects. Several innate like features of lymphocytes dominate the early pro-inflammatory events. Particularly atypical T cells such as γδ T cells could explain some of these discrepancies and targeted treatment against their signature cytokine IL-17 might be a promising treatment option.</p></sec><sec id="s8"><title>Conflict of interest statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec>
Retraction: Physiological basis and image processing in functional magnetic resonance imaging: neuronal and motor activity in brain	Could not extract abstract	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Sharma</surname><given-names>Rakesh</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>rpk1@columbia.edu</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Sharma</surname><given-names>Avdhesh</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>avdhesh_2000@yahoo.com</email></contrib>	BioMedical Engineering OnLine	<sec><title>Retraction</title><p>This article [<xref ref-type="bibr" rid="B1">1</xref>] has been retracted by the publisher because of significant overlap with figures from previously published work [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B3">3</xref>] without appropriate attribution or permissions. We were unable to contact the authors of the article, despite our best efforts. We apologise to all affected parties for the inconvenience caused.</p></sec>
Subcellular structural plasticity caused by the absence of the fast Ca<sup>2+</sup> buffer calbindin D-28k in recurrent collaterals of cerebellar Purkinje neurons	<p>Purkinje cells (PC) control spike timing of neighboring PC by their recurrent axon collaterals. These synapses underlie fast cerebellar oscillations and are characterized by a strong facilitation within a time window of <20 ms during paired-pulse protocols. PC express high levels of the fast Ca<sup>2+</sup> buffer protein calbindin D-28k (CB). As expected from the absence of a fast Ca<sup>2+</sup> buffer, presynaptic action potential-evoked [Ca<sup>2+</sup>]<sub>i</sub> transients were previously shown to be bigger in PC boutons of young (second postnatal week) CB-/- mice, yet IPSC mean amplitudes remained unaltered in connected CB–/– PC. Since PC spine morphology is altered in adult CB–/– mice (longer necks, larger spine head volume), we summoned that morphological compensation/adaptation mechanisms might also be induced in CB–/– PC axon collaterals including boutons. In these mice, biocytin-filled PC reconstructions revealed that the number of axonal varicosities per PC axon collateral was augmented, mostly confined to the granule cell layer. Additionally, the volume of individual boutons was increased, evidenced from z-stacks of confocal images. EM analysis of PC–PC synapses revealed an enhancement in active zone (AZ) length by approximately 23%, paralleled by a higher number of docked vesicles per AZ in CB–/– boutons. Moreover, synaptic cleft width was larger in CB–/– (23.8 ± 0.43 nm) compared to wild type (21.17 ± 0.39 nm) synapses. We propose that the morphological changes, <italic>i.e.,</italic> the larger bouton volume, the enhanced AZ length and the higher number of docked vesicles, in combination with the increase in synaptic cleft width likely modifies the GABA release properties at this synapse in CB–/– mice. We view these changes as adaptation/homeostatic mechanisms to likely maintain characteristics of synaptic transmission in the absence of the fast Ca<sup>2+</sup> buffer CB. Our study provides further evidence on the functioning of the Ca<sup>2+</sup> homeostasome.</p>	<contrib contrib-type="author"><name><surname>Orduz</surname><given-names>David</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="author-notes" rid="fn002"><sup></sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/49862"/></contrib><contrib contrib-type="author"><name><surname>Boom</surname><given-names>Alain</given-names></name><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/190202"/></contrib><contrib contrib-type="author"><name><surname>Gall</surname><given-names>David</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/1850"/></contrib><contrib contrib-type="author"><name><surname>Brion</surname><given-names>Jean-Pierre</given-names></name><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/9557"/></contrib><contrib contrib-type="author"><name><surname>Schiffmann</surname><given-names>Serge N.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/33636"/></contrib><contrib contrib-type="author"><name><surname>Schwaller</surname><given-names>Beat</given-names></name><xref ref-type="aff" rid="aff3"><sup>3</sup></xref><xref ref-type="author-notes" rid="fn002"><sup></sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/38834"/></contrib>	Frontiers in Cellular Neuroscience	<sec><title>INTRODUCTION</title><p>Activity-dependent synaptic plasticity, <italic>i.e.,</italic> modulation of the synaptic strength of two neurons coupled by chemical synapses is the hallmark of most synapses (<xref rid="B27" ref-type="bibr">Kim and Linden, 2007</xref>). Induction of plasticity occurs at highly variable time scales. Such changes may be (<italic>i</italic>) very brief and restricted to the time period from the start to the end of trains of APs and are named STP, (<italic>ii</italic>) longer lasting as in the case of Hebbian-type plasticity including LTP or depression (LTP and LTD, respectively), or (<italic>iii</italic>) operating at time scales of hours to days and the process is referred to as HSP (<xref rid="B56" ref-type="bibr">Turrigiano, 2012</xref>). The latter is thought to maintain network stability by fine-tuning global synaptic strength under conditions when its activity diverges from tolerant (stable) physiological levels, <italic>e.g.,</italic> as the consequence of an insult, chronic suppression of activity or mutations in genes implicated in synaptic transmission. Many different processes occurring in the pre- and/or postsynaptic compartment have been described for the various types of plasticity (<xref rid="B71" ref-type="bibr">Zucker and Regehr, 2002</xref>; <xref rid="B8" ref-type="bibr">Blitz et al., 2004</xref>; <xref rid="B59" ref-type="bibr">Vitureira and Goda, 2013</xref>). Since transmitter release at presynaptic terminals is a Ca<sup>2+</sup>-dependent process, the precise shape of Ca<sup>2+</sup> signals within presynaptic terminals is a critical determinant (<xref rid="B11" ref-type="bibr">Cavazzini et al., 2005</xref>). Among other components linked to Ca<sup>2+</sup> entry and extrusion, Ca<sup>2+</sup> buffers are considered as relevant modulators of these presynaptic Ca<sup>2+</sup> signals. Examples of such buffers characterized by either slow or fast Ca<sup>2+</sup>-binding kinetics include parvalbumin (PV) and CB, respectively (<xref rid="B51" ref-type="bibr">Schwaller, 2010</xref>). Both of these proteins were previously shown to modulate STP (<xref rid="B7" ref-type="bibr">Blatow et al., 2003</xref>; <xref rid="B14" ref-type="bibr">Collin et al., 2005</xref>; <xref rid="B41" ref-type="bibr">Orduz et al., 2013</xref>). The quantitative aspects of a presynaptic Ca<sup>2+</sup> signal then determine the time course and amount of neurotransmitter released into the synaptic cleft. This, in turn, leads to an appropriate response in the postsynaptic neuron, in the form of an inhibitory or excitatory postsynaptic response, depending on the type of neurotransmitter and on the type(s) of receptors. Also at the postsynaptic side, several mechanisms leading to modulation of synaptic transmission have been described and include receptor saturation, receptor desensitization, receptor distribution/clustering and/or phosphorylation (<xref rid="B61" ref-type="bibr">von Gersdorff and Borst, 2002</xref>), but also morphological changes (<italic>e.g.,</italic> spine shape; <xref rid="B3" ref-type="bibr">Alvarez and Sabatini, 2007</xref>). Thus, while pre- and postsynaptic compartments show a large degree of plasticity, the architecture of the synaptic cleft, most notably cleft width is considered as relatively resistant to plasticity. The cleft width is assumed to be essentially determined by interactions of proteins anchored in the pre- and postsynaptic membrane including neurexin family members and neuroligins, respectively. Consequently, the cleft width for a particular synapse shows extremely little variation, most often less than 5%, <italic>i.e.,</italic> 0.4 nm variation for a “typical” synapse cleft width of approximately 20 nm, as modeled based on Monte Carlo simulations (<xref rid="B46" ref-type="bibr">Savtchenko and Rusakov, 2007</xref>) or measured from EM images. Structural changes with respect to synaptic cleft width, dimension of the AZ/PSD, curvature and presynaptic bouton volume have been reported. These changes are often the result of short- or long-term pathological insults such as oxygen/glucose deprivation (OGD; <xref rid="B34" ref-type="bibr">Lushnikova et al., 2011</xref>), long-term exposure to lead (<xref rid="B21" ref-type="bibr">Han et al., 2014</xref>), aluminum (<xref rid="B25" ref-type="bibr">Jing et al., 2004</xref>) or biphenyl-A, a presumed endocrine disruptor (<xref rid="B66" ref-type="bibr">Xu et al., 2013a</xref>,<xref rid="B67" ref-type="bibr">b</xref>). But also exposure of ovarectomized (OVX) rats to estradiol benzoate increases the synaptic cleft width of CA1 hippocampal synapses in the pyramidal cell layer (<xref rid="B68" ref-type="bibr">Xu and Zhang, 2006</xref>). Most recently, a change in the synaptic cleft architecture was reported in mice deficient for the neurexin family member contactin associated protein-like 4 (CNTNAP4), a protein implicated in autism spectrum disorders (ASD; <xref rid="B26" ref-type="bibr">Karayannis et al., 2014</xref>). Based on the observation that the absence of the fast Ca<sup>2+</sup> buffer CB in presynaptic boutons of PC recurrent axon collaterals of PN7–PN12 mice had no effect on either basic synaptic transmission (mean IPSC amplitude) or on paired-pulse facilitation (PPF; <xref rid="B10" ref-type="bibr">Bornschein et al., 2013</xref>), we summoned adaptive/homeostatic mechanisms to be present at this synapse and focused on changes in the synapse morphology. We observed an increase in presynaptic bouton volume, paralleled by an increase in the AZ/PSD length, in the number of docked vesicles and moreover, an increase in synaptic cleft width. This is in line with the proposition of <xref rid="B33" ref-type="bibr">Loebel et al. (2013)</xref> that “ongoing synaptic plasticity results in matched presynaptic and postsynaptic modifications, in which elementary modules that span the synaptic cleft are added or removed as a function of experience.”</p></sec><sec sec-type="materials\|methods" id="s1"><title>MATERIALS AND METHODS</title><sec><title>ANIMALS</title><p>The experiments and procedures conformed to the regulations of the Institutional Ethical Committee of the School of Medicine of the Université Libre de Bruxelles, Belgium and the University of Fribourg, Switzerland. In this study we used C57Bl/6J mice as control animals and null-mutant mice for (CB–/– mice; <xref rid="B2" ref-type="bibr">Airaksinen et al., 1997</xref>).</p></sec><sec><title>CB IMMUNOHISTOCHEMISTRY AND WESTERN BLOT ANALYSES</title><p>To assess developmental CB expression in the cerebellum, immunohistochemistry with CB antibodies was performed with sections from mice of the same litters (PN5–PN25). Animals were perfused intracardially with phosphate-buffered saline (PBS), followed by paraformaldehyde (4%) and rinsed with PBS (0.1 M). Cerebella were removed and placed overnight in a 4% PFA solution. Sagittal slices of the cerebellar vermis (80 μm) were prepared in ice-cold PBS solution (4°C). Slices from animals belonging to the same litter were processed simultaneously. Primary antibody incubations were performed at 4°C for 24 h, using rabbit anti-CB 38a antiserum (1:500, Swant, Marly, Switzerland) and secondary antibody incubations for 1 h at room temperature with Alexa Fluor-labeled donkey anti-rabbit 488 (1:500, Invitrogen).</p><p>For Western blot analyses, mice from similar age groups (PN6–PN25; <italic>n</italic> = 3 animals for each age group/series, 2 series for all but one for the PN25 group) were deeply anesthetized by CO<sub>2</sub> inhalation and perfused transcardially with ice-cold PBS solution (4°C). Dissected cerebellum were homogenized in 10 mM Tris-HCl/1 mM EDTA, pH 7.4. Soluble protein fractions (supernatant) were obtained by centrifugation of homogenates at 15,000 × g for 30 min. Proteins (1 μg) were separated by SDS-PAGE (12%) and transferred on nitrocellulose membranes. Membranes were incubated with a blocking solution (LI-COR Biosciences GmbH, Bad Homburg, Germany) for 1 h. Next, they were incubated overnight with primary antibodies against CB (CB 38a; dilution, 1:1000; Swant) or against α-actin (mouse monoclonal, dilution 1:1000; Sigma). As secondary antibodies we used either anti-rabbit labeled with IRDye 800CW or anti-mouse labeled with IRDye 680RD (dilution 1:10,000, incubation for 1 h). The bands corresponding to CB and α-actin were visualized and quantified by the Odyssey®; Infrared Imaging System (LI-COR) and the corresponding software, respectively.</p></sec><sec><title>BIOCYTIN-FILLING OF PC AND SYNAPTOPHYSIN LABELING</title><p>Sagittal cerebellar slices (180 μm) were prepared from WT and CB–/– mice (PN18–PN25). Animals were anesthetized with halothane before decapitation. After rapid removal of the cerebellum, slices were cut with a Leica VT1000S vibratome (Leica Microsystems), in ice-cold bicarbonate-buffered saline (BBS) at 4°C, containing (in mM): 125 NaCl, 2.5 KCl, 1.25 NaH<sub>2</sub>PO<sub>4</sub>, 26 NaHCO<sub>3</sub>, 2 CaCl<sub>2</sub>, 1 MgCl<sub>2</sub>, 10 glucose and equilibrated with a 95%O<sub>2</sub>–5%CO<sub>2</sub> mixture (pH 7.3). Before experiments, slices were incubated at 34°C for 45 min in the same saline. After this period, a single slice was transferred to a recording chamber and submerged in continuously flowing BBS at 22–24°C with a flow rate of 1.5 ml/min.</p><p>To increase the probability of finding PC with intact axons and recurrent collateral arbors, we targeted PC located (<italic>i</italic>) in the second/third somata below the surface of the slice and (<italic>ii</italic>) in flat regions between the apex and base of a lobule. These regions were clearly less damaged during the slicing procedure. PC were visualized with a 63x water immersion objective placed in a Zeiss upright microscope (Axioskop 2FS Plus, Zeiss) and recorded using the whole-cell configuration of the patch-clamp technique (WCR) with a double EPC-10 operational amplifier (Heka Elektronik). Patch pipettes were made from borosilicate glass capillaries (Hilgenberg GmbH) with a two-stage vertical puller (PIP 5, Heka Elektronik) with resistances between 3.5 and 5 MΩ. The intracellular solution contained the following (in mM): 150 K-gluconate, 4.6 MgCl<sub>2</sub>, 10 K-Hepes, 0.1 K-EGTA, 0.4 Na-GTP, 4 Na-ATP, and 0.1 CaCl<sub>2</sub> (pH 7.2) and 0.4% biocytin.</p><p>WCR was maintained for at least 20 min after break-in to improve labeling up to distal regions of recurrent collaterals. Once this time period had elapsed, high-resistance outside-out patches were obtained during pipette withdrawal. An extra time period of 15 min was allotted for biocytin diffusion, then slices were fixed overnight in 4% PFA (4°C) and rinsed with PBS (0.1 M). For double immunostainings we first incubated slices with a primary rabbit anti-synaptophysin 1 antibody (1:200, Synaptic systems) for 24 h, followed by secondary anti-rabbit 633 antibodies (1:500, Invitrogen) and streptavidin-conjugated NL557 (1:5000, R&D systems) at room temperature during 1 h.</p></sec><sec><title>CONFOCAL MICROSCOPY AND IMAGE ANALYSIS</title><p>Confocal acquisitions were obtained by using an Axiovert 200M-LSM 510 META microscope (Zeiss) equipped with a Plan-Neofluar 10x/0.3 W or a C-Apochromat 40x/1.2 W objective. We used three laser beams (a 488 nm argon and 543 and 633 nm helium–neon laser lines) with filters to selectively detect emitted fluorescence from CB-positive regions (BP 500–550 nm), biocytin-filled cells (BP 565–615 nm) and synaptophysin-positive zones (BP 650–710).</p><p>For the CB quantification during PN development by immunofluorescence acquisition, parameters were optimized for non-saturated CB signals at PN20 on 60 μm-thick z-stacks composed of 115 × 115 μm images with a z-step of 3 μm. Subsequently similar stacks for sections from younger mice from the same litter were acquired using the identical parameters. Because fluorescence can be artificially reduced on optical slices as the result of incomplete antibody penetration, we used ImageJ software (<ext-link ext-link-type="uri" xlink:href="http://imagej.nih.gov/ij/">http://imagej.nih.gov/ij/</ext-link>) to identify the five images within each stack with the strongest fluorescence intensities. A maximal intensity z-projection was made with these five selected optical slices and we calculated the mean fluorescence values of the PC layer as percentage of fluorescence normalized to PN20. For colocalization studies, the fluorescence intensity profiles were obtained from single optical sections of a PC after biocytin-filling (green) and synaptophysin labeling (red).</p><p>For the localization of PC boutons in the different PC layers, biocytin-loaded PC were reconstructed from an area of 350 × 350 μm and consisting of 80 z-sections (0.8 μm each). To visualize recurrent collateral boutons on PC somata, z-stacks containing 30 images (thickness of 0.4 μm per section) and covering a region of 14 × 14 μm were selected. We applied a median filter to each z-projection from the stacks to reduce noise and to perform morphology measurements (<italic>e.g.,</italic> bouton volume) with ImageJ tools. To assess the distribution pattern of boutons from several PC, all PC somata were superimposed and axonal arbors were oriented in such a way as to respect their position within cerebellar layers and to position the main axon path toward the white matter. We then determined x and y positions for each bouton. These coordinates were used to build a 2D-matrix in MatLab environment permitting to quantify the number of boutons within 20-μm side length squares. This size was chosen, since it corresponds approximately to the diameter of a PC soma. We then calculated the bouton density as the number of boutons per μm<sup>2</sup> and used a color gradient scale (heat map) to visualize regions with high and low bouton densities. For bouton volume measurements, we first defined the “boundaries of a bouton” along the axon recurrent collateral as the points at which collaterals dilate to twice its axonal diameter. Next we counted the number of fluorescent voxels between these points (or in the case of terminal boutons only from the starting point). Because a single voxel corresponded to a cube of 0.00004 μm<sup>3</sup> (0.01 × 0.01 × 0.4 μm for x, y, and z dimensions), we were able to calculate the approximated value of bouton volumes. Changes in volume were more evident when we performed 3D-surface reconstructions of each bouton with Osirix software, as it is shown in <bold>Figure <xref ref-type="fig" rid="F3">3D</xref></bold>.</p></sec><sec><title>ELECTRON MICROSCOPY AND ULTRASTRUCTURAL MEASUREMENTS</title><p>Mice (PN20) were anaesthetized and were perfused transcardially with a heparinized saline solution followed by fixative containing 4% PFA and glutaraldehyde (0.25%) in 0.1 M phosphate buffer (pH 7.3). Brains were removed, cerebella were dissected and further fixed for 4 h in the same fixative. For pre-embedding immunolabeling, cerebella were sagittally sectioned on a vibrating blade microtome (Ted Pella, Inc., USA) at a thickness of 35–40 μm. The sections were processed for immunolabeling with the streptavidin–biotin-peroxidase complex (ABC, Vector, UK). In brief, the tissue sections were pre-incubated for 1 h in 5% normal goat serum with 0.5% Triton X-100 and then incubated for 24 h at 4°C with TBS containing anti-L7 polyclonal rabbit antibody (1:500, Santa Cruz). After washing in TBS, sections were incubated for 4 h with the goat anti-rabbit antibody conjugated to biotin (1:100, Vector, UK) followed by the streptavidin–biotin-peroxidase complex (Vector). The peroxidase activity was revealed using diaminobenzidine as chromogen (DAB; Dako, Belgium). 35–40 μm-thick cerebellar slices were then observed with transmitted light on an inverted microscope (Zeiss, Axiovert 200 M), which corroborated L7-labeling on PC somata and PC collateral boutons from both WT and CB–/– mice. After washing in Millonig’s buffer with 0.5% (w/v) sucrose for 24 h, sections were post-fixed in 2% (w/v) OsO<sub>4</sub> for 30 min, dehydrated in a graded series of ethanol and embedded in Epon-resin LX112 (Ladd Research Industries, Inc., USA). Semi-thin sections were stained with toluidine blue. Ultrathin sections (30 nm) collected on nickel grids were stained in 2% aqueous uranyl acetate and lead citrate and observed with a Zeiss EM 809 microscope at 80 kV coupled to a digital camera (Jenoptik, Prog Res C14, Germany).</p><p>PC – PC contacts were found by careful inspection of PC somata. PC slightly labeled by L7 antibodies presented a better ultrastructure, which facilitated the extraction of morphological parameters. In contrast, strong labeling hid the most important details by obscuring the synapses. Thus, we focused our analyses on morphologically well preserved, lightly stained PC <italic>e.g.,</italic> shown in <bold>Figure <xref ref-type="fig" rid="F4">4B</xref></bold>.</p><p>Morphological measurements were made with ImageJ tools on EM images at x 30,000 magnifications. To analyze PC–PC synapses, we first identified PC soma based on their characteristic soma shape and size. The PC soma was surveyed under high magnification and PC–PC synapses were recognized by the presence of (<italic>i</italic>) L7-labeling, (<italic>ii</italic>) symmetrical membrane appositions, (<italic>iii</italic>) pleomorphic synaptic vesicles, and (<italic>iv</italic>) eventually its myelinated axon, when the slicing procedure preserved it. We counted the number of AZ per bouton and their individual lengths. Only synapses that had a clear synaptic cleft separating the pre- and postsynaptic elements were taken for width measurements. These measurements required a more standardized method, because the synaptic cleft boundaries are not really well defined when measuring it on EM images. A true consensus is missing on how to reduce the observer’s bias in measurements of “the real cleft width.” It is often defined as the brightest region between pre and postsynaptic membranes and usually measured as the distance between pre- and postsynaptic electro-dense (black) peaks. However, this procedure can lead to mistakes, because those peaks can vary significantly from one synapse to the other.</p><p>To solve this problem we adapted the following method consisting of three steps: (1) on digital-zoomed images, we traced five 100 nm-length straight lines from the pre- to the postsynaptic side, perpendicular to the membranes. These lines were separated from each other by 10-nm intervals; (2) we extracted the intensity profiles for each line and averaged them. This reduced the noise to better resolve the valley between peaks that is considered to be the real synaptic width; and (3) we identified the point at which the intensity of the signal decreased by 10% from the presynaptic peak to the minimal signal value into the valley (as it is shown by the gray dotted line in <bold>Figure <xref ref-type="fig" rid="F4">4C</xref></bold>) and similarly for the postsynaptic peak with respect to the minimal value into the valley. The distance between both were assumed to represent the real synaptic width. We found that our method is more robust and less biased compared to measurements by experimenter’s eye-inspection as demonstrated by manual-blind measurements <italic>vs.</italic> the standardized method (see Figure <xref ref-type="supplementary-material" rid="SM1">S1</xref>).</p></sec><sec><title>DATA ANALYSIS AND STATISTICS</title><p>Data analyses were performed using Neuromatic software package and custom routines within the IgorPro (Wavemetrics, Lake Oswego, OR, USA) or Matlab environment. All values are expressed as mean ± SEM. Student’s <italic>t</italic>-test were performed within the Excel software package (Microsoft). The significance level was established at <italic>P</italic> < 0.05 (). Cumulative distributions were compared using the Kolmogorov–Smirnov test. A homoscedasticity test was used to determine whether measurements of synaptic cleft width within a group have equal variances or not (separately for either WT or CB–/–). This was important to justify that the homeostatic rearrangement in synaptic cleft values are indeed genotypically stable changes (low variability for each genotype, independent of the number of samples). For this the test initially consists of calculating the ratio of the largest to the smallest of several sample variances for a genotype (F<sub>max</sub>). If this ratio is equal to unit, then the null hypothesis is accepted (variances are not different), but this was not the case for both genotypes. In that case, one needs to know how high F<sub>max</sub> may be for each genotype before to accept the null hypothesis. This is given by a statistical value called critical-F<sub>max</sub> = F<sub>max</sub> α[k, n-1], where α = 0.05; <italic>k</italic> = number of mice; <italic>n</italic> = number of observations per mouse. The critical-F<sub>max</sub> for both genotypes, obtained from statistic tables (<xref rid="B15" ref-type="bibr">David, 1952</xref>), presented higher values compared to F<sub>max</sub> for each genotype, demonstrating that variances are equal within each genotype group. Coefficients of variation were calculated as the square root of the ratio of SD to mean. The box-and-whiskers graphs were performed using Prism 4.0 (GraphPad software) and show the median (the line in the middle) and the 25th – 75th percentile (the box extension).</p></sec></sec><sec><title>RESULTS</title><sec><title>DEVELOPMENTAL INCREASE IN CB EXPRESSION IN MOUSE CEREBELLAR PC REACHING A PLATEAU AT PC MATURITY</title><p>The first three postnatal (PN) weeks are critical for the establishment of a mature cerebellar circuit in rodents (<xref rid="B57" ref-type="bibr">van Welie et al., 2011</xref>). Only at the end of the second PN week PC axon collateral arbors stabilize their length, spatial distribution and number of axonal boutons, a process probably achieved by cerebellar myelination that limits the expansion of axon collaterals and conserves final and stable synapses (<xref rid="B20" ref-type="bibr">Gianola et al., 2003</xref>). CB has been commonly used as “the prototypical PC marker,” but little attention has been paid to the temporal aspect of CB expression during this critical period of cerebellar development. With this aim, we quantified CB expression levels in the cerebellum of WT mice by immunofluorescence and Western blot analyses from PN5 to PN20 (<bold>Figure <xref ref-type="fig" rid="F1">1</xref></bold>). Cerebellar slices at different ages were processed simultaneously and values for the two types of measurements were normalized to PN20 levels. Slices obtained from younger animals (PN5–PN14) were imaged with the same optical settings (see Materials and Methods). CB immunofluorescence was detected at the first PN week (PN5) and the staining intensity increased progressively until PN20 (<bold>Figure <xref ref-type="fig" rid="F1">1A</xref></bold>). Semi-quantification of CB expression measured by the intensity of fluorescence in the PC layer indicated a developmental upregulation of CB (<bold>Figures <xref ref-type="fig" rid="F1">1A,B</xref></bold>). As a second method to quantify the developmental increase in CB expression levels, we performed Western blot analyses with two additional time points that confirmed the increase in CB expression during that period; a plateau was reached during the third PN week (<bold>Figures <xref ref-type="fig" rid="F1">1C,D</xref></bold>). The combined results from immunofluorescence and Western blot analyses indicated an approximately twofold increase in CB expression levels taking place between the end of the first and the end of the third PN week. This strongly indicates that morphological stabilization of PC axon collaterals and CB-regulated intracellular Ca<sup>2+</sup>-signaling processes in PC recurrent collaterals, as well as in the soma and dendrites are likely to be temporally coordinated. We then focused our further investigation on mice from PN18–PN25, time points when CB expression levels had attained steady-state levels, thus all further experiments were carried out with animals of this age range.</p><fig id="F1" position="float"><label>FIGURE 1</label><caption><p><bold>Developmental regulation of calbindin-D28k (CB) expression in mouse cerebellar PC. (A)</bold> Confocal images of the cerebellar cortex at different ages from PN5–PN20 on sagittal slices treated with anti-CB antibodies. The expression of CB is detected early during the first PN week in dendrites, somata and axons of PC. The intensity of IHC signals increases progressively until PN20. <bold>(B)</bold> Mean CB fluorescence values of the PC layer at different ages (<italic>n</italic> = 3 mice per age). Data are normalized to PN20. <bold>(C)</bold> Age-dependent increase in the total CB protein expression levels from PN6–PN20 determined by quantitative Western blots (average from 2 independent experiments, <italic>n</italic> = 3 mice per time point). The actin signal was used for normalization (norm.). In each of the 2 experiments, the normalized signal at PN20 was set as 1.00. <bold>(D)</bold> Pooled data for Western-blots of cerebella from PN6–PN25 (<italic>n</italic> = 3 mice per age) reached a plateau in CB expression at the end of the third week (ML, molecular layer; PCL, Purkinje cell layer, GCL, granule cell layer).</p></caption><graphic xlink:href="fncel-08-00364-g001"/></fig></sec><sec><title>TERMINAL BOUTONS OF PC COLLATERALS ON NEIGHBOR PC ARE SYNAPSES AND CO-LOCALIZE WITH CB</title><p>The trilaminar architecture of the cerebellar cortex is easily distinguished by CB immunolabeling of a cerebellar lobule (<bold>Figure <xref ref-type="fig" rid="F2">2A</xref></bold>). PC dendrites, somata, and axons determine the boundaries of the molecular, PC and granule cell layers, respectively. This labeling also distinguishes recurrent collateral arbors and their terminal boutons on neighbor PC (<bold>Figure <xref ref-type="fig" rid="F2">2B</xref></bold>). The strong CB immunofluorescence of PC somata makes it difficult to resolve presynaptic boutons, a requirement for detailed morphological analyses of PC–PC connections. To surmount this difficulty we loaded PC with biocytin via the patch pipette during whole cell recordings in PN18–25 mice. This approach permitted us to identify from a single PC the following structures: the main axon, recurrent collaterals terminating on the PC layer and terminal boutons on neighbor PC somata (<bold>Figure <xref ref-type="fig" rid="F2">2C</xref></bold>). We confirmed that these boutons made synapses, since they co-localized with the synaptic vesicle glycoprotein, synaptophysin (<bold>Figure <xref ref-type="fig" rid="F2">2D</xref></bold>). With this tool at hand we proceeded to analyze numerous PC boutons from both WT and CB–/– mice in order to determine their spatial distribution patterns in the presence/absence of CB.</p><fig id="F2" position="float"><label>FIGURE 2</label><caption><p><bold>calbindin D-28k labels presynaptic PC collateral boutons onto neighbor PC. (A)</bold> Confocal image of a cerebellar lobule showing that in the cerebellar cortex, CB expression is restricted to PC (PN18, sagittal slice). <bold>(B)</bold> At high resolution, CB labels the recurrent axon collateral plexus and their boutons on PC somata (white arrowheads). <bold>(C)</bold> Confocal image of one biocytin-filled PC (PC1; green)) in which the main axon (white arrow) gives rise to a recurrent collateral that returns to the PC layer (red arrow) and terminates on a neighbor PC soma (PC2, dotted red line). To reveal the axon collateral, the contrast was adapted for three adjacent areas of the final stack. <bold>(D)</bold> Zoom of the white square in <bold>(C)</bold> shows a biocytin-loaded bouton (green) on the PC layer (white square) co-localizing with synaptophysin (red). Line analysis (inset) confirmed that boutons on PC layer are <italic>bona fide</italic> PC–PC synapses.</p></caption><graphic xlink:href="fncel-08-00364-g002"/></fig></sec><sec><title>ABSENCE OF CB INDUCES A CHANGE IN NUMBER AND VOLUME OF PRESYNAPTIC TERMINALS OF PC COLLATERALS</title><p>Dendrites and the main axon of PC are aligned in an almost perfect sagittal plane with minimal transversal deviations (<xref rid="B12" ref-type="bibr">Chan-Palay, 1971</xref>; <xref rid="B28" ref-type="bibr">King and Bishop, 1982</xref>). This is also the case for the recurrent collateral arbors that are entirely confined to very thin optical stacks from the confocal reconstructions of single biocytin-loaded PC, being the thickness of this stack generally <50 μm. This planar disposition of recurrent collaterals enabled us to examine confocal projections of PC in order to test whether morphological differences existed with respect to axon collateral synapses between WT and CB–/– mice.</p><p>We compared the distance to the first branch point, the angle that gave rise to the collateral and the total length of the collateral arbor; no statistically significant differences existed between genotypes (<bold>Table <xref ref-type="table" rid="T1">1</xref></bold>). Second, we evaluated the orientation/distribution of recurrent collateral arbors and boutons. For this we superimposed somata of biocytin-loaded PC and oriented their main axons toward the white matter, in order to align cortical layers (black traces in <bold>Figure <xref ref-type="fig" rid="F3">3A</xref></bold>, <italic>n</italic> = 10 cells per genotype). Independent of the genotype, the main axons pointed in the direction of the DCN (to the base of the lobule) with their recurrent collaterals also pointing toward that direction. None to very few of recurrent collaterals returned in the opposite direction, <italic>i.e.,</italic> in the direction toward the apex of the lobule. This indicates that CB deletion in CB–/– mice did not affect the spatial orientation either of the principal PC axon or their recurrent collaterals. This is in line with a previous report demonstrating that in connected PC pairs a majority of PC collaterals (>80%) projected from the apex to the base of the lobule in both genotypes (<xref rid="B10" ref-type="bibr">Bornschein et al., 2013</xref>).</p><table-wrap id="T1" position="float"><label>Table 1</label><caption><p>Morphological measurements of axonal recurrent collaterals from WT and CB–/– mice.</p></caption><table frame="hsides" rules="groups" cellspacing="5" cellpadding="5"><thead><tr><th valign="top" align="left" rowspan="1" colspan="1"><italic>n</italic></th><th valign="top" align="left" rowspan="1" colspan="1">WT 10</th><th valign="top" align="left" rowspan="1" colspan="1">CB–/– 10</th><th valign="top" align="left" rowspan="1" colspan="1"><italic>P</italic></th></tr></thead><tbody><tr><td valign="top" align="left" rowspan="1" colspan="1">Distance until first branch (μm)</td><td valign="top" align="left" rowspan="1" colspan="1">134.68 ± 14.61</td><td valign="top" align="left" rowspan="1" colspan="1">134.02 ± 25.6</td><td valign="top" align="left" rowspan="1" colspan="1">NS</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">Angle deviation of axon collateral (°)</td><td valign="top" align="left" rowspan="1" colspan="1">145.4 ± 3.51</td><td valign="top" align="left" rowspan="1" colspan="1">141.2 ± 6.26</td><td valign="top" align="left" rowspan="1" colspan="1">NS</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">Total length collateral arbor (μm)</td><td valign="top" align="left" rowspan="1" colspan="1">807.8 ± 81.15</td><td valign="top" align="left" rowspan="1" colspan="1">950.19 ± 112.71</td><td valign="top" align="left" rowspan="1" colspan="1">NS</td></tr></tbody></table><table-wrap-foot><attrib><italic>The three values presented in the table correspond to the distance from the axon hillock to the branching point, the angle of the collateral with respect to the main axon path and the length of the entire collateral branches from the branching point, respectively (from mice aged between PN18–PN25). Data are presented as mean ± SEM. No significant differences (NS) were observed between the two genotypes (Student’s <italic>t</italic> tests).</italic></attrib></table-wrap-foot></table-wrap><fig id="F3" position="float"><label>FIGURE 3</label><caption><p><bold>Homeostatic changes in presynaptic boutons of PC recurrent collaterals of CB–/– mice. (A)</bold> Axonal bouton density of PC collaterals in the cerebellar cortex. Biocytin-loaded PC were confocally imaged and superimposed for analysis of bouton distribution (<italic>n</italic> = 10 per genotype, PN18–25, see Materials and Methods). Somata were aligned (only one soma is shown, left) and axonal plexuses were oriented in the direction of the main PC axon pointing toward the DCN (right). As indicated by the white arrow, axonal plexuses ramified from the apex to the base of the cerebellar lobule. Heat maps were built and the number of boutons per area was calculated as bouton densities (boutons/μm<sup>2</sup>). Red and blue colors indicate maximal and minimal density of boutons, respectively. Axonal boutons from WT and CB–/– PC showed a similar spatial distribution by targeting more the PC layer and by respecting the apex-to-base orientation. <bold>(B)</bold> Pooled data for boutons and branches per PC collateral in the three layers of the cerebellar cortex from PC in A. CB–/– PC showed an increase in total number of boutons and branches compared to WT PC. This difference was essentially due to an increase in GCL boutons and branches without significant changes either in PCL or in ML. <bold>(C)</bold> Confocal projections of biocytin-loaded boutons terminating on neighbor PC somata (dotted white line). White asterisks show the half-radius of the targeted PC soma center. Note the size increase of CB–/– terminal boutons when compared to WT ones. <bold>(D)</bold> 3D-projections of the examples in C (right panels) corroborated the increase in volume. <bold>(E)</bold> Histogram for bouton volume values and the cumulative distributions showed a statistically significant difference between CB–/– compared to the WTs PC boutons. <italic>P</italic> < 0.05, Student’s <italic>t</italic>-test.</p></caption><graphic xlink:href="fncel-08-00364-g003"/></fig><p>Next, we calculated the bouton density on the superimposed PC per area represented as heat color maps shown in <bold>Figure <xref ref-type="fig" rid="F3">3A</xref></bold> (for details, see Materials and Methods). In both genotypes the highest bouton densities (warm colors in the heat map) were seen to the right of the parental somata, <italic>i.e.,</italic> in the direction to the base of the lobule. A high prevalence of boutons within the PC layer was evident in comparison to the other layers. This indicates that the main postsynaptic targets of PC collaterals in either genotype were neighbor PC somata located approximately 20–80 μm away from the parental soma. However, the total number of boutons per collateral was increased in CB–/– mice compared to WT animals (<bold>Figure <xref ref-type="fig" rid="F3">3B</xref></bold>, left panel). This difference was mostly due to an increase in GCL boutons (<italic>P</italic> < 0.05) that was accompanied by an increase in the number of branching points in the GCL of the cerebellar cortex (<bold>Figure <xref ref-type="fig" rid="F3">3B</xref></bold>, right panel).</p><p>Finally, we focused on the morphology of boutons located on neighbor PC somata to explore subtle presynaptic changes in PC–PC synapses resulting from CB deletion. On images of z-stack projections of these boutons an increase in the size of CB–/– boutons compared to WT ones was observed (<bold>Figure <xref ref-type="fig" rid="F3">3C</xref></bold>). A 3D-reconstruction pointed toward an increase in volume (<bold>Figure <xref ref-type="fig" rid="F3">3D</xref></bold>), which was confirmed by counting the number of voxels per bouton. The volume distribution histograms showed a right-shift indicative of bigger values for CB–/– boutons compared to the WT boutons, also evidenced in the cumulative distribution curve shown in <bold>Figure <xref ref-type="fig" rid="F3">3E</xref></bold>. In summary, these results imply that a likely homeostatic program of morphological axonal remodeling is induced in the absence of CB. This strongly hints toward changes at the subcellular level possibly in order to cope with the altered Ca<sup>2+</sup> transients in CB–/– presynaptic PC terminals reported before (<xref rid="B10" ref-type="bibr">Bornschein et al., 2013</xref>).</p></sec><sec><title>SUBCELLULAR ADAPTATIONS OF PC–PC SYNAPSES IN CB–/– MICE</title><p>To further explore putative changes at the subcellular level of PC–PC synapses from both WT and CB–/– mice (PN20), we used EM to visualize presynaptic terminals onto PC somata and to determine their ultra-structural profile. Evidently when analyzing CB–/– mice, CB immunohistochemistry could not be used as a means to identify presynaptic terminals. To circumvent this problem we used the L7 protein (also named PCP-2), a marker exclusively localized within PC, expressed in all neuronal compartments and detected from early stages of neuronal maturation (<xref rid="B40" ref-type="bibr">Oberdick et al., 1990</xref>; <xref rid="B5" ref-type="bibr">Berrebi and Mugnaini, 1992</xref>). Moreover the L7 gene promoter has been widely used to efficiently direct targeted gene expression to PC (<xref rid="B55" ref-type="bibr">Smeyne et al., 1995</xref>) and L7 protein expression is observed throughout the PC cytosol. At the light and electron microscopic levels, L7 was visualized on somata and axonal terminals of PC from both genotypes (<bold>Figure <xref ref-type="fig" rid="F4">4A</xref></bold>). Of note, L7<sup>+</sup> boutons surrounding the PC somata were thinner in WT mice compared to CB–/– animals, which confirmed our previous observations obtained in single biocytin-loaded PC (<bold>Figure <xref ref-type="fig" rid="F3">3C</xref></bold>).</p><fig id="F4" position="float"><label>FIGURE 4</label><caption><p><bold>Ultrastructural changes of PC–PC synapses in CB–/– mice. (A)</bold> L7-immunolabelings were used to visualize axonal boutons on PC somata (white arrowheads) in both WT and CB–/– mice (PN20). WT boutons exhibited a smaller size than those from CB–/– mice. <bold>(B)</bold> Electronmicrographs of PC somata (blue background) and their presynaptic PC collateral boutons (green background) from both genotypes were recognized by their L7<sup>+</sup> labeling (arrowheads) at X 30,000 magnification. Right panels are zoomed regions to better distinguish AZs (arrows). <bold>(C)</bold> Zoomed region of AZs from <bold>(B)</bold> (indicated by arrows with an asterisk on the right panels) showed an enlargement of synaptic cleft width in the CB–/– synapses compared to WT synapses (top). The line analysis method to properly measure synaptic cleft widths (bottom). Averages of five lines including pre/postsynaptic densities per synapse were normalized to peak/valley (see Materials and Methods). <bold>(D)</bold> Histograms and cumulative distributions showed that in CB–/– PC–PC synapses the right shift in the histogram (upper part) and cumulative distribution plot (lower part) indicate larger synaptic cleft values at CB–/– PC–PC synapses.</p></caption><graphic xlink:href="fncel-08-00364-g004"/></fig><p>As shown in <bold>Figure <xref ref-type="fig" rid="F4">4B</xref></bold>, we identified PC–PC synapses by a detailed inspection of the somatic PC plasma membrane in order to localize L7<sup>+</sup> boutons with symmetrical synaptic densities and pleomorphic synaptic vesicles as previously reported (<xref rid="B30" ref-type="bibr">Larramendi and Victor, 1967</xref>; <xref rid="B12" ref-type="bibr">Chan-Palay, 1971</xref>). The number of AZ per bouton was rather small (in the order of 1 – 3) and similar when comparing WT (1.19 ± 0.08; <italic>n</italic> = 32 from 2 mice) <italic>vs</italic>. CB–/– synapses (1.38 ± 0.1, <italic>n</italic> = 35 from 2 mice, <italic>P</italic> = NS). However, the length of individual AZ was 23% larger at CB–/– synapses (292.08 ± 15.89 nm, <italic>n</italic> = 35) compared to AZ of WT synapses (237.01 ± 14.43 nm, <italic>n</italic> = 32, <italic>P</italic> < 0.05). We also measured the synaptic cleft width between pre-post synaptic membranes of those synapses by using a newly developed method aimed to standardize the measurements and to maximally reduce experimenter bias (<bold>Figure <xref ref-type="fig" rid="F4">4C</xref></bold>; also see Materials and Methods and Figure <xref ref-type="supplementary-material" rid="SM1">S1</xref>). Interestingly, we found that CB–/– cleft widths were bigger than WT ones (23.8 ± 0.43 nm <italic>vs.</italic> 21.17 ± 0.39 nm, respectively; <italic>P</italic> < 0.001; <bold>Figure <xref ref-type="fig" rid="F4">4D</xref></bold>). To determine whether variances are equal or not within each group we performed homoscedasticity tests. In this test one calculates first the ratio of the largest to the smallest of the sample variances within one group (F<sub>max</sub>). If this ratio is bigger than unit, then one calculates what values F<sub>max</sub> can attain (critical-F<sub>max</sub>) before one needs to reject the null hypotheses, which is that “variances are different” (see Materials and Methods). Neither for WT nor for CB–/– mice, F<sub>max</sub> exceeded the critical-F<sub>max</sub> values (1.19 <italic>vs.</italic> 2.46 for WT and 1.47 <italic>vs.</italic> 3.72, for CB–/–). This corroborated that synaptic cleft width values were homogeneous within each genotype and are in support that CB deletion in CB–/– mice induced a steady and constant change in the pre-post synaptic distance, which is assumed to have an impact on released GABA reaching the postsynaptic PC somata and thus on PC–PC IPSC characteristics (<xref rid="B10" ref-type="bibr">Bornschein et al., 2013</xref>).</p><p>Based on the observed changes in AZ length in CB–/– mice, we determined the vesicle number within this region, since it constitutes the site of synaptic vesicle clustering, docking and transmitter release (<xref rid="B44" ref-type="bibr">Rizzoli and Betz, 2005</xref>). Vesicles were counted in (<italic>i)</italic> the vicinity of the AZ, a region defined as a half circle with a diameter of the AZ length and (<italic>ii</italic>) in proximity of the AZ membrane, the latter representing the docked vesicle population (<bold>Figure <xref ref-type="fig" rid="F5">5</xref></bold>). In both compartments, the number of vesicles was higher in CB–/– presynaptic terminals compared to WT, however, the overall vesicle density was similar in both genotypes: 8.97 ± 0.82 <italic>vs.</italic> 7.21 ± 0.58 vesicles per 100 nm<sup>2</sup> for WT and CB–/– synapses, respectively (<italic>P</italic> = NS). This indicates that the number of vesicles was proportional to the AZ length, irrespective of the genotype and the vesicle/AZ length relationship is shown in <bold>Figure <xref ref-type="fig" rid="F5">5C</xref></bold>. The vesicle number depended linearly on the AZ length, as it has been previously reported also for glutamatergic synapses (<xref rid="B24" ref-type="bibr">Holderith et al., 2012</xref>). Thus, the higher number of docked vesicles in PC–PC CB–/– synapses is in line with the observation of bigger <italic>q</italic> quanta values previously recorded in PC–PC CB–/– synapses (<xref rid="B10" ref-type="bibr">Bornschein et al., 2013</xref>). Of note the coefficient of determination (R) was smaller in CB–/– synapses compared to WT synapses, for both, the docked vesicles and the ones in the AZ vicinity. This might be an indication of a less tight organization of vesicles in the CB–/– presynaptic compartment and that potentially more vesicles were available for GABA release. This in turn, could underlie the reported increase in <italic>p</italic>. Thus, our results indicate that the absence of CB also affects the distribution/organization of vesicles at this synapse, likely also affecting the exocytotic machinery and consequently GABA release properties. Interestingly, in line with our results, an increase in the amount of Ca<sup>2+</sup> entering the presynaptic terminal, <italic>e.g.,</italic> via Ca<sub>V</sub>2-type Ca<sup>2+</sup> channels is often paralleled by an increase in the AZ size and the size of the ready-releasable pool of vesicles (<xref rid="B19" ref-type="bibr">Frank, 2014</xref>); other reported presynaptic HSP mechanisms include changes in <italic>p</italic> (<xref rid="B37" ref-type="bibr">Murthy et al., 2001</xref>).</p><fig id="F5" position="float"><label>FIGURE 5</label><caption><p><bold>The number of synaptic vesicles scales linearly with AZ length in both WT and CB–/– presynaptic boutons. (A)</bold> The scheme shows the two approaches to count vesicles within a bouton in respect to the AZ. Vesicles in the “AZ vicinity” are considered as those located in an area covered by a half-circle with the diameter of the AZ length (blue region). Vesicles localized at a distance <10 nm from the plasma membrane of the AZ were counted as “docked vesicles.” <bold>(B)</bold> Pooled data from the two defined regions. CB–/– boutons contained more vesicles in the zone defined as “AZ vicinity” as well as docked vesicles per AZ. <bold>(C)</bold> Comparison between vesicle numbers and AZ length for each bouton. A strong linear relationship between AZ length and either docked or associated vesicles strongly indicated that the AZ length dictates the number of vesicles immediately ready to be released as well as those belonging to the pool for slower release. Statistical differences were calculated by Student’s <italic>t</italic> tests (<italic>P</italic> < 0.01; *<italic>P</italic> < 0.001). A 2-tailed test of significance was used for linear correlations (Pearson correlation) in <bold>(C)</bold> showing <italic>P</italic> values < 0.001 with the exception of <italic>P</italic> < 0.01 for the right bottom panel. R, coefficient of determination; S, slope.</p></caption><graphic xlink:href="fncel-08-00364-g005"/></fig></sec></sec><sec><title>DISCUSSION</title><p>The brain is characterized by an extraordinary degree of plasticity, however, controlled by the presence of homeostatic signaling systems to keep the excitatory and inhibitory activity (E/I balance) within a “physiological” range thus allowing for a stable communication between neurons. From the viewpoint of information transfer at synapses, it is evident that synapses are under a bidirectional homeostatic control (<xref rid="B60" ref-type="bibr">Vitureira et al., 2012</xref>). By the process of HSP, neurons may modulate their excitability, firing properties and short- and/or long-term synaptic modulation in response to changes in activity in either direction, <italic>i.e.,</italic> as the result of increased or decreased net activity (<xref rid="B32" ref-type="bibr">Lee et al., 2014</xref>). Given the importance of Ca<sup>2+</sup> ions in the process of synaptic transmission, both pre- and postsynaptically, an involvement of many components of the Ca<sup>2+</sup>-signaling toolkit (<xref rid="B6" ref-type="bibr">Berridge et al., 2003</xref>) in essentially all forms of synaptic plasticity has been reported (<xref rid="B11" ref-type="bibr">Cavazzini et al., 2005</xref>). Examples include presynaptic high-voltage activated Ca<sub>V</sub>2-type Ca<sup>2+</sup> channels that gate forms of HSP. Weakened synapse activity leads to increased influx through Ca<sub>V</sub>2 channels, while enhanced influx via Ca<sub>V</sub>1-type channels elicits homeostatic adaptation by removal of postsynaptic excitatory receptors (for details, see <xref rid="B19" ref-type="bibr">Frank, 2014</xref>). The entity of all molecules that build the network of Ca<sup>2+</sup> signaling components, and that are involved in their own regulation as to maintain physiological Ca<sup>2+</sup> homeostasis resulting in phenotypic stability is named the Ca<sup>2+</sup> homeostasome (<xref rid="B50" ref-type="bibr">Schwaller, 2009</xref>, <xref rid="B52" ref-type="bibr">2012a</xref>).</p><p>The cerebellum represents a model system to investigate synaptic plasticity and the involved Ca<sup>2+</sup>-dependent processes, both accessible to experimental (reviewed in <xref rid="B29" ref-type="bibr">Lamont and Weber, 2012</xref>) and modeling approaches (<xref rid="B1" ref-type="bibr">Achard and Schutter, 2008</xref>). This highly repetitive structure consisting of relatively few distinct elements and a rather stereotyped wiring pattern, has allowed to investigate the role of proteins implicated in Ca<sup>2+</sup> signaling in the various forms of plasticity. While rather much attention was given to systems implicated in Ca<sup>2+</sup> entry from the extracellular side and to release mechanisms from internal stores, as well as systems leading to a decrease in [Ca<sup>2+</sup>]<sub>i</sub>, studies on the role of intracellular Ca<sup>2+</sup>-binding proteins in synaptic plasticity in the cerebellum are rather underrepresented. The “fast” buffer CB and the “slow buffer” PV are expressed in PC and PV additionally in stellate and basket cells. The effects that the absence of these proteins entail in PC and MLIs have been described in detail, both at the functional level (<xref rid="B49" ref-type="bibr">Schmidt et al., 2003</xref>; <xref rid="B14" ref-type="bibr">Collin et al., 2005</xref>; <xref rid="B18" ref-type="bibr">Franconville et al., 2011</xref>; <xref rid="B10" ref-type="bibr">Bornschein et al., 2013</xref>; reviewed in <xref rid="B53" ref-type="bibr">Schwaller, 2012b</xref>), but also at the level of PC morphology (<xref rid="B58" ref-type="bibr">Vecellio et al., 2000</xref>; <xref rid="B13" ref-type="bibr">Chen et al., 2006</xref>). The increased spine head volume and the longer spine shafts on CB–/– PC dendrites had been discussed as a HSP mechanism possibly contributing to unaltered LTD in CB–/– mice, although dendritic and spine [Ca<sup>2+</sup>]<sub>i</sub> dynamics were altered in the absence of CB (<xref rid="B4" ref-type="bibr">Barski et al., 2003</xref>). With respect to the presynaptic function of CB, previous studies on recurrent PC axon collaterals forming synapses on neighboring PC (<xref rid="B42" ref-type="bibr">Orduz and Llano, 2007</xref>) have lead to several unexpected findings in CB-deficient mice (<xref rid="B10" ref-type="bibr">Bornschein et al., 2013</xref>). Although [Ca<sup>2+</sup>]<sub>i</sub> amplitudes evoked by 10 APs delivered at 200 Hz were clearly larger in PC boutons of CB–/– PC, the evoked IPSC in the postsynaptic PC was unaltered and also PPF characteristics was unchanged in PN7–PN12 mice. This is at an age when CB expression levels have not yet attained adult levels and based on our results were estimated to be in the order of 50% compared to values in PN18–PN25 mice (<bold>Figure <xref ref-type="fig" rid="F1">1</xref></bold>). This relates to a CB concentration in the range of 50 – 180 μM, since CB levels in PC of adult mice were reported to be in the range of 100 – 360 μM (for details, see <xref rid="B48" ref-type="bibr">Schmidt, 2012</xref>). Based on a modeling approach, the authors proposed that slow Ca<sup>2+</sup> unbinding from the sensor for transmitter release was the main determinant for PPF dynamics and thus independent from CB (<xref rid="B10" ref-type="bibr">Bornschein et al., 2013</xref>). This is in contrast to the synapse between cortical CB-expressing multipolar bursting (MB) cells and pyramidal cells, where rapid Ca<sup>2+</sup> buffer (CB) saturation resulted in a decreased IPSC amplitude of the first response and increased PPF by this buffer effect, a process termed “facilitation by Ca<sup>2+</sup> buffer (CB) saturation (<xref rid="B35" ref-type="bibr">Maeda et al., 1999</xref>; <xref rid="B7" ref-type="bibr">Blatow et al., 2003</xref>) or “pseudo-facilitation” (<xref rid="B39" ref-type="bibr">Neher, 1998</xref>; <xref rid="B45" ref-type="bibr">Rozov et al., 2001</xref>; <xref rid="B71" ref-type="bibr">Zucker and Regehr, 2002</xref>). These experiments had been carried out in acute slices after washout of CB via the patch pipette (representing the CB–/– situation) and reverted by loading of terminals with CB or BAPTA, in both cases not allowing for homeostatic plasticity mechanisms to come into play. Additional experiments showed that PPF at MB cell terminals depend on Ca<sup>2+</sup> influx rather than on the initial <italic>p</italic>.</p><p>In PC–PC connections in CB–/– mice, both <italic>p</italic> and <italic>q</italic> are increased, findings that hinted toward an induction of likely homeostatic mechanisms possibly aimed at restoring IPSC amplitudes and STP as seen in WT terminals. Both parameters, <italic>p</italic> and <italic>q</italic>, strongly correlate with bouton volume (<xref rid="B47" ref-type="bibr">Schikorski and Stevens, 1997</xref>; <xref rid="B37" ref-type="bibr">Murthy et al., 2001</xref>) and peak amplitude of presynaptic [Ca<sup>2+</sup>] transients has been seen positively correlated with the AZ area (<xref rid="B24" ref-type="bibr">Holderith et al., 2012</xref>). Such a volume increase was also evident in PC boutons from CB–/– mice (<bold>Figure <xref ref-type="fig" rid="F3">3</xref></bold>); moreover an increase in bouton volume is also associated with a larger AZ, paralleled by an almost identical increase in the size of the PSD (<xref rid="B37" ref-type="bibr">Murthy et al., 2001</xref>). While a larger AZ encompasses more docked vesicles (<xref rid="B24" ref-type="bibr">Holderith et al., 2012</xref>), which is in line with an increase in the magnitude of <italic>q</italic>, such a straightforward correlation appears not to hold true for the <italic>p</italic>. Several possibilities were considered to explain differences in <italic>p</italic> at CF and PF synapses, despite the number of docked vesicles being similar (<xref rid="B69" ref-type="bibr">Xu-Friedman et al., 2001</xref>). Alterations in the priming process of docked vesicles or in the phosphorylation state of proteins implicated in the exocytotic machinery, but also changes in the Ca<sup>2+</sup> signal resulting from changes in Ca<sup>2+</sup> influx or Ca<sup>2+</sup> buffering were discussed to affect <italic>p</italic>. An example of the former was observed in boutons of hippocampal neurons, where an increase in Ca<sup>2+</sup> influx by TTX pretreatment increased <italic>p</italic>, while the number of readily releasable vesicles was only marginally reduced (<xref rid="B70" ref-type="bibr">Zhao et al., 2011</xref>).</p><p>Although EM 3D-reconstructions from serial ultrathin sections could permit a most accurate view of the ultrastructure (<xref rid="B43" ref-type="bibr">Pierce and Mendell, 1993</xref>), we considered the AZ/PSD length of recurrent PC axon collaterals as a proxy measure for the area of these appositions, <italic>i.e.,</italic> we assumed these contacts to be circular in shape. Accepting these shortcomings, the area of a PC–PC contact was calculated to be approximately 50% larger in CB–/– terminals. Studies that reported on the width of the synaptic cleft found the cleft width to show extremely little variation (SD in the order of <5%) for a given synapse type (<xref rid="B25" ref-type="bibr">Jing et al., 2004</xref>; <xref rid="B68" ref-type="bibr">Xu and Zhang, 2006</xref>; <xref rid="B66" ref-type="bibr">Xu et al., 2013a</xref>,<xref rid="B67" ref-type="bibr">b</xref>; <xref rid="B21" ref-type="bibr">Han et al., 2014</xref>). Changes in cleft width are most often the result of experimental manipulations including OGD, “metal-poisoning” by lead and aluminum (<xref rid="B25" ref-type="bibr">Jing et al., 2004</xref>; <xref rid="B34" ref-type="bibr">Lushnikova et al., 2011</xref>; <xref rid="B21" ref-type="bibr">Han et al., 2014</xref>), but were also observed in genetic models, <italic>e.g.,</italic> CNTNAP4-KO mice (<xref rid="B26" ref-type="bibr">Karayannis et al., 2014</xref>). Of note cleft width changes appear to be independent from changes in the length/area of the AZ. After 60 min OGD, AZ area increased by 58% (<xref rid="B34" ref-type="bibr">Lushnikova et al., 2011</xref>) and also the volume of the presynaptic terminal was augmented. In the other models, where cleft width was increased, the length of the AZ was either unaltered, as in the case of estradiol benzoate treatment (Xu et al., 2006) or even reduced as in the cases of bisphenol-A-mediated inhibition of synaptogenesis (<xref rid="B66" ref-type="bibr">Xu et al., 2013a</xref>,<xref rid="B67" ref-type="bibr">b</xref>) and exposure to either lead (<xref rid="B21" ref-type="bibr">Han et al., 2014</xref>) or aluminum (<xref rid="B25" ref-type="bibr">Jing et al., 2004</xref>).</p><p>One of the problems with measuring cleft width on EM images is the procedure used to accurately measure this parameter. Different approaches have been chosen. Here we have compared a manual and a standardized procedure for measuring cleft width. Although absolute values are different when applying the 2 methods (Figure <xref ref-type="supplementary-material" rid="SM1">S1</xref>), synaptic cleft width was clearly larger in CB–/– PC–PC synapses. The volume of the synaptic cleft was then estimated to be a cylinder defined by the surface area of the AZ/PSD and the height of the synaptic cleft; this volume was increased by 70% in CB–/– PC–PC synapses. Thus, it is foreseeable that the increased volume would suffice to diminish the GABA concentration at the surface of the postsynaptic side and thus IPSC amplitude, as previously modeled for excitatory synaptic transmission (<xref rid="B62" ref-type="bibr">Wahl et al., 1996</xref>). This, in turn may lead to a postsynaptic response (IPSC) of similar magnitude as observed in WT PC (<bold>Figure <xref ref-type="fig" rid="F6">6</xref></bold>).</p><fig id="F6" position="float"><label>FIGURE 6</label><caption><p><bold>Summary of the observed and proposed alterations at PC–PC synapses due to CB deletion. (A)</bold> An AP arrives at the presynaptic terminal and causes opening of voltage-operated Ca<sup>2+</sup> channels (VOCC). The resulting intracellular rise in [Ca<sup>2+</sup>]<sub>i</sub> is shaped by the presence of CB (<xref rid="B10" ref-type="bibr">Bornschein et al., 2013</xref>), which determines the efficacy of synaptic transmission. In the absence of CB, the maximal amplitude of Ca<sup>2+</sup> transients is increased in CB–/– presynaptic boutons compared to WT controls. <bold>(B)</bold> To possibly reduce the impact of the Ca<sup>2+</sup> transients on the amount of GABA released into the synaptic cleft, the volume (and surface) of presynaptic boutons of CB–/– mice is increased. <bold>(C)</bold> This is also reflected by a stretching of the AZ length, which is accompanied by a proportional increase in the number of ready releasable vesicles. <bold>(D)</bold> This might explain the enhancement in synaptic efficacy characterized by increased <italic>p</italic> and <italic>q</italic> previously observed at CB–/– PC–PC synapses (<xref rid="B10" ref-type="bibr">Bornschein et al., 2013</xref>). Yet postsynaptic PC responses with respect to IPSC mean amplitudes or STP are indistinguishable between PC pairs from WT or CB–/– mice (<xref rid="B10" ref-type="bibr">Bornschein et al., 2013</xref>). <bold>(E)</bold> An enlargement of synaptic cleft width and AZ length resulting in an approximately 70% larger volume of the synaptic cleft might reduce the GABA concentration reaching the postsynaptic GABA receptors finally resulting in unaltered IPSC amplitudes and PPF characteristics at CB–/– PC–PC synapses. It can’t be excluded that other mechanisms, <italic>e.g.,</italic> subtle alterations in the subunit composition of GABA<sub>A</sub> receptors might also contribute to this likely homeostatic mechanism of plasticity.</p></caption><graphic xlink:href="fncel-08-00364-g006"/></fig><p>Besides the changes in the architecture of PC–PC contacts in CB–/– mice, the number of boutons per PC collateral was increased. Studies on cerebellar synapse formation during postnatal development have mostly focused on the formation of PF–PC synapses and CF–PC synapses. In the mouse the process of CF–PC synapse formation/elimination is a multistep process consisting of initially multiple innervation of a PC soma by several CF (PN3) followed by functional differentiation (PN7). At this time point also PF synapses, as well as MLI synapses start to form. The later stages consist of CF synapse translocation essentially to PC dendrites (PN9) and an early (PN8–PN11) and late (PN12–PN17) phase of CF elimination, resulting in a single CF innervating each PC (reviewed in <xref rid="B23" ref-type="bibr">Hashimoto and Kano, 2013</xref>). What is currently known about the temporal development of PC–PC synapses? The PC intracortical plexus starts to form during the first postnatal week characterized by sprouting with a maximal branching reached around PN6 (<xref rid="B20" ref-type="bibr">Gianola et al., 2003</xref>). PC collateral start to form synapses around PN7, just at the time when functional differentiation of CF synapses is finished. CF synapse formation is considered to be the first step during cerebellar synaptogenesis (<xref rid="B31" ref-type="bibr">Larramendi, 1969</xref>). The second postnatal week is then exemplified by augmented structural plasticity consisting of trimming of collateral branches to less than half compared to PN6 and in remodeling of terminal arbors. Excess fibers that make up this tangled plexus disappear largely around PN20, but stable long-term connections are preserved from PN15 onward in rats (<xref rid="B20" ref-type="bibr">Gianola et al., 2003</xref>). Paired PC–PC recordings in young rodents revealed a decrease in the probability of connections between the first and second PN weeks compared to the third PN week: 26% at PN4–14 (<xref rid="B63" ref-type="bibr">Watt et al., 2009</xref>) and 10% at PN15–19 (<xref rid="B42" ref-type="bibr">Orduz and Llano, 2007</xref>). This decrease may be viewed as an advantage of CF–PC somatic innervation <italic>vs.</italic> PC–PC connections during postnatal development (<xref rid="B31" ref-type="bibr">Larramendi, 1969</xref>). Of interest, a mathematical model recapitulating <italic>in vivo</italic> recordings of cerebellar oscillations in adult rats, in which PC–PC connections are fully operational, requires a connection probability of 20% between PC in order to reproduce this oscillatory behavior (<xref rid="B16" ref-type="bibr">de Solages et al., 2008</xref>). Thus, the higher number of axonal branches and the increased number of boutons in the GCL of CB–/– mice might, in part, explain the strong 160-Hz oscillations observed in adult CB–/– mice that are essentially absent in WT mice (<xref rid="B54" ref-type="bibr">Servais et al., 2005</xref>). Thus, one might speculate that the situation of PC collaterals in CB–/– mice resembles the situation of early development, when collaterals are more numerous, inhibition via MLI has not taken place yet, resulting in synchronization in the sagittal plane in the form of traveling waves (<xref rid="B63" ref-type="bibr">Watt et al., 2009</xref>).</p><p>We reasoned that the time window of the formation of PC–PC synapses would be the same as for spine development, <italic>i.e.,</italic> in conjunction with the functional maturation of the cerebellum. In developing hippocampal neurons during the period of rapid synaptogenesis <italic>in vitro</italic>, the blocking of spike activity by tetrodotoxin (TTX) reduces the density of inhibitory synapses, both onto glutamatergic and GABAergic target neurons, without affecting density of glutamatergic synapses (<xref rid="B22" ref-type="bibr">Hartman et al., 2006</xref>). Yet in other experimental settings using cortical or hippocampal cultures, also the density of excitatory synapses was found to be increased after TTX treatment (<xref rid="B64" ref-type="bibr">Wierenga et al., 2006</xref>) and decreasing neuronal circuit activity in hippocampal cultures resulted in an increase in the number of connected neuron pairs (<xref rid="B38" ref-type="bibr">Nakayama et al., 2005</xref>). Thus, both excitatory as well as inhibitory synapses are receptive for activity-dependent modulation by modifying synapse numbers. Here we propose that the increase in CB–/– PC synaptic boutons might be viewed as a means to increase inhibition onto close neighbor PC. The most prominent increase in axonal boutons was observed in the GCL and less in the PCL indicating that possibly axo-axonic and not axo-somatic inhibition might be increased. Thus, these morphological findings indicate that the output of an individual PC to the DCN is possibly more strongly controlled by neighbor PC. Another interesting point concerns the fact that myelin formation shapes cerebellar connections by removing excess collateral branches of Purkinje neurons (<xref rid="B20" ref-type="bibr">Gianola et al., 2003</xref>). In the case of CB–/– mice, the increased branching in the GCL, the layer in which more myelin is wrapped around the PC axons might be indicative of a permissive signal generating branching and possibly more synaptic contacts onto postsynaptic targets of PC axon collaterals, <italic>e.g.,</italic> on other PC axons, on interneurons or even on NG2<sup>+</sup> progenitors (<xref rid="B9" ref-type="bibr">Boda and Buffo, 2014</xref>).</p><p>What are the consequences of the absence of CB with respect to cerebellar function and do the reported HSP mechanisms suffice to “prevent” a CB–/– motor phenotype? As mentioned above, LTD is not affected in CB–/– mice, while motor coordination and motor learning are impaired (<xref rid="B4" ref-type="bibr">Barski et al., 2003</xref>; <xref rid="B17" ref-type="bibr">Farré-Castany et al., 2007</xref>), as shown in the runway assay, and by measuring the optokinetic reflex OKR (<xref rid="B4" ref-type="bibr">Barski et al., 2003</xref>). This has, in part, been attributed to the presence of 160 Hz cerebellar oscillations recorded from alert CB–/– mice that are essentially absent in WT mice (<xref rid="B54" ref-type="bibr">Servais et al., 2005</xref>). These oscillations are the result of synchronous activity along the PF beam with a likely contribution of the recurrent PC axon collaterals (<xref rid="B36" ref-type="bibr">Maex and Schutter, 2005</xref>). Whether higher cognitive functions are impaired in CB–/– mice is currently unknown. A decrease in CB expression levels and/or CB-ir neurons has been reported in several neurological diseases including schizophrenia, bipolar disorder and autism spectrum disorders (<xref rid="B65" ref-type="bibr">Wondolowski and Dickman, 2013</xref>). It remains to be investigated whether CB–/– mice also show behavioral changes reminiscent of these pathologies.</p></sec><sec><title>AUTHOR CONTRIBUTIONS</title><p>David Orduz, David Gall, Serge N. Schiffmann, and Beat Schwaller conceived and designed the experiments; David Orduz, Alain Boom, and Beat Schwaller performed research; David Orduz and Beat Schwaller analyzed the data; all authors interpreted the data; David Orduz and Beat Schwaller drafted the manuscript, all authors critically revised the manuscript and approved the final version of the manuscript.</p></sec><sec><title>Conflict of Interest Statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec>
Addiction and reward-related genes show altered expression in the postpartum nucleus accumbens	<p>Motherhood involves a switch in natural rewards, whereby offspring become highly rewarding. Nucleus accumbens (NAC) is a key CNS region for natural rewards and addictions, but to date no study has evaluated on a large scale the events in NAC that underlie the maternal change in natural rewards. In this study we utilized microarray and bioinformatics approaches to evaluate postpartum NAC gene expression changes in mice. Modular Single-set Enrichment Test (MSET) indicated that postpartum (relative to virgin) NAC gene expression profile was significantly enriched for genes related to addiction and reward in five of five independently curated databases (e.g., Malacards, Phenopedia). Over 100 addiction/reward related genes were identified and these included: <italic>Per1</italic>, <italic>Per2</italic>, <italic>Arc</italic>, <italic>Homer2</italic>, <italic>Creb1</italic>, <italic>Grm3</italic>, <italic>Fosb</italic>, <italic>Gabrb3</italic>, <italic>Adra2a</italic>, <italic>Ntrk2</italic>, <italic>Cry1</italic>, <italic>Penk</italic>, <italic>Cartpt</italic>, <italic>Adcy1</italic>, <italic>Npy1r</italic>, <italic>Htr1a</italic>, <italic>Drd1a</italic>, <italic>Gria1</italic>, and <italic>Pdyn</italic>. ToppCluster analysis found maternal NAC expression profile to be significantly enriched for genes related to the drug action of nicotine, ketamine, and dronabinol. Pathway analysis indicated postpartum NAC as enriched for RNA processing, CNS development/differentiation, and transcriptional regulation. Weighted Gene Coexpression Network Analysis (WGCNA) identified possible networks for transcription factors, including <italic>Nr1d1</italic>, <italic>Per2</italic>, <italic>Fosb</italic>, <italic>Egr1</italic>, and <italic>Nr4a1</italic>. The postpartum state involves increased risk for mental health disorders and MSET analysis indicated postpartum NAC to be enriched for genes related to depression, bipolar disorder (BPD), and schizophrenia. Mental health related genes included: <italic>Fabp7</italic>, <italic>Grm3</italic>, <italic>Penk</italic>, and <italic>Nr1d1</italic>. We confirmed via quantitative PCR <italic>Nr1d1</italic>, <italic>Per2</italic>, <italic>Grm3</italic>, <italic>Penk</italic>, <italic>Drd1a</italic>, and <italic>Pdyn</italic>. This study indicates for the first time that postpartum NAC involves large scale gene expression alterations linked to addiction and reward. Because the postpartum state also involves decreased response to drugs, the findings could provide insights into how to mitigate addictions.</p>	<contrib contrib-type="author"><name><surname>Zhao</surname><given-names>Changjiu</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/133721"/></contrib><contrib contrib-type="author"><name><surname>Eisinger</surname><given-names>Brian Earl</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/128292"/></contrib><contrib contrib-type="author"><name><surname>Driessen</surname><given-names>Terri M.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/128976"/></contrib><contrib contrib-type="author"><name><surname>Gammie</surname><given-names>Stephen C.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><xref ref-type="author-notes" rid="fn001"><sup>*</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/29054"/></contrib>	Frontiers in Behavioral Neuroscience	<sec sec-type="introduction" id="s1"><title>Introduction</title><p>The bond of mother to offspring has been suggested to be the primary social bond in most mammals (Broad et al., <xref rid="B8" ref-type="bibr">2006</xref>) and is a core component of the maternal phenotype. Conserved reward CNS circuitries underlie key life history traits, including sex and eating, and these CNS regions include: medial prefrontal cortex (mPFC), ventral tegmental area (VTA), medial preoptic area (MPOA), and nucleus accumbens (NAC; Kelley and Berridge, <xref rid="B32" ref-type="bibr">2002</xref>). Multiple studies indicate that maternal and social bonding are also regulated by these circuits (Insel, <xref rid="B30" ref-type="bibr">2003</xref>; Numan and Insel, <xref rid="B58" ref-type="bibr">2003</xref>; Burkett and Young, <xref rid="B11" ref-type="bibr">2012</xref>; Olazábal et al., <xref rid="B59" ref-type="bibr">2013</xref>). Among the evidence that mothering involves reward and incentive pathways is that these CNS regions show brain activation in response to offspring cues in humans (Lorberbaum et al., <xref rid="B46" ref-type="bibr">2002</xref>; Bartels and Zeki, <xref rid="B3" ref-type="bibr">2004</xref>; Noriuchi et al., <xref rid="B57" ref-type="bibr">2008</xref>; Strathearn et al., <xref rid="B82" ref-type="bibr">2008</xref>) and rodents (Febo et al., <xref rid="B22" ref-type="bibr">2005</xref>; Ferris et al., <xref rid="B23" ref-type="bibr">2005</xref>; Febo, <xref rid="B21" ref-type="bibr">2011</xref>). Also, rodent mothers will bar press (Wilsoncroft, <xref rid="B87" ref-type="bibr">1968</xref>; Hauser and Gandelman, <xref rid="B27" ref-type="bibr">1985</xref>; Lee et al., <xref rid="B41" ref-type="bibr">1999</xref>) and show a place preference for pups (Mattson et al., <xref rid="B51" ref-type="bibr">2001</xref>, <xref rid="B52" ref-type="bibr">2003</xref>) as they would for other rewarding stimuli. Additionally, drugs of abuse act by coopting natural reward circuitries (Kelley and Berridge, <xref rid="B32" ref-type="bibr">2002</xref>) and mothering mitigates the rewarding properties of addictive drugs, such as cocaine (Mattson et al., <xref rid="B51" ref-type="bibr">2001</xref>, <xref rid="B52" ref-type="bibr">2003</xref>; Mattson and Morrell, <xref rid="B50" ref-type="bibr">2005</xref>; Seip and Morrell, <xref rid="B75" ref-type="bibr">2007</xref>).</p><p>Nucleus accumbens’ role in maternal incentive/reward processes has been evaluated from multiple approaches. For example, NAC shows elevated fMRI activation when either human mothers receive cues from children (Lorberbaum et al., <xref rid="B46" ref-type="bibr">2002</xref>; Atzil et al., <xref rid="B2" ref-type="bibr">2011</xref>) or postpartum rats are nursed (Febo, <xref rid="B21" ref-type="bibr">2011</xref>). Dopamine release increases in NAC in rat mothers when they interact with pups [22]. Further, cocaine-induced NAC activation is significantly reduced in rat mothers (Ferris et al., <xref rid="B23" ref-type="bibr">2005</xref>), suggesting reward salience has been shifted to pups whereby pups have become a natural form of addiction. Additionally, pharmacological manipulations and lesions of NAC modulate maternal care (Hansen et al., <xref rid="B26" ref-type="bibr">1991</xref>; Stolzenberg et al., <xref rid="B80" ref-type="bibr">2007</xref>). Pair bonding in prairie voles has many similarities to maternal bonding and pair bonding is linked to NAC function (Young et al., <xref rid="B90" ref-type="bibr">2011</xref>).</p><p>Many studies of NAC in natural and drug reward have focused on dopamine and opioid signaling (Pettit et al., <xref rid="B65" ref-type="bibr">1984</xref>; Spanagel et al., <xref rid="B78" ref-type="bibr">1990</xref>; Leone et al., <xref rid="B42" ref-type="bibr">1991</xref>; Peciña and Berridge, <xref rid="B64" ref-type="bibr">2013</xref>), but the human and mouse genome both have more than 20,000 genes and it is becoming increasingly clear that hundreds of genes are linked to drug addiction and reward behaviors (Li et al., <xref rid="B45" ref-type="bibr">2011</xref>; Spanagel, <xref rid="B77" ref-type="bibr">2013</xref>). Further, despite NAC’s links to maternal attachment and addictive processes, no study to date has examined large scale gene expression changes that occur in NAC during this switch in natural reward. In this study, we use microarrays to evaluate large scale gene expression changes that occur in NAC in postpartum females (compared to non-maternal females) using a mouse model. In order to identify genes with connections to reward and addiction, we used our recently developed software tool, Modular Single-set Enrichment Test (MSET), that allows one to evaluate any large scale gene expression dataset against any disease, disorder, pathway, or hand curated database (Eisinger et al., <xref rid="B19" ref-type="bibr">2013a</xref>, <xref rid="B18" ref-type="bibr">2014</xref>; Driessen et al., <xref rid="B16" ref-type="bibr">2014a</xref>). We identified five independently curated databases that maintain lists of genes related to addiction or reward and then tested our significant array results for enrichment against each database. ToppCluster was used to determine whether array results were similar to genes associated with the action of specific drugs or chemicals. Because the maternal state involves an increased risk for mental health disorders (Brockington, <xref rid="B9" ref-type="bibr">2004</xref>), we also used MSET to evaluate whether array results had similarities to mental health disorders using a range of independently curated databases as in recent studies (Eisinger et al., <xref rid="B19" ref-type="bibr">2013a</xref>, <xref rid="B18" ref-type="bibr">2014</xref>; Driessen et al., <xref rid="B16" ref-type="bibr">2014a</xref>). NIH DAVID and ToppCluster were both employed to conduct additional pathway analysis. Weighted Gene Coexpression Network Analysis (WGCNA) identified clusters of genes whose expression may be linked transcriptionally.</p></sec><sec sec-type="methods" id="s2"><title>Methods</title><sec id="s2-1"><title>Animals</title><p>Outbred hsd:ICR female mice (Harlan, Indianapolis, IN, USA) (~70 days of age) were used and procedures were almost identical to those described in detail in our recent microarray study on mPFC (Eisinger et al., <xref rid="B18" ref-type="bibr">2014</xref>). In brief, mice were nulliparous when obtained and half were housed with breeder males for a 2 week mating period, while the other half were housed with age-matched female littermates to minimize isolation-induced stress. After mating, all females were housed individually until parturition (postpartum day 0) so that all subjects had a similar social environment. We have previously used these groups to identify gene expression changes that correspond to collective experiences (pregnancy, parturition, and postpartum) that generate the maternal phenotype (Zhao et al., <xref rid="B95" ref-type="bibr">2012b</xref>; Eisinger et al., <xref rid="B20" ref-type="bibr">2013b</xref>; Driessen et al., <xref rid="B16" ref-type="bibr">2014a</xref>,<xref rid="B17" ref-type="bibr">b</xref>). <italic>Ad libitum</italic> breeder chow (Harlan) and water were provided along with precut nesting material. Polypropylene cages were changed weekly prior to parturition, after which cages were not changed again until dissection. On day 0, litters were culled, if necessary, to standardize litter size to 11. All subjects were kept on a 12:12 light:dark cycle with lights on at 6:00 CST. All procedures followed guidelines set by the National Institutes of Health Guide for the Care and use of Laboratory Animals, and were approved by the University of Wisconsin Animal Care and Use Committee.</p></sec><sec id="s2-2"><title>Tissue collection and RNA extraction</title><p>Virgin and postpartum females were lightly anesthetized with isuflurane and decapitated between 9:00 and 12:00 CST on postpartum Day 7. Brains from age-matched virgin and postpartum females were collected on the same day and dissections were alternated between groups. After decapitation, vaginal lavage allowed for determination of estrous state. To minimize effects of estrous cycling on gene expression (Romano et al., <xref rid="B71" ref-type="bibr">1988</xref>; Arosh et al., <xref rid="B1" ref-type="bibr">2002</xref>), only diestrous virgins were used in the microarray experiment. Brains were frozen in isopentane, stored at −80°C, sectioned via cryostat (Leica CM1850, Bannockburn, IL, USA) at 200 micrometer thickness, and NAC collected via a micropunch technique (Makino et al., <xref rid="B48" ref-type="bibr">1994</xref>) using a Brain Punch Set (Stoelting, Wood Dale, IL, USA) under a dissecting microscope. Nucleus accumbens tissue was collected from Bregma 1.54 mm to Bregma 1.045 mm as shown in Figure <xref ref-type="fig" rid="F1">1</xref> and included both core and shell regions of NAC. Samples were collected from 10 postpartum females and 10 virgin females, and were subsequently stored at −80°C until RNA extraction. RNA extraction and purification was exactly as recently described (Eisinger et al., <xref rid="B18" ref-type="bibr">2014</xref>) and involved the Aurum Total RNA Fatty and Fibrous Tissue Kit (Bio-Rad, Hercules, CA, USA) and the NanoDrop 2000 spectrophotometer (Thermo Scientific, Wilmington, DE, USA) and RNA was stored at −80°C until further processing. For the microarray studies, six mice from each group were randomly selected for analysis as previous microarray studies indicate six per group is sufficient to detect differences in treatment. An N of 10 per group was used for follow up quantitative PCR (qPCR) analysis.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Representative section with NAC dissection for microarray analysis</bold>. Distance from Bregma in the rostrocaudal plane is indicated. Modified from the Allen Mouse Brain Atlas (reference atlas version 1, 2008). Abbreviations: aco, anterior commissure; NAC, nucleus accumbens.</p></caption><graphic xlink:href="fnbeh-08-00388-g0001"/></fig></sec><sec id="s2-3"><title>High-density oligonucleotide array hybridization</title><p>Microarray analysis was performed with the GeneChip Mouse Gene 2.0 ST Array (Affymetrix, Santa Clara, CA, USA) with targets derived from total RNA from NAC. Approaches were identical to those recently described (Eisinger et al., <xref rid="B18" ref-type="bibr">2014</xref>) and included the Ambion GeneChip WT Expression Kit (Ambion, Austin, TX, USA), the Affymetrix WT Terminal Labeling Kit (Affymetrix), and an Affymetrix GC3000 G7 Scanner. Data were extracted and processed in the Affymetrix Command Console v. 3.1.1.1.229 and cDNA synthesis, fragmentation, labeling, array hybridization, staining, and scanning were performed by the Gene Expression Center at the University of Wisconsin-Madison as in previous studies (Eisinger et al., <xref rid="B20" ref-type="bibr">2013b</xref>, <xref rid="B18" ref-type="bibr">2014</xref>; Driessen et al., <xref rid="B16" ref-type="bibr">2014a</xref>).</p></sec><sec id="s2-4"><title>Probeset level summarization and microarray statistical analysis</title><p>Probe logarithmic intensity error (PLIER) algorithm in Affymetrix Expression Console, build 1.2.1.20 was used for probeset level summarization and normalization. The BioConductor package limma v3.14.4 was used to perform statistical analysis. The nominal PLIER <italic>p</italic>-value of 0.01 was used for analysis as in previous studies (Saul et al., <xref rid="B72" ref-type="bibr">2012</xref>; Eisinger et al., <xref rid="B20" ref-type="bibr">2013b</xref>, <xref rid="B18" ref-type="bibr">2014</xref>; Driessen et al., <xref rid="B16" ref-type="bibr">2014a</xref>) and a <italic>p</italic>-value of 0.03 was also used for additional data mining. Fold change was calculated for each gene as the ratio of the limma-calculated average maternal expression coefficient divided by average virgin expression coefficient.</p></sec><sec id="s2-5"><title>Modular Single-set Enrichment Test (MSET)</title><p>MSET was used to test significant microarray results for enrichment of gene lists with multiple databases. To evaluate links to addiction, we identified five independent online gene-disease association databases, including the HuGE Navigator’s Phenopedia (Yu et al., <xref rid="B91" ref-type="bibr">2010</xref>), the DISEASES database by the Novo Nordisk Foundation Center for Protein Research at the University of Copenhagen (Pletscher-Frankild et al., <xref rid="B83" ref-type="bibr">2014</xref>), the NIH’s Genetic Association Database (GAD; Becker et al., <xref rid="B4" ref-type="bibr">2004</xref>), the Weizmann Institute of Science’s MalaCards compendium (Rappaport et al., <xref rid="B69" ref-type="bibr">2013</xref>), and Gemma’s Phenocarta (Zoubarev et al., <xref rid="B96" ref-type="bibr">2012</xref>). Addiction research often focuses on drugs of abuse and within each database multiple categories are provided, including general addiction or chemical dependency or genes known to be linked to different types of addiction, including cocaine, methamphetamine, heroin, nicotine, and alcohol. Within each database there are some overlaps of the different subsets. For Phenocarta, we kept together genes linked to all types of addiction (total gene number is 359), but for the other four we separately analyzed alcohol because these lists were often larger than all other lists combined (range from 337 to 675). Further, for GAD, the nicotine list alone was 2945 genes, so this was also analyzed separately. The level of overlap between the different addiction databases is shown in Figure <xref ref-type="fig" rid="F2">2</xref>. As an additional analysis, we created a database that included all unique genes from each of the five addiction databases (<italic>n</italic> = 986) and created an additional database for genes that showed up in two or more lists (<italic>n</italic> = 304). We also used MSET to test for enrichment against various mental health disorders and diseases as we recently performed using gene expression results from other maternal brain regions (Eisinger et al., <xref rid="B19" ref-type="bibr">2013a</xref>, <xref rid="B18" ref-type="bibr">2014</xref>; Driessen et al., <xref rid="B16" ref-type="bibr">2014a</xref>).</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Modular Single-set Enrichment Test evaluation of enrichment for addiction related gene sets within significantly altered genes in maternal NAC. (A)</bold> Y-axis represents the probability of X matches to database appearing in a randomly generated set of simulated results from the microarray background. The red arrow shows how many matches were found in the actual significant postpartum NAC expression changes and where that number falls on the probability density distribution. The enrichment <italic>p</italic>-value is derived from the number of simulated results that contained at least as many matches to database as the actual results. The number of genes within each database and the overlap (smaller sized list divided by larger list) is provided. <bold>(B)</bold> ToppCluster results showing enrichment of significant postpartum NAC genes with genes associated with three drugs, ketamine, nicotine, and dronabinol. Graph on right shows the general pattern whereby some genes are only linked to one drug, but others are linked to two or three.</p></caption><graphic xlink:href="fnbeh-08-00388-g0002"/></fig><p>Modular Single-set Enrichment Test uses a randomization testing algorithm to assess overrepresentation of any database (e.g., any mental health disorder, any disease) within significant microarray results. It generates a null distribution of expected number of matches to a given database of interest in randomly generated results built by sampling without replacement from the microarray background and this has been described in detail (Eisinger et al., <xref rid="B19" ref-type="bibr">2013a</xref>). In this study, we tested enrichment within microarray results with <italic>p</italic>-values less than 0.01 using 10,000 randomized sets of results. Current and previous evaluations indicate that almost all of the genes in this range can be confirmed via qPCR and are biologically relevant (Saul et al., <xref rid="B72" ref-type="bibr">2012</xref>; Zhao et al., <xref rid="B95" ref-type="bibr">2012b</xref>; Eisinger et al., <xref rid="B20" ref-type="bibr">2013b</xref>, <xref rid="B18" ref-type="bibr">2014</xref>). In this study we also confirmed a number of genes with <italic>p</italic>-values less than 0.03, but greater than the <italic>p</italic> < 0.01 cutoff and for additional data mining we ran MSET using a <italic>p-value</italic> cutoff of 0.03. As with any array, there can be a small number of false positive and in the present study, for example, we could not confirm <italic>Grm7</italic> changes. MSET works with larger datasets and in this study the removal of one or two genes (from 1052 annotated genes in the microarray) when compared with databases with hundreds of genes had little contribution to the overall <italic>p</italic>-value. Therefore, we report here the <italic>p</italic>-value using all 1052 genes, but remove any unconfirmed gene from further analysis.</p></sec><sec id="s2-6"><title>Analysis with ToppCluster and NIH DAVID</title><p>Microarray targets with <italic>p</italic>-values less than 0.01 were used for both ToppCluster (Kaimal et al., <xref rid="B31" ref-type="bibr">2010</xref>) and NIH DAVID analysis (Huang da et al., <xref rid="B28" ref-type="bibr">2009</xref>). The default statistical <italic>p</italic>-value analysis was used for both. While some databases are shared between the two, each also provides unique analysis. For example, ToppCluster allows for analysis of enrichment against genes associated with various drug actions, and this analysis was used in this study.</p></sec><sec id="s2-7"><title>WGCNA and transcription factor analysis</title><p>Weighted Gene Coexpression Network Analysis was used to identify modules of genes whose expression changes are highly correlated to one another. R software was used for all WGCNA analysis (Zhang and Horvath, <xref rid="B93" ref-type="bibr">2005</xref>; Langfelder and Horvath, <xref rid="B39" ref-type="bibr">2008</xref>) and the array <italic>p</italic>-value cutoff was <0.01. To generate a weighted network of genes (nodes) and their expression correlations (edges), correlations were raised to a soft thresholding power β, chosen such that the network approximates a model of scale-free topology (<italic>R</italic><sup>2</sup> > 0.8), which is a necessary assumption for WGCNA accuracy. Using unsupervised hierarchical clustering, a minimum module size of 30 genes, and a threshold setting for merging modules of 0.25, WGCNA identified two modules. The modules were exported as a Cytoscape network file, which was manually trimmed to consist only of transcription factor nodes and their gene-to-gene correlations. The finalized transcription factor module was visualized with Cytoscape v3.0.1.</p></sec><sec id="s2-8"><title>Quantitative real-time PCR</title><p>Polymerase chain reaction was performed on genes of interest (<italic>n</italic> = 10 per group) in order to confirm expression changes detected by microarray analysis. Two stable reference genes were used to normalize relative expression results of genes of interest; Tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein, zeta polypeptide (<italic>Ywhaz</italic>), and peptidylprolyl isomerase A (<italic>Ppia</italic>) (also known as CycA). Control genes were chosen because they have been shown by others to act as reliable control genes for qPCR in the rodent brain (Bonefeld et al., <xref rid="B6" ref-type="bibr">2008</xref>; Nelissen et al., <xref rid="B56" ref-type="bibr">2010</xref>) and have been found by us to be reliable markers for the postpartum brain (Zhao et al., <xref rid="B94" ref-type="bibr">2012a</xref>,<xref rid="B95" ref-type="bibr">b</xref>; Eisinger et al., <xref rid="B20" ref-type="bibr">2013b</xref>, <xref rid="B18" ref-type="bibr">2014</xref>; Driessen et al., <xref rid="B17" ref-type="bibr">2014b</xref>).</p><p>A SuperScript III First-Strand Synthesis System for RT-PCR (Invitrogen, Carlsbad, CA, USA) was used to reverse transcribe 100 ng of RNA to cDNA in an Eppendorf MasterCycler Personal PCR Machine (Eppendorf, Hamburg, Germany) with poly-T 20mer primers. The thermal profile used is as follows: an initial melting step of 95°C for 30 s, followed by 40 cycles of a 5-s 95°C melt, a 20-s 58°C annealing step, and a 20-s 72°C elongation step. A melt curve was performed from 60–95°C at 5-s 0.5°C increments to confirm specificity of primer binding, and relative expression values were calculated with REST 2009 (Pfaffl et al., <xref rid="B66" ref-type="bibr">2002</xref>).</p></sec></sec><sec sec-type="results" id="s3"><title>Results</title><sec id="s3-1"><title>Genes with altered expression in maternal NAC</title><p>High density oligonucleotide microarray (41,346 probes) was performed on NAC and results were analyzed with the PLIER algorithm. Altered gene expression was found in postpartum NAC for 1260 probes (1052 unique, annotated genes) using a <italic>p</italic> < 0.01 cutoff and for 2740 probes (2164 unique, annotated genes) using a <italic>p</italic> < 0.03 cutoff. The full set of results is available in Supplementary Table 1. All microarray results, including CEL files, have been uploaded to NCBI’s Gene Expression Omnibus (Accession number: GSE62258).</p></sec><sec id="s3-2"><title>MSET and ToppCluster analysis for enrichment of gene sets associated with addiction</title><p>Modular Single-set Enrichment Test was used to assess enrichment for genes associated with addiction within genes displaying altered expression (<italic>p</italic> < 0.01) in maternal NAC. Details on the databases are provided in the Methods section. As shown in Figure <xref ref-type="fig" rid="F2">2A</xref>, the postpartum NAC showed enrichment for addiction and reward related genes in five of five independent datasets (<italic>p</italic> < 0.05). Information on the number of genes in each list and the extent of overlap between lists is provided in Figure <xref ref-type="fig" rid="F2">2A</xref>. Furthermore, enrichment was found when either all addiction related genes were pooled to create a novel database (<italic>N</italic> = 986; <italic>p</italic> < 0.01) or when only genes in two or more lists were pooled to create a novel database (<italic>N</italic> = 304; <italic>p</italic> < 0.01). Notable genes found in more than one database were: <italic>Per1</italic>, <italic>Per2</italic>, <italic>Penk</italic>, <italic>Homer2</italic>, <italic>Creb1</italic>, <italic>Ntrk2</italic>, <italic>Nr4a2</italic>, and <italic>Fosb</italic>. Further, we ran MSET using an array <italic>p</italic> < 0.03 cutoff (see rationale above) and also found significant enrichment for all addiction/reward datasets (see Supplementary Table 2). Among the interesting genes with <italic>p</italic>-values between 0.01 and 0.03 were: <italic>Cartpt</italic>, <italic>Adcy1</italic>, <italic>Npy1r</italic>, <italic>Htr1a</italic>, <italic>Drd1a</italic>, <italic>Gria1</italic>, and <italic>Pdyn</italic>. Supplementary Table 2 provides a list of all genes (<italic>N</italic> = 113) with <italic>p</italic>-values less than 0.03 that were associated with one or more addiction database. When analyzed with ToppCluster, significant postpartum NAC genes showed a significant overlap with genes involved in the action of drugs of abuse, including nicotine, ketamine, or dronabinol (a synthetic cannabinoid analog; <italic>p</italic> < 0.001 each; Figure <xref ref-type="fig" rid="F2">2B</xref>). A GeneMania diagram of interactions of a subset of addiction/reward related genes in postpartum NAC is shown in Figure <xref ref-type="fig" rid="F3">3</xref>. Although <italic>Nr1d1</italic> was not in any of the five addiction lists, it was included here because it has recently been implicated in addiction processes (see Section Discussion below), its expression is significantly downregulated in NAC (see below), and it is centrally connected to addiction/reward related genes.</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Interaction network of addiction related genes with altered expression in postpartum NAC</bold>. The red circles indicate upregulation and blue circles indicate downregulation of genes within significant maternal NAC expression results. Results were visualized as an interaction network with genes in gray circles added by GeneMania due to interactions with specified genes. <italic>Nr1d1</italic> was added to network due to recent studies linking it to addiction pathways and because of its multiple interactions with known addiction related genes. The nature of the interaction data linking any two genes is encoded by color (blue lines = colocalization, purple lines = coexpression, red lines = physical interactions, light green = shared protein domains, and orange lines = predicted interaction). Distance between genes is proportional to the strength of evidence for their interaction.</p></caption><graphic xlink:href="fnbeh-08-00388-g0003"/></fig><p>Because alcohol addiction related genes had large datasets, these were analyzed independently, but were commonly just above significance (<italic>p</italic> = 0.06, 0.09, 0.06, and 0.34 for the four available databases; see Supplementary Table 3). Also, the nicotine related database from GAD was large (<italic>N</italic> = 2945), so this was analyzed independently and was significantly linked to maternal NAC expression (<italic>p</italic> < 0.001; see Supplementary Table 3).</p></sec><sec id="s3-3"><title>MSET analysis for enrichment of gene sets associated with mental health disorders and other diseases</title><p>Figure <xref ref-type="fig" rid="F4">4</xref> shows MSET analysis for genes linked to depression, bipolar disorder (BPD), autism, schizophrenia, arthritis, Alzheimer’s and multiple sclerosis. Details on the databases are provided in the Methods section. Postpartum NAC array genes (<italic>p</italic> < 0.01) showed consistent significant enrichment for depression (two of three databases), BPD (three of three databases), and schizophrenia (five of five databases). For autism, only three of seven databases showed significant enrichment. A list of postpartum NAC genes with links to mental health disorders is provided in Supplementary Table 4. For arthritis, Alzheimer’s disease, and multiple sclerosis, no enrichment was found (Figure <xref ref-type="fig" rid="F4">4</xref>).</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>Modular Single-set Enrichment Test evaluation of enrichment for mental health related gene sets within significantly altered genes in maternal NAC. (A)</bold> Y-axis represents the probability of X matches to depression and BPD databases appearing in a randomly generated set of simulated results from the microarray background. The red arrow shows how many matches were found in the actual significant postpartum NAC expression changes and where that number falls on the probability density distribution. The enrichment <italic>p</italic>-value is derived from the number of simulated results that contained at least as many matches to database as the actual results. <bold>(B)</bold> Modular Single-set Enrichment Test <italic>p</italic>-values for significant postpartum NAC genes against databases for autism, schizophrenia, arthritis, Alzheimer’s disease, and multiple sclerosis.</p></caption><graphic xlink:href="fnbeh-08-00388-g0004"/></fig></sec><sec id="s3-4"><title>Pathway analysis with NIH DAVID, ToppCluster, and MSET</title><p>NIH DAVID and ToppCluster identified enrichment for RNA binding and NIH DAVID further found enrichment for genes involved with CNS development/differentiation, transcriptional regulation, and circadian rhythms. A list of 41 postpartum NAC genes with connections to various aspects of CNS development, such as axon growth, axonogenesis, neurogenesis, and neuron differentiation is provided in Supplementary Table 5. Using MSET and a database of known transcriptional regulators in mice, the Animal Transcription Factor Database, we found significant enrichment of postpartum NAC for transcriptional regulators (<italic>p</italic> < 0.00001). A total of 128 genes that contribute to transcriptional regulation were found within significant postpartum array results (~12% of all genes). A list of these genes along with additional transcriptional regulation genes from NIH DAVID is provided in Supplementary Table 5. NIH DAVID analysis also revealed enrichment for acetylation (172 genes), histone deacetylation complex (genes included: <italic>Hdac4</italic>, <italic>Hdac5</italic>, <italic>Nrip1</italic>, <italic>Morf4l1</italic>, <italic>Sap30</italic>, and <italic>Suds3</italic>) and histone acetyltransferase complex (genes included: <italic>Ep300</italic>, <italic>Taf12</italic>, <italic>Yeats4</italic>, <italic>Eny2</italic>, and <italic>Morf4l1</italic>) (see Supplementary Table 5). KEGG pathway analysis via NIH DAVID identified enrichment for circadian rhythm and the genes identified were: <italic>Csnk1e</italic>, <italic>Cry1</italic>, <italic>Npas2</italic>, <italic>Nr1d1</italic>, <italic>Per1</italic>, and <italic>Per2</italic>.</p></sec><sec id="s3-5"><title>qPCR analysis of a subset of genes</title><p>Genes of interest for real-time qPCR confirmation were selected based on biological importance and concordance with recent studies in other brain regions. We confirmed array expression changes (<italic>p</italic> < 0.01) via qPCR (all <italic>p</italic> < 0.05) for <italic>Grm3</italic>, <italic>Flt1</italic>, <italic>Penk</italic>, <italic>Nr1d1</italic>, <italic>Uhrf2</italic>, and <italic>Per2</italic> and PCR direction was always the same as array direction (Figure <xref ref-type="fig" rid="F5">5</xref>). The array decreases in <italic>Fabp7</italic> in NAC were confirmed in a separate study (unpublished observations). We did not confirm <italic>Grm7</italic>, <italic>Oprk1</italic>, or <italic>Dscam</italic> (<italic>p</italic> > 0.05). We also confirmed via qPCR (<italic>p</italic> < 0.05) expression changes for genes with array <italic>p</italic>-values between 0.01 and 0.03 that have strong connections to addiction and reward, including <italic>Pdyn</italic> (array <italic>p</italic> = 0.026) and <italic>Drd1a</italic> (array <italic>p</italic> = 0.016; Figure <xref ref-type="fig" rid="F5">5</xref>). Further, elevated <italic>Oxtr</italic> (array <italic>p</italic> = 0.07) and decreased <italic>Pde4b</italic> (array <italic>p</italic> = 0.043) were confirmed as significant in postpartum NAC increases using qPCR (<italic>p</italic> < 0.05) (not shown). Primer information on tested genes can be viewed in Supplementary Table 6.</p><fig id="F5" position="float"><label>Figure 5</label><caption><p><bold>Quantitative real-time PCR confirmation of expression changes for genes of interest in maternal NAC compared to virgin</bold>. Relative expression distribution (Y-axis) represented as a ratio of postpartum vs. virgin (<italic>n</italic> = 10 per group) normalized against two references genes, <italic>Ppia</italic> and <italic>Ywhaz</italic>, and shown by box-and-whisker plots as medians (black dashed lines), interquartile ranges (boxes), and ranges (whiskers). Ratios over 1 indicate genes that are more highly expressed in postpartum NAC than in virgin, while ratios less than 1 indicate genes with lower expression in postpartum females. <italic>p</italic> < 0.05; **<italic>p</italic> < 0.001.</p></caption><graphic xlink:href="fnbeh-08-00388-g0005"/></fig></sec><sec id="s3-6"><title>WGCNA analysis</title><p>We additionally used WGCNA to identify modules of genes that cluster together based on co-expression. This approach can provide indirect insight into transcriptional regulatory networks. To investigate a possible role for transcription factors in the coordinated expression of this gene network, expression correlations for transcription factors found within two modules were visualized in Cytoscape. As seen in Figure <xref ref-type="fig" rid="F6">6</xref>, <italic>Nr1d1</italic> (in blue module) may have an important role in transcriptional regulation of some postpartum genes, including <italic>Penk</italic> and <italic>Grm3</italic>. Other circadian transcriptional regulators, including <italic>Cry1</italic>, <italic>Per1</italic>, and <italic>Per2</italic> (turquoise module), and genes they may regulate in the postpartum state are also shown in Figure <xref ref-type="fig" rid="F6">6</xref>. A subset of transcriptional regulators, including <italic>Fosb</italic>, <italic>Jun</italic>, <italic>Egr1</italic>, <italic>Ar</italic>, <italic>Dmtf1</italic>, <italic>Nr4a3</italic>, <italic>Nr4a1</italic>, <italic>Ncoa7</italic>, <italic>Egr2</italic>, <italic>Hdac5</italic>, and <italic>Irf2</italic> are themselves clustered together and are indicated to act as a network to regulate ~100 postpartum genes (Figure <xref ref-type="fig" rid="F6">6</xref>).</p><fig id="F6" position="float"><label>Figure 6</label><caption><p><bold>Weighted Gene Coexpression Network Analysis network of transcription factors and genes with highly correlated expression changes in postpartum NAC</bold>. Within the gene module identified by WGCNA as having highly correlated expression, transcription factors (light blue or blue circles with black labels) are visualized in relation to other module genes. The significance of the correlation is indicated by the width of the line, with thicker lines reflecting more significant correlation.</p></caption><graphic xlink:href="fnbeh-08-00388-g0006"/></fig></sec></sec><sec sec-type="discussion" id="s4"><title>Discussion</title><p>In this study we find evidence that the large-scale gene expression changes in NAC in postpartum females include numerous genes linked to addiction and reward. The finding is consistent with NAC’s role in addiction or reward/incentive related behaviors and that offspring are highly rewarding to mothers (see Section Introduction). The novelty of this study is the identification of over a hundred addiction/reward-related genes that are recruited and likely promote the emergence of a new natural reward, the offspring. The finding of multiple postpartum NAC genes that are involved in CNS plasticity and development is consistent with findings in other maternal brain regions and suggests the maternal brain may be a developmental endpoint. That a number of genes with altered expression are involved in transcription regulation provides a means for understanding how the maternal NAC is produced as does the WGCNA analysis that locates coexpression modules. The links of maternal NAC expression to some mental health disorders is similar to recent findings in other regions and has multiple implications as discussed below.</p><sec id="s4-1"><title>Genes with altered expression in postpartum NAC linked to addiction and reward</title><p>One of the most striking findings of the present study was the high enrichment of significant postpartum NAC genes with multiple databases with genes linked to addiction, dependency, and reward. Further when all genes from all the databases were brought together or genes that appeared in two or more lists were brought together to make novel databases, the significant enrichment was also found using MSET (Figure <xref ref-type="fig" rid="F2">2A</xref>). Although <italic>p</italic> < 0.01 is a typical array cutoff, we confirmed via qPCR genes with <italic>p</italic>-values less than 0.03, but greater than 0.01, including <italic>Drd1a</italic> and <italic>Pdyn</italic>. We also confirmed decreased <italic>Pde4b</italic> and elevated <italic>Oxtr</italic> via qPCR although the respective array <italic>p</italic>-values were 0.043 and 0.07. <italic>Oxtr</italic> is of interest because of previous work showing a positive association of the receptor in NAC to maternal care (Olazabal and Young, <xref rid="B60" ref-type="bibr">2006</xref>). Together, these findings and those of other arrays (Saul et al., <xref rid="B72" ref-type="bibr">2012</xref>; Zhao et al., <xref rid="B95" ref-type="bibr">2012b</xref>) suggested that a number of genes with <italic>p</italic>-values above 0.01 have biologically meaningful expression changes. To provide a broader survey of large scale changes we also tested for enrichment using MSET for genes with array <italic>p</italic> < 0.03. Again, significant enrichment of postpartum NAC genes with databases for addiction was found and a list of those genes is provided in Supplemental Table 2. The addiction/reward databases used in this study are independently curated and it is possible that the evidence for some of the genes is weak and they may prove to have little contribution to reward processes. However, it is also possible that some lesser known genes are included in these lists that do play an important role in addiction/reward processes. If we and others continue to find associations with such genes, it is possible that these previously unnoticed, but important genes, will now receive attention.</p><p>The idea of the postpartum NAC being in a new natural reward state was supported by ToppCluster analysis that found significant links of postpartum NAC genes with genes associated with either nicotine, ketamine, or dronabinol (a synthetic cannabinoid analog; Figure <xref ref-type="fig" rid="F2">2B</xref>). Modular Single-set Enrichment Test analysis of the one nicotine (GAD) database was significant, but none of the four databases for alcohol were significant. Although it is speculative, aspects of nicotine addiction and dependency appear to have greater similarities to the maternal NAC than does alcohol addiction.</p><p>Using an array <italic>p</italic>-value cutoff of 0.03, we identified over a hundred genes that are related to reward/incentive/addiction related processes. One approach to understand this multitude of genes is to evaluate each gene independently and to determine how they interact with one another. Some genes have been well studied in terms of addiction and/or maternal function. For example <italic>Cartpt</italic> (also known as Cart) derives its name from its response to cocaine and amphetamine and is implicated in reward processes (Hurd et al., <xref rid="B29" ref-type="bibr">1999</xref>; Rogge et al., <xref rid="B70" ref-type="bibr">2008</xref>). Further, Cart expression in NAC is modulated by pup cues in rat mothers (Mattson and Morrell, <xref rid="B50" ref-type="bibr">2005</xref>). <italic>Creb1</italic> (also known as Creb) and variants of <italic>Fosb</italic>, including delta Fosb, are involved in addictive responses (McClung and Nestler, <xref rid="B53" ref-type="bibr">2003</xref>) and maternal care is disrupted in <italic>Fosb</italic> mutant mice (Brown et al., <xref rid="B10" ref-type="bibr">1996</xref>). <italic>Drd1a</italic> (also known as Drd1) is involved in addictive responses (Comings et al., <xref rid="B14" ref-type="bibr">1997</xref>; Le Foll et al., <xref rid="B40" ref-type="bibr">2009</xref>) and maternal care in rats (Parada et al., <xref rid="B63" ref-type="bibr">2008</xref>) and humans (Mileva-Seitz et al., <xref rid="B54" ref-type="bibr">2012</xref>). Genemania provides gene connections based on numerous datasets, including coexpression and physical interactions. As shown in Figure <xref ref-type="fig" rid="F3">3</xref>, some interesting links occur between these genes and Genemania identifies additional genes (in gray) that may act as mediators between the focus genes. <italic>Nr1d1</italic> (also known as rev erb alpha) was not in any of the five addiction databases, but it is significantly downregulated in maternal NAC and recent studies are beginning to suggest an important role for this gene in addiction and reward related behaviors (Belluardo et al., <xref rid="B5" ref-type="bibr">2005</xref>; Wang et al., <xref rid="B86" ref-type="bibr">2008</xref>; Piechota et al., <xref rid="B67" ref-type="bibr">2012</xref>; Wongchitrat et al., <xref rid="B88" ref-type="bibr">2013</xref>). <italic>Nr1d1</italic> is a transcriptional factor integral to function of circadian rhythm genes and to mental health disorders (see below). As seen in Figure <xref ref-type="fig" rid="F3">3</xref>, <italic>Nr1d1</italic> is tightly linked to other addiction related gene and as highlighted by WGCNA analysis, it may be a critical transcriptional regulator in the maternal brain. It is possible that <italic>Nr1d1</italic> may provide a newer avenue for understanding the strong link between mental health disorders and the likelihood of addiction, but this would need to be addressed in subsequent studies.</p><p>Nucleus accumbens is a central region in reward related behaviors, but it is part of a network and it will be valuable to understanding of how modifications of NAC are coordinated with other regions. Further, NAC includes both core and shell regions (Kelley and Berridge, <xref rid="B32" ref-type="bibr">2002</xref>) and a detailed understanding of how gene expression changes in NAC are manifested in terms of subregions will be important for future work on natural shifts in reward. Further, how genes are expressed in given cell types is of great importance. For example, while <italic>Fabp7</italic> is found mostly in glial cells along with neural progenitors (Matsumata et al., <xref rid="B49" ref-type="bibr">2012</xref>; Yun et al., <xref rid="B92" ref-type="bibr">2012</xref>),<italic> Penk</italic> is clearly neuron-specific. A detailed understanding of gene expression changes within cell subtype will also be key to moving our understanding forward.</p><p>In this study we avoided any maternal testing as tests themselves can alter gene expression. All pups were healthy at the time of tissue collection, indirectly indicating that maternal care, including nursing, was sufficient. However, variance in levels of maternal care has been documented in rats and mice (Champagne et al., <xref rid="B13" ref-type="bibr">2003</xref>, <xref rid="B12" ref-type="bibr">2007</xref>) and even with seemingly similar maternal profiles, the reward responding can differ significantly among mothers at day 10 in rats (Mattson et al., <xref rid="B52" ref-type="bibr">2003</xref>). Reward responding to pups is high on day 8 in rats and shifts away from offspring at day 10 and beyond as they mature (Mattson et al., <xref rid="B51" ref-type="bibr">2001</xref>). In this study we evaluated female mice on postpartum day 7 when it would be expected that reward responding to pups was still high. Because we did not test for reward responding in this study, we cannot draw any direct conclusions, but suggest that the composite brain changes we observed in postpartum NAC correspond to the composite changes that occur in reward responding at this time (even though individual differences may occur). It would be of interest in future work to evaluate how individual differences in maternal reward responding may relate to individual differences in some of the highlighted genes.</p><p>The magnitude of gene expression change observed in this study is consistent with those in other postpartum brain studies (Gammie et al., <xref rid="B24" ref-type="bibr">2005</xref>; Xiao et al., <xref rid="B89" ref-type="bibr">2005</xref>; Zhao et al., <xref rid="B95" ref-type="bibr">2012b</xref>; Eisinger et al., <xref rid="B20" ref-type="bibr">2013b</xref>, <xref rid="B18" ref-type="bibr">2014</xref>; Driessen et al., <xref rid="B16" ref-type="bibr">2014a</xref>), suggesting the findings are real and biologically significant. Further, in the previous array studies and here, the magnitude of change assessed using qPCR matches that of the array, suggesting a significant and real biological level of change. In recent work, we found protein level changes also to be significant and to match those of mRNA analysis for glutamic acid decarboxylase in lateral septum (LS; Zhao et al., <xref rid="B94" ref-type="bibr">2012a</xref>) and for <italic>Nr1d1</italic> and <italic>Fabp7</italic> in multiple brain regions (unpublished observations). Although the changes may not appear as dramatic as those found using cell cultures, they are similar to a wide range of other studies on manipulations (e.g., knockout, drug treatment, exercise, or aging) in the CNS (Piechota et al., <xref rid="B68" ref-type="bibr">2010</xref>; Kohman et al., <xref rid="B35" ref-type="bibr">2011</xref>; Kõks et al., <xref rid="B36" ref-type="bibr">2011</xref>). Thus, the 21% decrease in Nr1d1, a transcription factor, would be expected to have an important effect. Given the fine-tuned nature of the CNS, even a 10% change of a gene could be expected to have a biologically significant effect, but the effect of any change would need to be tested directly.</p></sec><sec id="s4-2"><title>Postpartum NAC genes involved in transcriptional regulation</title><p>One notable feature of postpartum NAC genes was the large number involved in transcriptional regulation. This was seen with both MSET analysis using a transcriptional database and with NIH DAVID pathway analysis. One-hundered and fifty-three genes with an array <italic>p</italic> < 0.01 are involved in transcriptional regulation (Supplementary Table 5). WGCNA analysis provides indirect information on which of these transcriptional regulators may be contributing most to altered expression of other genes. Interestingly, <italic>Nr1d1</italic> (Figure <xref ref-type="fig" rid="F6">6</xref>) was identified as a possible key regulator of other maternal genes such as <italic>Penk</italic> and <italic>Grm3</italic>. Other circadian transcriptional regulators, including <italic>Cry1</italic>, <italic>Per1</italic>, and <italic>Per2</italic> were part of a different module, but also suggested to regulate some postpartum genes. As also seen in Figure <xref ref-type="fig" rid="F6">6</xref>, a large number of postpartum genes were suggested to be regulated by a small subset of transcriptional regulators, including <italic>Fosb</italic>, <italic>Jun</italic>, <italic>Egr1</italic>, <italic>Ar</italic>, <italic>Dmtf1</italic>, <italic>Nr4a3</italic>, <italic>Nr4a1</italic>, <italic>Ncoa7</italic>, <italic>Egr2</italic>, <italic>Hdac5</italic>, and <italic>Irf2</italic>. Weighted Gene Coexpression Network Analysis provides a starting point for investigating how natural changes in NAC occur. Whether or to what extent the transcriptional modules produced by WCGNA reflect how the postpartum NAC is produced can be evaluated in subsequent studies.</p></sec><sec id="s4-3"><title>Genes with altered expression in postpartum NAC enriched for genes linked to mental health disorders</title><p>Postpartum NAC gene expression showed particularly high enrichment for BPD (three of three databases), schizophrenia (five of five databases), and depression (two of three databases). Autism showed modest enrichment (three of seven databases). Together, this profile is similar to that recently found for postpartum mPFC, whereby schizophrenia and BPD showed the strongest enrichment (Eisinger et al., <xref rid="B18" ref-type="bibr">2014</xref>). One explanation for such enrichment is that the same genes that are actively regulated to produce a maternal phenotype are also ones that can be dysregulated in a mental health disorder. As an example, it could be possible that hundreds of genes have altered expression to elevate sociability and bonding in a mother, but if those same genes are dysregulated, then decreases in sociability or bonding could occur. For most mental health disorders, deficits in sociability are an endophenotype. Another view is that hundreds of genes are needed for the complex maternal phenotype, but if any of those changes go amiss, then a disorder could occur. This outlook is consistent with the finding of increased risk in mothers for depression, BPD, and postpartum psychosis (with connections to schizophrenia) (Brockington, <xref rid="B9" ref-type="bibr">2004</xref>; Sit et al., <xref rid="B76" ref-type="bibr">2006</xref>; Spinelli, <xref rid="B79" ref-type="bibr">2009</xref>; Maina et al., <xref rid="B47" ref-type="bibr">2014</xref>). Among genes of interest with high links to mental disorders are <italic>Grm3</italic>, <italic>Fabp7</italic>, and <italic>Nr1d1</italic>. Recent work suggests <italic>Nr1d1</italic> regulation of <italic>Fabp7</italic> expression (Schnell et al., <xref rid="B74" ref-type="bibr">2014b</xref>) and as indicated above, a role for <italic>Nr1d1</italic> in addictive and reward related processes are now being established. Whether <italic>Nr1d1</italic> could play an integral role in the high rates of addiction in those with mental health disorders is an interesting, but untested hypothesis. How to interpret or understand the links of the maternal brain genes with those for mental health disorders will take time to determine, but knowledge of shared genes can help inform both our view of the maternal brain, but also of how mental health disorders occur.</p></sec><sec id="s4-4"><title>Enrichment of developmental processes within postpartum NAC</title><p>A number of postpartum NAC genes were found to be linked to CNS plasticity and development via NIH DAVID. These processes included axon growth, axonogenesis, neurogenesis, and neuron differentiation. The finding of enrichment for CNS plasticity and development is similar to our recent postpartum microarray findings for LS, medial preoptic area (MPOA) and mPFC (Eisinger et al., <xref rid="B20" ref-type="bibr">2013b</xref>, <xref rid="B18" ref-type="bibr">2014</xref>; Driessen et al., <xref rid="B16" ref-type="bibr">2014a</xref>). These cumulative findings suggest the maternal brain represents a developmental endpoint and are consistent with previous work demonstrating CNS plasticity in lactating females (Gregg et al., <xref rid="B25" ref-type="bibr">2007</xref>; Kim et al., <xref rid="B33" ref-type="bibr">2010</xref>; Leuner and Gould, <xref rid="B43" ref-type="bibr">2010</xref>; Lévy et al., <xref rid="B44" ref-type="bibr">2011</xref>). Differentiation and plasticity can occur in the absence of neurogenesis or gliogenesis and it is possible that many of the postpartum CNS developmental events are occurring within intact cells as opposed to within newly generated cells. For example, decreases of <italic>Fabp7</italic> can lead to altered identity of cells expressing <italic>Fabp7</italic>, but also of neighboring cells (Owada, <xref rid="B61" ref-type="bibr">2008</xref>; Boneva et al., <xref rid="B7" ref-type="bibr">2011</xref>; Kipp et al., <xref rid="B34" ref-type="bibr">2011</xref>; Mitchell and Hatch, <xref rid="B55" ref-type="bibr">2011</xref>). The extent of CNS plasticity is likely quite large, but focused studies would be needed to help untangle how specific gene expression changes translate into function CNS changes.</p></sec><sec id="s4-5"><title>One model for the emergence of reward and addiction to offspring with a focus on opioids</title><p>As detailed in the Introduction section, a large body of work has highlighted the maternal state as involving and increases reward responding to offspring. Although we highlight numerous genes in NAC that may contribute to the maternal phenotype, some of these are particularly noteworthy. In postpartum NAC, both endogenous opioids, <italic>Pdyn</italic> and<italic> Penk</italic>, show highly significant decreases in expression. Further, these postpartum decreases of <italic>Penk</italic> and <italic>Pdyn</italic> are also found in mPFC and in LS (Eisinger et al., <xref rid="B20" ref-type="bibr">2013b</xref>, <xref rid="B18" ref-type="bibr">2014</xref>). Normally, the decreases of endogenous opioid signaling in NAC and associated regions would be associated with a decreased ability of reward responding (Kelley and Berridge, <xref rid="B32" ref-type="bibr">2002</xref>; Koob and Volkow, <xref rid="B37" ref-type="bibr">2010</xref>). Interestingly, MPOA has significant increase in <italic>Penk</italic> (Driessen et al., <xref rid="B16" ref-type="bibr">2014a</xref>) and MPOA is considered a critical maternal brain region. One model, then, is that in the transition to motherhood there is a general decrease in reward (via decreased endogenous opioids) for other rewarding natural stimuli. However, because MPOA enkephalin is elevated, there is now a heightened rewarding effect of any maternally related signal. A role for endogenous opioids in maternal care has previously been suggested (Panksepp et al., <xref rid="B62" ref-type="bibr">1994</xref>) and here a new insight may be that rewards from other signals are reduced in mothers, while rewards from offspring are enhanced. In earlier work we found elevated enkephalin in a large tissue region that included both hypothalamus and MPOA (Gammie et al., <xref rid="B24" ref-type="bibr">2005</xref>). Likewise, it is possible that increases in <italic>Penk</italic> from feeding associated regions, such as arcuate nucleus, has the similar effect so that in the maternal state, offspring and eating are the primary forms of reward type signaling. Although this is speculative, it does provide a testable model for how a natural switch in reward and addiction can occur with a focus on opioids.</p></sec><sec id="s4-6"><title>Circadian rhythm and acetylation genes</title><p>Circadian rhythms are intertwined with multiple processes, including addictions and mental health disorders, for recent reviews, see Damaggio and Gorman (<xref rid="B15" ref-type="bibr">2014</xref>), Landgraf et al. (<xref rid="B38" ref-type="bibr">2014</xref>) and Schnell et al. (<xref rid="B73" ref-type="bibr">2014a</xref>). We found enrichment for circadian rhythm genes and those included: <italic>Csnk1e</italic>, <italic>Cry1</italic>, <italic>Npas2</italic>, <italic>Nr1d1</italic>, <italic>Per1</italic>, and <italic>Per2</italic>. <italic>Nr1d1</italic> is of particular interest because it is a transcription factor that interacts with clock proteins (Ueda et al., <xref rid="B84" ref-type="bibr">2005</xref>). Also, <italic>Nr1d1</italic> has now been found to be reliably and significantly reduced in four maternal brain regions, NAC, mPFC, MPOA, and LS (Eisinger et al., <xref rid="B20" ref-type="bibr">2013b</xref>, <xref rid="B18" ref-type="bibr">2014</xref>; Driessen et al., <xref rid="B16" ref-type="bibr">2014a</xref>), suggesting its downregulation is critical for production of the maternal brain. We also found enrichment for genes related to acetylation events and some of these may have relevance to a recent finding for involvement of deacetylation in supporting gene expression and maternal care in mice (Stolzenberg et al., <xref rid="B81" ref-type="bibr">2014</xref>) and acetylation processes in gene expression and pair bonding in prairie voles (Wang et al., <xref rid="B85" ref-type="bibr">2013</xref>).</p></sec><sec id="s4-7"><title>Concluding remarks</title><p>In this study, we evaluate for the first time large scale gene expression changes that occur in NAC in mothers when pups have become highly rewarding. Using bioinformatics tools we find enrichment for genes involved in addiction and reward using multiple independently curated databases. A novelty of this study is the identification of over a hundred addiction/reward-related genes that are likely recruited and promote the emergence of a new natural reward, the offspring. Interestingly, in rodents, the rewarding properties of pups begins waning about half way through lactation (Seip and Morrell, <xref rid="B75" ref-type="bibr">2007</xref>), so the maternal rodent brain provides an example of how an animal can both increase and lessen a natural reward. Further, mothering decreases the rewarding properties of drugs (Mattson et al., <xref rid="B51" ref-type="bibr">2001</xref>, <xref rid="B52" ref-type="bibr">2003</xref>; Ferris et al., <xref rid="B23" ref-type="bibr">2005</xref>), so it is possible that insight from the maternal NAC could provide new insights into how to mitigate addictions.</p></sec></sec><sec id="s6"><title>Conflict of interest statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec>
Phytochemical, toxicological and antimicrobial evaluation of lawsonia inermis extracts against clinical isolates of pathogenic bacteria	<sec><title>Background</title><p>The emerging resistance of pathogen against the currently available antimicrobial agents demands the search of new antimicrobial agents. The use of medicinal plants as natural substitute is the paramount area of research to overwhelm the drug resistance of infectious agents. Scientists have not made enough effort on the evaluation of safety of medicinal plant yet.</p></sec><sec><title>Methods</title><p>In the present study antimicrobial activity of <italic>Lawsonia inermis</italic> is investigated against clinical isolates of seven bacteria including four Gram negative (<italic>Escherichia coli</italic>, <italic>Salmonella typhi</italic>, <italic>Klebsiella spp.</italic>, <italic>Shigella sonnei</italic>) and three Gram positive (<italic>Bacillus subtilis, Staphylococcus aureus</italic>, <italic>Staphylococcus epidermidis</italic>) using disc diffusion method. Four types of <italic>Lawsonia inermis</italic> extracts were prepared using methanol, chloroform, acetone and water as extraction solvents, while DMSO (Dimethyl sulfoxide) and water as dissolution solvents. The rate and extent of bacterial killing was estimated by time-kill kinetic assay at 1× MIC of each bacterial isolate. The overall safety of <italic>Lawsonia inermis</italic> extracts was assessed in mice.</p></sec><sec><title>Results</title><p><italic>Lawsonia inermis</italic> displayed noteworthy antimicrobial activity against both gram positive and gram negative bacterial strains used in the study. The minimum value of MIC for different bacterial strains ranged from 2.31 mg/ml to 9.27 mg/ml. At 1x MIC of each bacterial isolate, 3log<sub>10</sub> decrease in CFU was recorded after 6 hours of drug exposure and no growth was observed in almost all tested bacteria after 24 hours of exposure. No sign of toxidrome were observed during <italic>in vivo</italic> toxicity evaluation in mice at 300 mg/kg concentration.</p></sec><sec><title>Conclusion</title><p>In conclusion, the present study provides the scientific rational for medicinal use of <italic>Lawsonia inermis</italic>. The use of <italic>Lawsonia inermis</italic> extracts is of great significance as substitute antimicrobial agent in therapeutics.</p></sec>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Gull</surname><given-names>Iram</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>iramgull86@yahoo.com</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Sohail</surname><given-names>Maria</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>twinklestar_47@yahoo.com</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Aslam</surname><given-names>Muhammad Shahbaz</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>shahbazaslam111@gmail.com</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Athar</surname><given-names>Muhammad Amin</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>amatmian@yahoo.com</email></contrib>	Annals of Clinical Microbiology and Antimicrobials	<sec sec-type="intro"><title>Introduction</title><p>In traditional herbal medicine, plants have been used for many years [<xref ref-type="bibr" rid="B1">1</xref>]. Therefore, plants have attained status of natural source of new and potent antimicrobial agents [<xref ref-type="bibr" rid="B2">2</xref>]. Medicinal plants are used as ethnomedicine in different countries around the world [<xref ref-type="bibr" rid="B3">3</xref>] and are source of natural products providing unlimited opportunity for new drugs because of readily available medicinal diversity [<xref ref-type="bibr" rid="B4">4</xref>].</p><p><italic>Lawsonia Inermis</italic> (<italic>L. inermis</italic>) is a scientific name of a tall shrub plant commonly known as Henna or Mehndi [<xref ref-type="bibr" rid="B5">5</xref>] or mignonette tree [<xref ref-type="bibr" rid="B6">6</xref>] belongs to Kingdom: Plantae, Division: Angiospermae, Class: Dicotyledoneae, Order: Myrtales, Family: Lythraceae, Genus: Lawsonia, Species: L. inermis [<xref ref-type="bibr" rid="B7">7</xref>]. Henna is a flowering plant, having a height of 5 meters, natal to subtropical and tropical regions of world including South Asia, Africa, oases of Sahara Dessert and even in northern regions of Australia. Leaves of henna plant are entire, opposite, sub-sessile, oval-shaped and smooth [<xref ref-type="bibr" rid="B8">8</xref>]. Leaves have length of 2–3 cm with 1–2 cm width [<xref ref-type="bibr" rid="B5">5</xref>]. Henna shrub is highly branched and has greyish-brown barks [<xref ref-type="bibr" rid="B9">9</xref>].</p><p>Main chemical constituents of henna are Lawsone (2-hydroxynaphthoquinone), mucilage, mannite, gallic acid and tannic acid [<xref ref-type="bibr" rid="B10">10</xref>]. Henna is known to be used as a cosmetic agent for dyeing hair, nails and skin [<xref ref-type="bibr" rid="B11">11</xref>].</p><p>In traditional medicine, henna plant is used to treat many diseases like oedema, bronchitis, menstrual disorder, rheumatism, hemorrhoids and even in jaundice, leprosy, pain, spleen enlargement, dysentery and skin problems [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B12">12</xref>-<xref ref-type="bibr" rid="B14">14</xref>]. Henna can also be used as an astringent and antihemorragic agent and is also known for its hypotensive, cardio inhibitory and sedative effects [<xref ref-type="bibr" rid="B9">9</xref>]. In addition, henna is reported to show some other properties including hypoglycemic [<xref ref-type="bibr" rid="B15">15</xref>], immunostimulant [<xref ref-type="bibr" rid="B16">16</xref>], hepatoprotective [<xref ref-type="bibr" rid="B17">17</xref>], anti-inflammatory [<xref ref-type="bibr" rid="B18">18</xref>], tuberculostatic [<xref ref-type="bibr" rid="B19">19</xref>], anti-cancer and antioxidant properties [<xref ref-type="bibr" rid="B20">20</xref>].</p><p>The present research is designed to determine the antimicrobial activity of leaves of <italic>Lawsonia inermis</italic> available locally in Pakistan against certain pathogenic bacterial strains.</p></sec><sec sec-type="materials\|methods"><title>Materials and methods</title><sec><title>Bacterial cultures</title><p>Bacterial cultures used in the present study were clinical isolates collected from the Shaikh Zayed Hospital and Jinnah Hospital, Lahore, Pakistan. The cultures comprise of four Gram negative bacterial isolates namely <italic>Escherichia coli</italic>, <italic>Salmonella typhi</italic>, <italic>Klebsiella spp.</italic>, <italic>Shigella sonnei</italic> and three Gram positive bacterial isolates namely <italic>Bacillus subtilis, Staphylococcus aureus</italic>, <italic>Staphylococcus epidermidis</italic>.</p></sec><sec><title>Maintenance of bacterial cultures</title><p>All the bacterial isolates were cultured and maintained in LB (Luria Bertani) medium (1% Tryptone, 1% sodium chloride, 0.5% yeast extract) during all the experiments of the study until mentioned. The bacterial cultures were refreshed fortnightly.</p></sec><sec><title>Plant material</title><p>The dried leaves of <italic>Lawsonia inermis</italic> (Henna) used in the present study were purchased from the local market of Lahore, Pakistan.</p></sec><sec><title>Preparation of plant extract</title><p>To determine the <italic>in vitro</italic> antimicrobial activity of <italic>Lawsonia inermis</italic>, four different types of extracts including methanol extract, aqueous extract, chloroform extract and acetone extract were prepared.</p><p>For the preparation of extracts, dried leaves of Henna were ground to fine powder mechanically in electric grinder. Powdered leaves (10 g) were added in four flasks of 100 ml volume and 50ml of each solvent was added to each flask separately. The flasks were kept in incubator at 37°C overnight with shaking at 180 rpm. The contents of flask were first filtered through four layers of muslin cloth and then through Whatman filter paper. The filtrate was evaporated in rotary evaporator at 50°C. The weight of residues was recorded. DMSO (Dimethyl sulfoxide) was used to dissolve the residues of methanol, acetone and chloroform extracts while aqueous extract residues were dissolved in distilled water at different concentrations. The resulting extracts were stored at 4°C for further use in experiments.</p></sec><sec><title>Inoculum preparation</title><p>Before performing antimicrobial activity assay each bacterial strain was refreshed in 5 ml of LB broth (pH 7) separately under sterile conditions. Cultures were incubated in shaking incubator at 160 rpm for 16 hours at 37°C. Each bacterial culture was maintained at the concentration of 10<sup>7</sup> CFU/ml.</p></sec><sec><title>Antimicrobial sensitivity test using disc diffusion method</title><p>The assay of antimicrobial activity of <italic>Lawsonia inermis</italic> extracts was performed by Disc diffusion method [<xref ref-type="bibr" rid="B21">21</xref>]. Disc impregnated with DMSO were used as control. The diameter of zones of inhibition formed was measured in mm (millimeters). The test was performed in triplicate with each bacterial strain and mean zone of inhibition was recorded.</p></sec><sec><title>Determination of minimum inhibitory concentration (MIC)</title><p>MIC of four different extracts of <italic>Lawsonia inermis</italic> was determined against the test bacterial cultures using the method described by Natta et al. [<xref ref-type="bibr" rid="B22">22</xref>] with slight modifications. Briefly, starting from highest concentration of each extract of <italic>Lawsonia inermis</italic> serial dilution were prepared ranging from 544–17 mg/ml for methanol extract, 70–2.18 mg/ml for chloroform extract, 660–20.65 mg/ml for aqueous extract and 74.2-2.31 mg/ml for acetone extract. DMSO was used as diluent for all extract except for aqueous extract where water was used instead of DMSO. Sterile discs were dipped in different dilutions for 1 min and placed on LB agar plates seeded with each bacterial culture separately. The whole experiment was performed under aseptic conditions. Plates were then incubated at 37°C for 16 hrs. The minimum concentration of each extract with clear zone of inhibition was considered as MIC. The zone of inhibition in each case was measured as the diameter of the clear zone and results were recorded. Each experiment was performed in triplicate.</p></sec><sec><title>Phytochemical analysis</title><p>The extracts of <italic>Lawsonia inermis</italic> prepared in the present study were screened for phytochemicals including carbohydrates, cardioglycosides, terpenoids, tannins, phenolic compounds, proteins and quinones by phytochemical analysis as below [<xref ref-type="bibr" rid="B23">23</xref>].</p><sec><title><italic>Carbohydrates</italic></title><p>1 ml of each of four different extracts was taken in test tubes separately and treated with 5 ml of Fehling’s solution (Solution A: 34.6 g of copper (II) sulfate pentahydrate dissolved in 500 ml distilled water, Solution B: 125 g of potassium hydroxide and 173 g of potassium sodium tartrate tetrahydrate dissolved in 500 ml of distilled water, combine solution A and solution B (1:1) just before use). The test tubes were placed in boiling water bath for 5 min. The tubes were observed for appearance of yellow or red color precipitates indicating the presence of reducing sugars.</p></sec><sec><title><italic>Cardioglycosides</italic></title><p>5 ml of each of the four extracts was taken in test tubes separately and treated with 2 ml of glacial acetic acid having a drop of ferric chloride solution. 1 ml of the concentrated sulphuric acid was added to each test tube. Test tubes were observed for the appearance of brown coloured ring at the interface indicating the presence of cardioglycosides.</p></sec><sec><title><italic>Terpenoids</italic></title><p>5 ml of each of the four extracts was taken in test tubes separately and mixed with 2 ml of chloroform. Concentrated sulphuric acid was added to form a layer. Test tubes were observed for the appearance of reddish brown colour at the interface.</p></sec><sec><title><italic>Tannins</italic></title><p>2 ml of each extract of <italic>Lawsonia inermis</italic> was mixed with few drops of 0.1% ferric chloride solution in test tubes separately. Test tubes were observed for the appearance of brownish green colour.</p></sec><sec><title><italic>Phenolic compounds</italic></title><p>1 ml of each extract of <italic>Lawsonia inermis</italic> was mixed with 4 drops of ethanol and 3 drops of 0.1% ferric chloride solution in test tube separately. Test tubes were observed for the appearance of red color.</p></sec><sec><title><italic>Proteins</italic></title><p>1 ml of each extract was taken in test tubes separately. 2 drops of freshly prepared 0.2% ninhydrin reagent (2.5 g of ninhydrin dissolved in 50 ml n-butyl alcohol on mild heating and stirring and diluted 500 ml with n-butyl alcohol) were added. Test tubes were heated for few minutes. Test tubes were observed for the appearance of blue color.</p></sec><sec><title><italic>Quinones</italic></title><p>Few drops of 1 N sodium hydroxide solution were mixed with 1 ml of each extract of <italic>Lawsonia inermis</italic> in test tubes separately. Test tubes were observed for the appearance of red colour indicating the presence of quinones.</p></sec></sec><sec><title>Time-Kill kinetic analysis</title><p>The rate of bacterial killing was determined using <italic>Lawsonia inermis</italic> extracts with least MIC value for each bacterial isolate by time-kill kinetic assay as described by Miyasaki et al. [<xref ref-type="bibr" rid="B24">24</xref>] with slight modifications. Briefly, overnight bacterial cultures were diluted to the 5 × 10<sup>5</sup> CFU/mL with LB broth supplemented with 1× MIC of <italic>Lawsonia inermis</italic> extract for each bacterial isolate. The cultures were grown at 37°C with agitation at 160 rpm. The aliquots of cultures were collected at different time intervals (0 hr, 1 hr, 6 hr, 12 hrs, 24 hrs, 36 hrs and 48 hrs), serially diluted in LB broth and plated onto LB agar plates. After incubating the plates at 37°C for 16 hours viable colonies were enumerated. The results were recorded in terms of log<sub>10</sub> CFU and plotted vs. time for each bacterial isolate.</p></sec><sec><title>Toxicological evaluation of Lawsonia inermis extract</title><p>The experiments of acute toxicity in animals were performed in the animal house of Institute of Biochemistry and Biotechnology, University of the Punjab, Lahore, Pakistan after the approval of departmental ethical committee. Each of the two groups comprised of 10 male albino mice (weighing 150-200 g) were used to evaluate the acute toxicity of <italic>Lawsonia inermis</italic> ethanol extract. The animals were housed in cages and served with proper diet according to the international standards. One group was administered with ethanol extract of <italic>Lawsonia inermis</italic> (300 mg/Kg) and other group with equal volume of vehicle (DMSO) daily through subcutaneous route for 2 weeks by subcutaneous injection. The animals were continuously observed for signs of toxidromes such as aggression, sedation, rising fur, increased respiration, altered cardiac rate, excitation, convulsion, stupor, vomiting, etc. or death in first 2 hours and then after 24 hours.</p></sec><sec><title>Statistical analysis</title><p>Values are mean of ± standard deviation of three triplicates.</p></sec></sec><sec><title>Results and discussion</title><p>Dried leaves of <italic>Lawsonia inermis</italic> were used to prepare the extracts as it has been reported that dried preparation have more concentrated active phytochemical compounds than fresh plant material [<xref ref-type="bibr" rid="B25">25</xref>]. Four different types of extracts were prepared including methanol extract, chloroform extract, aqueous extract and acetone extract. The results revealed that all extracts exhibited antimicrobial activity against all bacterial strains used in the present study. However bacterial strains showed differential sensitivity for each extract (Figure <xref ref-type="fig" rid="F1">1</xref>). Antimicrobial activity was not observed with controls (DMSO and water). According to the study of Papageorgiou et al. [<xref ref-type="bibr" rid="B26">26</xref>], phytochemical constituents of <italic>Lawsonia inermis</italic> exhibit antimicrobial activity only against gram positive bacteria while ineffective for gram negative bacteria. In our study, it was interested to note that <italic>Lawsonia inermis</italic> had antimicrobial activity against both gram positive (<italic>S. aureus</italic>, <italic>B. subtilis</italic> and <italic>S. epidermidis</italic>) and gram negative (<italic>E. coli</italic>, <italic>S. typhi</italic>, <italic>Klebsiella spp.</italic> and <italic>Shigella</italic>) bacteria. The studies of Bhuwaneshwari et al. [<xref ref-type="bibr" rid="B13">13</xref>]; Habbal et al. [<xref ref-type="bibr" rid="B27">27</xref>] and Hussain et al. [<xref ref-type="bibr" rid="B28">28</xref>] support our findings.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Zone of inhibition (mm) of </bold><bold> <italic>Lawsonia inermis </italic> </bold><bold>extracts against tested bacterial isolates.</bold></p></caption><graphic xlink:href="1476-0711-12-36-1"/></fig><p>The values of MIC for each tested bacterial strain had been shown in Figure <xref ref-type="fig" rid="F2">2</xref>. From the data in Figure <xref ref-type="fig" rid="F2">2</xref>, it was illustrated that all of the tested bacterial isolates showed minimum value of MIC for chloroform extract except <italic>E. coli</italic> and <italic>B. subtilis</italic>. The minimum MIC value of <italic>E. coli</italic> (9.27 mg/ml) and <italic>B. subtilis</italic> (2.31 mg/ml) was observed using acetone extract. The MIC values of our study were less than the MIC values reported by Al-kurashy et al. [<xref ref-type="bibr" rid="B29">29</xref>]. They found MIC values in the range of 8–64 mg/ml for aqueous extract and 32–64 mg/ml for alcoholic extract of <italic>Lawsonia inermis</italic> against <italic>E. coli</italic>, <italic>S. aureus</italic>, <italic>P. aureginosa</italic> and <italic>E. faecalis</italic>. It was established that chloroform extract of <italic>Lawsonia inermis</italic> was more promising antimicrobial agent for <italic>S. aureus</italic>, <italic>S. epidermidis, S. typhi</italic>, <italic>Klebsiella spp.</italic> and <italic>Shigella</italic> while its acetone extract for <italic>E. coli</italic> and <italic>B. subtilis</italic> at least in <italic>in vitro</italic> antimicrobial assay.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Minimum inhibitory concentration (MIC) of different extracts of </bold><bold> <italic>Lawsonia inermis </italic> </bold><bold>against tested bacterial isolates.</bold></p></caption><graphic xlink:href="1476-0711-12-36-2"/></fig><p>The secondary metabolites mainly attribute the antimicrobial activity of plants [<xref ref-type="bibr" rid="B30">30</xref>]. The active constituents of these secondary metabolites include phenolic compounds and tannins [<xref ref-type="bibr" rid="B31">31</xref>]. In order to identify the metabolites present in different extracts of <italic>Lawsonia inermis</italic> phytochemical analysis were performed. The data in Table <xref ref-type="table" rid="T1">1</xref> depicted that methanol, acetone and aqueous extracts had cardioglycosides, terpenoids, carbohydrates, phenols, quinones and tannins. While proteins were absent in all three of these extracts. The chloroform extract had cardioglycosides, carbohydrates, phenols, quinones and tannins while proteins and terpenoids were absent. These metabolites solubilized in solvent on the bases for polarity. In most of the plant materials, water soluble components includes starches, tannins, saponins, polypeptides, terpenoids, lectins and different ions [<xref ref-type="bibr" rid="B32">32</xref>], while alcoholic extract includes flavonol, alkaloids, tannins, sterols polyphenols etc. [<xref ref-type="bibr" rid="B33">33</xref>]. Main chemical constituents of henna include Lawsone (2-hydroxynaphthoquinone), mucilage, mannite, gallic acid and tannic acid [<xref ref-type="bibr" rid="B10">10</xref>].</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p><bold>Phytochemical analysis of different extracts of </bold><bold> <italic>Lawsonia inermis</italic> </bold></p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th colspan="5" align="center" valign="bottom"><bold>Extracts of </bold><bold> <italic>Lawsonia inermis</italic> </bold><hr/></th></tr><tr><th align="center"> </th><th align="center"><bold>Methanol</bold></th><th align="center"><bold>Chloroform</bold></th><th align="center"><bold>Acetone</bold></th><th align="center"><bold>Aqueous</bold></th></tr></thead><tbody valign="top"><tr><td align="center" valign="bottom">Cardioglycosides<hr/></td><td align="center" valign="bottom"><bold>+</bold><hr/></td><td align="center" valign="bottom"><bold>+</bold><hr/></td><td align="center" valign="bottom"><bold>+</bold><hr/></td><td align="center" valign="bottom"><bold>+</bold><hr/></td></tr><tr><td align="center" valign="bottom">Terpenoids<hr/></td><td align="center" valign="bottom"><bold>+</bold><hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom"><bold>+</bold><hr/></td><td align="center" valign="bottom"><bold>+</bold><hr/></td></tr><tr><td align="center" valign="bottom">Carbohydrates<hr/></td><td align="center" valign="bottom"><bold>+</bold><hr/></td><td align="center" valign="bottom"><bold>+</bold><hr/></td><td align="center" valign="bottom"><bold>+</bold><hr/></td><td align="center" valign="bottom"><bold>+</bold><hr/></td></tr><tr><td align="center" valign="bottom">Proteins<hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom">-<hr/></td></tr><tr><td align="center" valign="bottom">Phenols<hr/></td><td align="center" valign="bottom"><bold>+</bold><hr/></td><td align="center" valign="bottom"><bold>+</bold><hr/></td><td align="center" valign="bottom"><bold>+</bold><hr/></td><td align="center" valign="bottom"><bold>+</bold><hr/></td></tr><tr><td align="center" valign="bottom">Quinones<hr/></td><td align="center" valign="bottom"><bold>+</bold><hr/></td><td align="center" valign="bottom"><bold>+</bold><hr/></td><td align="center" valign="bottom"><bold>+</bold><hr/></td><td align="center" valign="bottom"><bold>+</bold><hr/></td></tr><tr><td align="center">Tannins</td><td align="center"><bold>+</bold></td><td align="center"><bold>+</bold></td><td align="center"><bold>+</bold></td><td align="center"><bold>+</bold></td></tr></tbody></table><table-wrap-foot><p>(+ Sign indicating presence of compound; - sign indicating absence of compound).</p></table-wrap-foot></table-wrap><p>The rate of bacterial killing after exposure to the 1× MIC of respective extract of <italic>Lawsonia inermis</italic> for each isolate is summarized in Figure <xref ref-type="fig" rid="F3">3</xref>. The time required to achieve 3log<sub>10</sub> decrease in CFU is an acceptable index of bactericidal activity from time-kill analysis [<xref ref-type="bibr" rid="B34">34</xref>]. The results illustrated that not a single bacterial isolate showed significant bactericidal activity in first hour. Whereas, 3log<sub>10</sub> reduction in viability of all tested bacterial isolates was observed after 12 hours of exposure. The log<sub>10</sub> CFU of all bacterial isolates was reduced to zero after 24 hours of exposure except <italic>S. epidermidis</italic> (36 hours) and <italic>B. subtilis</italic> (48 hours).</p><fig id="F3" position="float"><label>Figure 3</label><caption><p>Time-Kill kinetic analysis of tested bacterial isolates.</p></caption><graphic xlink:href="1476-0711-12-36-3"/></fig><p>About 25% of all medicines available in the market have been derived directly or indirectly from plants [<xref ref-type="bibr" rid="B35">35</xref>,<xref ref-type="bibr" rid="B36">36</xref>]. Herbal medicines are generally believed as safe. However, it is important to evaluate their biological safety before use to avoid fatal consequences [<xref ref-type="bibr" rid="B37">37</xref>]. There is no doubt in pharmacological properties of <italic>Lawsonia inermis</italic> but its toxicological assessment is also indispensable. <italic>In vivo</italic> acute toxicity of <italic>Lawsonia inermis</italic> extracts was checked in mice. No mortality was observed during the study. All the signs of toxidrome were negative.</p><p>In conclusion, the present study provides the scientific rational for medicinal use of <italic>Lawsonia inermis</italic>. The use of <italic>Lawsonia inermis</italic> extracts is of great significance as substitute antimicrobial agent in therapeutics.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>All authors equally participated in designing experiments, acquisition, analysis and interpretation of data. Prof. Dr. M. Amin Athar critical revise the manuscript and approved the final version of manuscript. All authors read and approved the final manuscript.</p></sec>
Slicing, sampling, and distance-dependent effects affect network measures in simulated cortical circuit structures	<p>The neuroanatomical connectivity of cortical circuits is believed to follow certain rules, the exact origins of which are still poorly understood. In particular, numerous nonrandom features, such as common neighbor clustering, overrepresentation of reciprocal connectivity, and overrepresentation of certain triadic graph motifs have been experimentally observed in cortical slice data. Some of these data, particularly regarding bidirectional connectivity are seemingly contradictory, and the reasons for this are unclear. Here we present a simple static geometric network model with distance-dependent connectivity on a realistic scale that naturally gives rise to certain elements of these observed behaviors, and may provide plausible explanations for some of the conflicting findings. Specifically, investigation of the model shows that experimentally measured nonrandom effects, especially bidirectional connectivity, may depend sensitively on experimental parameters such as slice thickness and sampling area, suggesting potential explanations for the seemingly conflicting experimental results.</p>	<contrib contrib-type="author"><name><surname>Miner</surname><given-names>Daniel C.</given-names></name><xref ref-type="author-notes" rid="fn001"><sup>*</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/124853"/></contrib><contrib contrib-type="author"><name><surname>Triesch</surname><given-names>Jochen</given-names></name><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/1023"/></contrib>	Frontiers in Neuroanatomy	<sec id="s1"><title>1. Introduction</title><p>Synaptic connectivity forms the anatomical substrate which gives rise to our cognitive abilities. It has been shown that much of the lateral recurrent connectivity of the cortex is significantly nonrandom. That is to say that the statistics of local connectivity do not follow that of a directed Erdős-Rényi graph, i.e., a graph in which all possible connections exist with equal and independent probability (Erdős and Rényi, <xref rid="B2" ref-type="bibr">1960</xref>). For example, Holmgren et al. (<xref rid="B6" ref-type="bibr">2003</xref>), Song et al. (<xref rid="B22" ref-type="bibr">2005</xref>), and Ko et al. (<xref rid="B8" ref-type="bibr">2011</xref>) note the presence of greater than expected bidirectional connectivity, a feature that has been suggested as a key requirement for the sort of large-scale recurrent excitation that is seen and computation that is believed to take place in the neocortex (Douglas et al., <xref rid="B1" ref-type="bibr">1995</xref>). Lefort et al. (<xref rid="B9" ref-type="bibr">2009</xref>), on the other hand, notes no excess of bidirectional connectivity. Song et al. (<xref rid="B22" ref-type="bibr">2005</xref>) additionally notes greater than expected counts of certain triangular or triadic network motifs (three-neuron connectivity patterns) (Milo et al., <xref rid="B13" ref-type="bibr">2002</xref>). Yoshimura et al. (<xref rid="B31" ref-type="bibr">2005</xref>) examines specific microstructure, including bidirectional connections, within cortical columns. Perin et al. (<xref rid="B15" ref-type="bibr">2011</xref>) notes greater than expected common neighbor clustering, a phenomenon in which pairs of neurons sharing a greater number of common neighbors are more likely to be connected themselves, while Perin et al. (<xref rid="B16" ref-type="bibr">2013</xref>) further examines the structural implications of this above-chance common neighbor clustering. Morgan and Soltesz (<xref rid="B14" ref-type="bibr">2008</xref>), Litwin-Kumar and Doiron (<xref rid="B11" ref-type="bibr">2012</xref>), and McDonnell and Ward (<xref rid="B12" ref-type="bibr">2014</xref>) highlight some of the functional implications of clustering in balanced cortex-like networks. Rubinov and Sporns (<xref rid="B18" ref-type="bibr">2010</xref>) provides an overview of graph measures that might be applied to brain networks.The abundance of nonrandom features suggests that there may be some computational or metabolic advantage to the particular connectivity structure of the cortex. It is an open question which nonrandom features are developed as a result of direct genetic programming, neural plasticity under structured input, and spontaneous self-organization (Prill et al., <xref rid="B17" ref-type="bibr">2005</xref>).</p><p>The connectome, which we take here to refer to the micro-scale, or neuron-and-synapse connectivity of the brain Sporns et al. (<xref rid="B23" ref-type="bibr">2005</xref>) is a detailed and difficult thing to study. Numerous methods exist for its study, including (but not limited to) increasingly detailed histological techniques (Kleinfeld et al., <xref rid="B7" ref-type="bibr">2011</xref>, for example) and, more commonly, as they allow access to synaptic strengths and dynamics in addition to structure, electrophysiological recordings. We focus here on the most common implementation of the latter, involving the preparation of and recording from <italic>in vitro</italic> slices of cortical tissue. Though it provides more information about individual connections, the overall picture provided by electrophysiological techniques is affected by sampling biases and constraints (Seung, <xref rid="B20" ref-type="bibr">2009</xref>). Traditionally, the primary concern regarding such biases and constraints has been accurate reconstruction of very small sections of circuitry. However, as techniques improve and the available sections get larger and more densely sampled, and in particular as statistical network measures are utilized more and more, it becomes important to study the effect of these biases and constraints on the network measures as well.</p><p>We examine here a simple model for horizontal connectivity in the cortex under intersomatic distance-dependent connection constraints. This simple distance-dependence results in the formation of several nonrandom features including, but not limited to, common neighbor clustering, excess reciprocal or bidirectional connectivity, and an overrepresentation of certain triadic motifs. We perform virtual slicing and sampling on this model, similar to what would be done in a physiological experiment, and examine how the results depend on slice thickness and the size of the sampling area from which cells are probed. We find, encouragingly, that such complex nonrandom features can be seeded (if not fully instantiated to the degree at which they are experimentally observed) by such simple distance-dependent phenomenon. We also find, more troublingly, that the observed representation of some of these features depends strongly on interactions of scale between the connectivity profiles, the cortical structures, and the slicing and sampling thereof. We discuss in this paper the implications of these phenomena and conclude that in order to correctly interpret data on cortical connectivity and its nonrandom features, close attention has to be paid to the exact experimental parameters such as slice thickness and sampling area.</p></sec><sec id="s2"><title>2. Materials and methods</title><p>Our model is designed to represent a virtual slab of cortical layer V in rodents. The slab's dimensions are 1000 × 1000 μm, with a thickness of 300 μm (the lattermost dimension describing the approximate thickness of layer V of the rodent cortical sheet (Schüz and Palm, <xref rid="B19" ref-type="bibr">1989</xref>) (see Figure <xref ref-type="fig" rid="F1">1</xref>). We assume a cortical neuronal density of at least 20000 excitatory neurons per cubic mm, resulting in a total population of 6000 neurons, which are populated into the volume in a random, uniform fashion. This is a slight reduction in neuronal density from biological values, but is sufficient to demonstrate the phenomena we wish to explore and is necessary for rapid computational tractability. Though is is known that horizontal cortical axonal projections can reach lengths of several millimeters (Hirsch and Gilbert, <xref rid="B5" ref-type="bibr">1991</xref>), we choose to focus on local, sub-millimeter connectivity, as this is the scale of the microstructure typically being examined in network measure studies of cortical wiring. Various connectivity models, ranging in complexity from simple piecewise dense and sparse connectivity radii (Voges et al., <xref rid="B28" ref-type="bibr">2010a</xref>,<xref rid="B29" ref-type="bibr">b</xref>) to detailed reconstructions based on axonal and dendritic structure (Stepanyants et al., <xref rid="B24" ref-type="bibr">2008</xref>; Kleinfeld et al., <xref rid="B7" ref-type="bibr">2011</xref>), have been produced from experimental data. We select a continuous radial function for distance-dependent connectivity as solution between these two extremes. Our profile is a Gaussian with a half-width of 200 μm. This particular profile is chosen as a middle ground between the results of Song et al. (<xref rid="B22" ref-type="bibr">2005</xref>), who find no distance dependence up to a scale of 80–100 μm, and the results of Holmgren et al. (<xref rid="B6" ref-type="bibr">2003</xref>) and Perin et al. (<xref rid="B15" ref-type="bibr">2011</xref>), who find exponential distance dependence at a scale of 150–300 μm. The Gaussian compromise coarsely approximates both the flat top of the former result and the decay of the latter.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>An example of simulated slicing and sampling geometry, using a 300 μm slice and a 50 μm radius sampling area</bold>.</p></caption><graphic xlink:href="fnana-08-00125-g0001"/></fig><p>To produce the model graph, first, a 6000 × 6000 element distance matrix is constructed, with each element representing the euclidian distance between each pair of neurons. The boundary conditions are non-periodic, corresponding to slice boundary truncation. The connectivity profile function is then applied to each element, producing an unnormalized probability matrix, with each entry representing the pairwise connection probability. Self-connection probabilities are set to zero. The matrix is flattened into a vector and then the cumulative sum of the vector is taken and normalized, producing a cumulative distribution function (CDF). A look up table map is generated mapping each interval in the CDF to a particular pair of neurons.</p><p>The network is treated as a directed graph. A global connection fraction <italic>F</italic><sub><italic>C</italic></sub> is chosen upon model initialization, and the model is populated by generating random numbers in the interval [0,1] against the CDF and instantiating the edge mapped to the CDF interval in which each random number falls (rejecting already-instantiated edges) until the total number of edges reaches <italic>N</italic><sub>edges</sub> = <italic>F</italic><sub><italic>C</italic></sub> × (<italic>N</italic><sup>2</sup><sub>nodes</sub> − <italic>N</italic><sub>nodes</sub>).</p><p>Two sequential reduction procedures are then performed on the graph in order to simulate experimental sampling of the network. The first procedure simulates slicing. The virtual volume of the network is truncated along the X axis in Figure <xref ref-type="fig" rid="F1">1</xref> to correspond to the dimensions of a typical slice (50–500 μm, depending on the experiment). Edges and nodes that fall outside the truncation region are eliminated from the graph. The second procedure roughly simulates probing and sampling. In this procedure, a subset of nodes <italic>N</italic><sub>sample</sub> are randomly selected from a centered cylindrical volume within the slice of radius <italic>r</italic><sub>sample</sub> (50–300 μm, depending on the experiment), and a subgraph is constructed from these nodes and their respective edges. This subgraph is then taken to be equivalent to an electrophysiologically obtained sample. An example geometry of this virtual slicing and sampling is shown in Figure <xref ref-type="fig" rid="F1">1</xref>).</p><p>For any selected network, be it complete, a virtual slice, or a virtual sample, we compare properties against ensembles of two types of control graphs. The first control is a comparison against a purely random graph. It is a directed Erdős-Rényi graph (Erdős and Rényi, <xref rid="B2" ref-type="bibr">1960</xref>) parametrized by the same number of nodes and number of edges as the selected network.</p><p>The second control is a graph that naturally and randomly attains the amount of overrepresented bidirectional connections induced by the distance dependent connectivity, but contains no higher order effects. It is essentially a modified directed Erdős-Rényi-like graph parametrized by the number of nodes and the two independent probabilities of unidirectional connections and reciprocal or bidirectional connections. More explicitly, from the model graph, the fraction of node pairs that are unidirectionally connected and the fraction of node pairs that are bidirectionally connected is calculated. A new graph is then randomly populated with the same fractions of unidirectionally connected and bidirectionally connected edge pairs in an Erdős-Rényi-like fashion. This controls against an overrepresentation of motifs driven solely by excess bidirectional connectivity while preserving overrepresentation of motifs driven by higher order or more subtle forms of clustering.</p><p>The Python package NetworkX (Hagberg et al., <xref rid="B4" ref-type="bibr">2008</xref>) and a publicly available software script that counts triadic motifs in a directed graph (Levenson and van Liere, <xref rid="B10" ref-type="bibr">2011</xref>) are used to assist in the construction and analysis of graphs.</p><p>We will make comparisons between different sample and slice sizes based on overall connection fraction, bidirectional connection fraction, triadic motif count, and common neighbor clustering. We will demonstrate that sampling scale has a notable effect on how such properties are observed.</p></sec><sec id="s3"><title>3. Results</title><p>We select a global target connection fraction of 0.025 for the 1000 × 1000 × 300 μm layer V slab, as this produces a local connection fraction of 0.1 for a medium-sized slice and sample, as observed in numerous layer V slice studies (Thomson and Deuchars, <xref rid="B25" ref-type="bibr">1997</xref>; Thomson et al., <xref rid="B26" ref-type="bibr">2002</xref>). We select three slice thicknesses (in addition to the complete network) and three sampling radii with 100 neuron subsamples (except in the case of small sections, in which case the maximum number of neurons in the section is sampled). We will examine the complete network and complete slice statistics, as well as the sample statistics for each condition, and note how they vary. Unless otherwise specified, we average over five network samples.</p><p>The global connection fraction and bidirectional connection fraction for each condition is given in Tables <xref ref-type="table" rid="T1">1</xref>, <xref ref-type="table" rid="T2">2</xref>. We note that in general, for a given slice size, the overall connection fraction decreases with increasing sampling radius. This is an obvious result of local clustering due to the distance-dependent connection probability. Similarly, we note that as sampling radius increases, the number of bidirectional connections over chance (as compared to an Erdős-Rényi graph) increases. This is also a result of local clustering due to the distance-dependent connection probability.</p><table-wrap id="T1" position="float"><label>Table 1</label><caption><p><bold>Overall connection fraction (standard error)</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="left" rowspan="1" colspan="1"><bold>Slice size</bold></th><th align="center" rowspan="1" colspan="1"><bold>50 μm radius sample</bold></th><th align="center" rowspan="1" colspan="1"><bold>150 μm radius sample</bold></th><th align="center" rowspan="1" colspan="1"><bold>250 μm radius sample</bold></th><th align="center" rowspan="1" colspan="1"><bold>Complete section</bold></th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">Complete network</td><td align="center" rowspan="1" colspan="1">0.1343 (0.0063)</td><td align="center" rowspan="1" colspan="1">0.1066 (0.0014)</td><td align="center" rowspan="1" colspan="1">0.0749 (0.0030)</td><td align="center" rowspan="1" colspan="1">0.0250 (0.0000)</td></tr><tr><td align="left" rowspan="1" colspan="1">500 μm slice</td><td align="center" rowspan="1" colspan="1">0.1343 (0.0063)</td><td align="center" rowspan="1" colspan="1">0.1057 (0.0024)</td><td align="center" rowspan="1" colspan="1">0.0720 (0.0012)</td><td align="center" rowspan="1" colspan="1">0.0401 (0.0001)</td></tr><tr><td align="left" rowspan="1" colspan="1">300 μm slice</td><td align="center" rowspan="1" colspan="1">0.1343 (0.0063)</td><td align="center" rowspan="1" colspan="1">0.1060 (0.0021)</td><td align="center" rowspan="1" colspan="1">0.0827 (0.0034)</td><td align="center" rowspan="1" colspan="1">0.0495 (0.0001)</td></tr><tr><td align="left" rowspan="1" colspan="1">100 μm slice</td><td align="center" rowspan="1" colspan="1">0.1343 (0.0063)</td><td align="center" rowspan="1" colspan="1">0.1151 (0.0016)</td><td align="center" rowspan="1" colspan="1">0.0936 (0.0025)</td><td align="center" rowspan="1" colspan="1">0.0566 (0.0007)</td></tr></tbody></table></table-wrap><table-wrap id="T2" position="float"><label>Table 2</label><caption><p><bold>Bidirectional connection fraction (standard error) [fraction of chance – Erdős-Rényi control]</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="left" rowspan="1" colspan="1"><bold>Slice size</bold></th><th align="center" rowspan="1" colspan="1"><bold>50 μm radius sample</bold></th><th align="center" rowspan="1" colspan="1"><bold>150 μm radius sample</bold></th><th align="center" rowspan="1" colspan="1"><bold>250 μm radius sample</bold></th><th align="center" rowspan="1" colspan="1"><bold>Complete section</bold></th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">Complete network</td><td align="center" rowspan="1" colspan="1">0.0195 (0.0042) [1.0828]</td><td align="center" rowspan="1" colspan="1">0.0124 (0.0012) [1.0962]</td><td align="center" rowspan="1" colspan="1">0.0066 (0.0014) [1.1832]</td><td align="center" rowspan="1" colspan="1">0.0020 (0.0000) [3.1705]</td></tr><tr><td align="left" rowspan="1" colspan="1">500 μm slice</td><td align="center" rowspan="1" colspan="1">0.0195 (0.0042) [1.0828]</td><td align="center" rowspan="1" colspan="1">0.0126 (0.0013) [1.1228]</td><td align="center" rowspan="1" colspan="1">0.0065 (0.0006) [1.2561]</td><td align="center" rowspan="1" colspan="1">0.0034 (0.0000) [2.1140]</td></tr><tr><td align="left" rowspan="1" colspan="1">300 μm slice</td><td align="center" rowspan="1" colspan="1">0.0195 (0.0042) [1.0828]</td><td align="center" rowspan="1" colspan="1">0.0115 (0.0010) [1.0253]</td><td align="center" rowspan="1" colspan="1">0.0084 (0.0013) [1.2185]</td><td align="center" rowspan="1" colspan="1">0.0046 (0.0001) [1.8877]</td></tr><tr><td align="left" rowspan="1" colspan="1">100 μm slice</td><td align="center" rowspan="1" colspan="1">0.0195 (0.0042) [1.0828]</td><td align="center" rowspan="1" colspan="1">0.0143 (0.0014) [1.0841]</td><td align="center" rowspan="1" colspan="1">0.0101 (0.0017) [1.1517]</td><td align="center" rowspan="1" colspan="1">0.0060 (0.0001) [1.8915]</td></tr></tbody></table></table-wrap><p>We examine the common neighbor behavior in <bold>Figures 3–6</bold>. The common neighbor effect is measured as follows. Pairs of neurons sharing each possible number of commonly connected neighbors (up to some maximum value) are counted, ignoring directionality (see Figure <xref ref-type="fig" rid="F2">2</xref>). For each number of commonly connected neighbors, the number of connected neuron pairs is divided by the total number of neuron pairs, resulting in a connection probability conditioned on the number of common neighbors. The the steeper the slope of this measure as a function of number of common neighbors is, the stronger the effect (Perin et al., <xref rid="B15" ref-type="bibr">2011</xref>). For an Erdős-Rényi graph, this common neighbor effect measure will have, on average, a slope of zero and a value equal to the overall connection probability (up until the maximum number of neighbors). Common neighbor clustering should not be confused with more traditional clustering measures (Watts and Strogatz, <xref rid="B30" ref-type="bibr">1998</xref>; Fagiolo, <xref rid="B3" ref-type="bibr">2007</xref>). Common neighbor effect is taken here as an undirected measure for two reasons: alignment with the convention of Perin et al. (<xref rid="B15" ref-type="bibr">2011</xref>), and because our simple structural model has no directional preference, and can thus make no prediction about it. In an actual biological or more complex simulated system, it is likely that in and out (to and from) common neighbor effects would produce different results, as is suggested in the supplementary material of Perin et al. (<xref rid="B15" ref-type="bibr">2011</xref>).</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Common neighbor clustering illustrated</bold>. Tested nodes are red; common neighbors are blue.</p></caption><graphic xlink:href="fnana-08-00125-g0002"/></fig><p>Figure <xref ref-type="fig" rid="F3">3</xref> shows the total common neighbor effect for each entire slice. We note, firstly, that the slope of the common neighbor clustering increases with decreasing section size, and secondly, that the saturation point decreases with decreasing section size. We speculate that this occurs due to the truncation of connections that occurs upon slicing, and the resulting tendency of only nearby neurons to be well-connected. Similarly, for each individual slice thickness (Figures <xref ref-type="fig" rid="F4">4</xref>–<xref ref-type="fig" rid="F6">6</xref>), the saturation point increases with decreasing sampling radius. The overall effect also becomes less pronounced for the smaller (in this case, 100 neuron) samples, as would be expected. The strength of common neighbor clustering is sensitive to both the neuronal and connection densities, and the size of the distance-dependent connection probability, particularly as it relates to the sampling scale. It is the sensitivity to the relationship between these scales that we wish to emphasize in these results.</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Common neighbor clustering for complete network and slices (full sampling): pairwise connection probability as a function of number of commonly connected neighbors</bold>. Error bars indicate standard error of the mean. Average over five populations.</p></caption><graphic xlink:href="fnana-08-00125-g0003"/></fig><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>Common neighbor clustering for 500 μm slice: pairwise connection probability as a function of number of commonly connected neighbors</bold>. Error bars indicate standard error of the mean. Average over five populations.</p></caption><graphic xlink:href="fnana-08-00125-g0004"/></fig><fig id="F5" position="float"><label>Figure 5</label><caption><p><bold>Common neighbor clustering for 300 μm slice: pairwise connection probability as a function of number of commonly connected neighbors</bold>. Error bars indicate standard error of the mean. Average over five populations.</p></caption><graphic xlink:href="fnana-08-00125-g0005"/></fig><fig id="F6" position="float"><label>Figure 6</label><caption><p><bold>Common neighbor clustering for 100 μm slice: pairwise connection probability as a function of number of commonly connected neighbors</bold>. Error bars indicate standard error of the mean. Average over five populations.</p></caption><graphic xlink:href="fnana-08-00125-g0006"/></fig><p>Experimental data (Perin et al., <xref rid="B15" ref-type="bibr">2011</xref>) shows an above-chance common neighbor effect stronger than the one demonstrated by our model for similar sampling conditions, suggesting the presence of additional clustering mechanisms in the cortex beyond the simple geometric ones examined in our model. One prediction our model makes is that after a linear or near-linear rise in connection probability as function of common neighbor count, the connection probability saturates for some large number of common neighbors. It can be extrapolated, despite the increased common neighbor effect seen in physiological data, that this sort of turnover and saturation effect will still necessarily occur for a large number of common neighbors given a sufficiently thorough sampling of a section of cortical tissue.</p><p>We examine the counts of occurrences of directed triadic motifs (possible directed triangular subgraph configurations; see Figure <xref ref-type="fig" rid="F7">7</xref>) in the simulated tissue sections compared with Erdős-Rényi random graphs for complete sections and for a sampled 300 μm slice in Figures <xref ref-type="fig" rid="F8">8</xref>, <xref ref-type="fig" rid="F9">9</xref> (which is representative of sliced and sampled behavior, as it is observed that sliced and sampled behavior does not vary much between slice sizes; only sample radii). We note an excess of motifs with bidirectional connections. This is trivially expected from distant-dependent connection probabilities; since each direction in an edge is treated independently it will of course be the case that many minimally separated nodes will be bidirectionally connected, and, more generally that inter-group connectivity will be increased among tight groups of neurons. Furthermore, it is trivially the case that given an excess of bidirectional connections, triads containing them will be overrepresented. We wish to correct for this second effect, and do so via the bidirectionality corrected control described in the Materials and Methods section and elucidated below.</p><fig id="F7" position="float"><label>Figure 7</label><caption><p><bold>Triadic motif key</bold>.</p></caption><graphic xlink:href="fnana-08-00125-g0007"/></fig><fig id="F8" position="float"><label>Figure 8</label><caption><p><bold>Triadic motif counts for complete sections (full sampling)</bold>. Error bars indicate standard error of the mean. Average over five populations.</p></caption><graphic xlink:href="fnana-08-00125-g0008"/></fig><fig id="F9" position="float"><label>Figure 9</label><caption><p><bold>Triadic motif counts for 300 μm slice</bold>. Error bars indicate standard error of the mean. Average over five populations.</p></caption><graphic xlink:href="fnana-08-00125-g0009"/></fig><p>We examine triadic motif counts against bidirectionality-corrected random graphs for complete sections and for a sampled 300 μm slice in Figures <xref ref-type="fig" rid="F10">10</xref>, <xref ref-type="fig" rid="F11">11</xref>. Again, sliced and sampled behavior does not vary much between slice sizes; only sample radii. We note that even after bidirectionality correction, excesses of closed-loop (i.e., connected on all sides) triadic motifs containing bidirectionally connected pairs remain. Of interest as well is the excess of closed but non-bidirectional triadic motifs (numbers 10 and 11) remaining. We note, in general, that motifs 10 -16 remain overrepresented, a phenomenon seen as well in Song et al. (<xref rid="B22" ref-type="bibr">2005</xref>). An underrepresentation of motif 8, which is observed in Song et al. (<xref rid="B22" ref-type="bibr">2005</xref>) with a similar strength to the aforementioned overrepresentations, is not seen in our model. However, the purpose of this paper is not to fully analyze the more subtle effects of distant-dependent clustering, but rather to examine the implications of similar clustering occurring at the same spatial scale as variations in sampling. We note, firstly, that as slice size decreases, the statistics of the complete slice approach the statistics of the sample. This follows logically from the fact that the sample occupies an increasing fraction of the slice by volume for a smaller slice. Along similar lines, we note that thinner slices exhibit less variation in the counts between sampling radii. For a sufficiently thin slice, one could hypothetically move from a three-dimensional to a two-dimensional reference model, approximating a sheet. We also note that post-bidirectionality correction in the control, the variation between slice sizes and sample radii is smaller than it was pre-bidirectionality correction in the control. This is a strong indicator that any motif surveys undertaken would benefit from using a bidirectionality or similar (as in Song et al., <xref rid="B22" ref-type="bibr">2005</xref>) correction on the control in order to maximize consistency and universality in results.</p><fig id="F10" position="float"><label>Figure 10</label><caption><p><bold>Bidirectionally corrected triadic motif counts for complete sections (full sampling)</bold>. Error bars indicate standard error of the mean. Average over five populations.</p></caption><graphic xlink:href="fnana-08-00125-g0010"/></fig><fig id="F11" position="float"><label>Figure 11</label><caption><p><bold>Bidirectionally corrected triadic motif counts for 300 μm slice</bold>. Error bars indicate standard error of the mean. Average over five populations.</p></caption><graphic xlink:href="fnana-08-00125-g0011"/></fig></sec><sec id="s4"><title>4. Discussion</title><p>As we are able to access larger and denser subsamples of the connectome, complex network measures (Rubinov and Sporns, <xref rid="B18" ref-type="bibr">2010</xref>) are becoming an increasingly important way of understanding both the structure and function. Such measures have already been applied to the complete connectome of <italic>C. elegans</italic> (Varshney et al., <xref rid="B27" ref-type="bibr">2011</xref>). While elements of this study are highly telling, they do not provide a direct comparison to cortical slice studies, which are subsampled portions of a very different structure, even if the individual elements are similar. Currently, cortical slice studies provide some of the best information we have about the wiring structure of the cortex on a microscopic scale.</p><p>In order to understand this microstructure, it is very important to study and examine the statistics of connectivity at scales of tens to hundreds of μm—this will be vital to understanding the self-organizational and computational principles underlying the structure of the brain (Prill et al., <xref rid="B17" ref-type="bibr">2005</xref>; Sporns et al., <xref rid="B23" ref-type="bibr">2005</xref>; Seung, <xref rid="B20" ref-type="bibr">2009</xref>). However, at the same time, extreme care must be taken, as relatively small variations in section size and sampling density can lead to significantly differing results, as this is also the scale at which naturally occurring simple clustering may occur, and at which the statistical transition from microstructure to macrostructure may take place as well.</p><p>It is thus of great importance that experimenters take this into account and, accordingly, provide all available information regarding neuron type and approximate density, sampling space distribution, slice thickness, and other parameters that might lead to sampling biases. Various studies of such microstructure have shown conflicting results. Reiterating, Song et al. (<xref rid="B22" ref-type="bibr">2005</xref>) and Holmgren et al. (<xref rid="B6" ref-type="bibr">2003</xref>) noted an excess of bidirectional connectivity in layer V and layer II / III, respectively. However, Lefort et al. (<xref rid="B9" ref-type="bibr">2009</xref>) noted no such excess. It is possible that this could be a result of sampling from different parts of the cortex which exhibit significantly different micro-organization, or that small differences in sectioning size and sampling procedure could lead to such differences. It is this latter concern that we would like to emphasize.</p><p>We have not reproduced the sampling procedures used in these studies exactly, but rather provided a generic sampling simulation from which we can gain some qualitative insight into real-world experimental results. Examining the aforementioned studies, we note that Song et al. (<xref rid="B22" ref-type="bibr">2005</xref>) used a 300 μm slice (Sjöström et al., <xref rid="B21" ref-type="bibr">2001</xref>) with a roughly ellipsoid sampling area with radii of approximately 100 and 50 μm on the major and minor axes, respectively. Holmgren et al. (<xref rid="B6" ref-type="bibr">2003</xref>) also used a 300 μm slice, recording in an irregular shape out to a maximal radius of nearly 300 μm. Our model does not reproduce the high degree of excess bidirectional connections observed under these parameters, but it does result in an above-chance representation. Lefort et al. (<xref rid="B9" ref-type="bibr">2009</xref>), who noted no excess of bidirectional connections, used a 300 μm slice as well, further subdividing these into 100 μm sections, which would correspond to a centered recording radius of 50 μm—a radius at which our model does not exhibit a noteworthy excess of bidirectional connectivity, and suggesting an explanation for why their results appear potentially at odds with other cortical slice studies.</p><p>Our model demonstrating this concern is a simple graph model that, while it does not completely reproduce the nonrandom features noted in electrophysiological surveys, does reproduce some of them at a presumably natural scale. It is our belief that such a model provides a more reasonable, realistic, and general baseline for measuring the statistics of nonrandom cortical connectivity than a simple Erdős-Rényi graph. Certain observed complex features have been necessarily excluded to avoid an overly <italic>ad-hoc</italic> model. For example, our model does not reproduce the common neighbor clustering asymmetry in the in- and out-degree noted in the supplementary materials of Perin et al. (<xref rid="B15" ref-type="bibr">2011</xref>).</p><p>That the examined features depend so sensitively on section size in the presence of order 100 μm scale clustering should be both enlightening and concerning, particularly when most sampling procedures operate around this scale. Other factors such as neuronal type and local density almost certainly play into such effects as well. The model is not exhaustive, and numerous parameters, including the exact size and form of the connection probability profile and neuronal connection densities, could be varied. The thrust of the example provided in this paper is not to provide an exhaustive catalog of scenarios, but to demonstrate how sensitive the observed nonrandom effects of clustering mechanisms are to small variations in sampling. With this brief and simple demonstration in mind, the authors encourage experimenters to include all available information about neuronal and connection density and scale, as well as the full extent of exact sampling techniques in any study of such nonrandom features so that they can be best understood in the context of a complete graph.</p></sec><sec><title>Author contributions</title><p>Dr. Miner performed the programming, analysis, and initial writing. Research direction was shared. Significant background expertise and guidance was provided by Dr. Triesch, as was significant input into the writing and revision process.</p></sec><sec><title>Funding</title><p>This work was supported by the Quandt Foundation and the LOEWE-Program Neuronal Coordination Research Focus Frankfurt (NeFF).</p><sec><title>Conflict of interest statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec></sec>
Therapeutic Impact of Human Serum Albumin–Thioredoxin Fusion Protein on Influenza Virus-Induced Lung Injury Mice	<p>Reactive oxygen species (ROS) are the primary pathogenic molecules produced in viral lung infections. We previously reported on the use of a recombinant human serum albumin (HSA)–thioredoxin 1 (Trx) fusion protein (HSA–Trx) for extending the half-life Trx, an endogenous protein with anti-oxidant properties. As a result, it was possible to overcome the unfavorable pharmacokinetic and short pharmacological properties of Trx. We hypothesized that HSA–Trx would attenuate the enhanced ROS production of species such as hydroxyl radicals by neutrophils during an influenza viral infection. The levels of 8-hydroxy-2′-deoxyguanosine and 3-nitrotyrosine were used as indices of the anti-oxidant activity of HSA–Trx. In addition, the cytoprotective effects of HSA–Trx were examined in PR8 (H1N1) influenza virus-induced lung injured mice. The findings show that HSA–Trx reduced the number of total cells, neutrophils, and total protein in BALF of influenza virus-induced lung injured mice. The HSA–Trx treatment significantly decreased the level of 8-hydroxy-2′-deoxyguanosine and 3-nitrotyrosine, but failed to inhibit inducible nitric oxide synthase expression, in the lungs of the virus-infected mice. On the other hand, Tamiflu<sup>®</sup> treatment also significantly suppressed the production of inflammatory cells and neutrophil infiltration, as well as the protein level in BALF and lung histopathological alterations caused by the influenza virus. The suppressive effect of Tamiflu<sup>®</sup> was slightly stronger than that of HSA–Trx. Interestingly, Tamiflu<sup>®</sup> significantly decreased virus proliferation, while HSA–Trx had no effect. These results indicate that HSA–Trx may be of therapeutic value for the treatment of various acute inflammatory disorders such as influenza-virus-induced pneumonia, by inhibiting inflammatory-cell responses and suppressing the overproduction of NO in the lung.</p>	<contrib contrib-type="author"><name><surname>Tanaka</surname><given-names>Ryota</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="author-notes" rid="fn001"><sup>†</sup></xref></contrib><contrib contrib-type="author"><name><surname>Ishima</surname><given-names>Yu</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><xref ref-type="author-notes" rid="fn001"><sup>†</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/181471"/></contrib><contrib contrib-type="author"><name><surname>Enoki</surname><given-names>Yuki</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref></contrib><contrib contrib-type="author"><name><surname>Kimachi</surname><given-names>Kazuhiko</given-names></name><xref ref-type="aff" rid="aff3"><sup>3</sup></xref></contrib><contrib contrib-type="author"><name><surname>Shirai</surname><given-names>Tatsuya</given-names></name><xref ref-type="aff" rid="aff3"><sup>3</sup></xref></contrib><contrib contrib-type="author"><name><surname>Watanabe</surname><given-names>Hiroshi</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="aff" rid="aff2"><sup>2</sup></xref></contrib><contrib contrib-type="author"><name><surname>Chuang</surname><given-names>Victor T. G.</given-names></name><xref ref-type="aff" rid="aff4"><sup>4</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/190538"/></contrib><contrib contrib-type="author"><name><surname>Maruyama</surname><given-names>Toru</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><xref ref-type="corresp" rid="cor1"></xref></contrib><contrib contrib-type="author"><name><surname>Otagiri</surname><given-names>Masaki</given-names></name><xref ref-type="aff" rid="aff5"><sup>5</sup></xref><xref ref-type="aff" rid="aff6"><sup>6</sup></xref><xref ref-type="corresp" rid="cor1"></xref></contrib>	Frontiers in Immunology	<sec sec-type="introduction" id="S1"><title>Introduction</title><p>Influenza virus infections cause a broad array of illnesses, including acute lung injuries (ALI), which are responsible for vital morbidity and mortality both in children and adults (<xref rid="B1" ref-type="bibr">1</xref>). Although most influenza A viruses cause a transitory localized debilitating respiratory illness in human beings, more severe respiratory and systemic complications that are sometimes fatal can arise, as evidenced by the pandemic influenza of 1918 caused by the H1N1 virus and the highly pathogenic avian H5N1 virus (<xref rid="B2" ref-type="bibr">2</xref>). In addition, the novel swine-origin influenza virus H1N1 that spread globally in 2009 was an attenuated strain, but children and patients with a pre-existing disease developed more severe infections (<xref rid="B3" ref-type="bibr">3</xref>).</p><p>There are several, currently available antiviral therapeutic agents, including neuraminidase inhibitors, oseltamivir (Tamiful<sup>®</sup>) and zanamivir (Relenza<sup>®</sup>) to combat seasonal and pandemic influenza A virus outbreaks (<xref rid="B4" ref-type="bibr">4</xref>). These antiviral agents that prevent the release of the influenza virus from already infected cells and the transmission of the virus are clearly effective in the treatment of such infections. However, emerging evidence suggests that certain strains of the influenza virus are developing resistance to these antivirals, and sometimes cause severe ALI, requiring intensive care and mechanical ventilation in clinical settings (<xref rid="B5" ref-type="bibr">5</xref>). Therefore, the development of newer, more effective therapeutic approaches that differ from traditional virus based strategies is of great importance.</p><p>Recent studies have suggested that oxidative stress plays an important role in the pathogenesis and development of influenza-induced ALI (<xref rid="B6" ref-type="bibr">6</xref>). At the time of an influenza infection, several cells of the innate immune system, such as alveolar macrophages and neutrophils release reactive oxygen species (ROS) and reactive nitrogen species (RNOS), including as superoxide <inline-formula><mml:math id="M1"><mml:mrow><mml:mo class="MathClass-open">(</mml:mo><mml:mrow><mml:msubsup><mml:mrow><mml:mi mathvariant="normal">O</mml:mi></mml:mrow><mml:mrow><mml:mn>2</mml:mn></mml:mrow><mml:mrow><mml:mi>•</mml:mi><mml:mo class="MathClass-bin">−</mml:mo></mml:mrow></mml:msubsup></mml:mrow><mml:mo class="MathClass-close">)</mml:mo></mml:mrow></mml:math></inline-formula>, hydroxyl radical (<sup>⋅</sup>OH), nitric oxide (NO), and peroxynitrite (ONOO<sup>−</sup>) in order to combat the invading virus. However, the resulting overproduced ROS and RNOS not only exert a direct cytopathic effect on viral replication in the infected cells but also induced the apoptotic cell death of non-infected cells. That ROS and RNOS play critical roles as mediators of virus-induced lung damage is supported studies in which exogenous anti-oxidants, such as superoxide dismutase (SOD), catalase (anti-oxidative enzymes), allopurinol (xanthine oxidase inhibitor), or <italic>N</italic>-monomethyl-<sc>l</sc>-arginine (NO synthase inhibitor) decreased the extent of lung damage and mortality in influenza-infected mice (<xref rid="B7" ref-type="bibr">7</xref>–<xref rid="B9" ref-type="bibr">9</xref>). In addition, the overexpression of extracellular SOD or heme oxygenase-1 was reported to suppress lung injury and inflammation, resulting in an improved survival rate (<xref rid="B10" ref-type="bibr">10</xref>, <xref rid="B11" ref-type="bibr">11</xref>).</p><p>Thioredoxin-1 (Trx) is a small redox-active protein (12 kDa) that is ubiquitously present in the human body and is one of the defense proteins that are induced in response to various oxidative stress conditions (<xref rid="B12" ref-type="bibr">12</xref>). In addition to its potent anti-oxidative effect, which is derived from dithiol-disulfide exchange in its active site, Trx also has anti-inflammatory properties, due mainly to its ability to inhibit neutrophil chemotaxis to inflammatory sites (<xref rid="B13" ref-type="bibr">13</xref>). Because of its desirable anti-oxidative and anti-inflammatory properties, Trx represents a new and potentially effective therapeutic agent for the treatment of influenza virus-induced ALI. However, since Trx is eliminated extensively via glomerular filtration, its plasma half-life is only about 1 h in mice and 2 h in rats, which is extremely short in terms of producing a significant therapeutic impact (<xref rid="B13" ref-type="bibr">13</xref>, <xref rid="B14" ref-type="bibr">14</xref>). In order to obtain a satisfactory therapeutic outcome, a sustainable therapeutic concentration of Trx would be needed. To achieve this, a constant infusion or frequent repeated administrations of Trx would be required. In fact, Yashiro et al. reported that exogenous recombinant Trx was effective in inhibiting influenza virus-induced lung damage, when administered intraperitoneally 2 day intervals (<xref rid="B15" ref-type="bibr">15</xref>).</p><p>In an attempt to increase the blood retention time of Trx, we recently produced a genetically engineered fusion protein comprises human serum albumin (HSA) and Trx (HSA–Trx) using a <italic>Pichia</italic> expression system (<xref rid="B16" ref-type="bibr">16</xref>). The incentives for targeting albumin are that it constitutes the most abundant serum protein in blood, and albumin has an extraordinary long half-life of 2–3 weeks in human beings. In addition to having a molecular size above the renal clearance threshold, the long half-lives are attributed to the efficient receptor-mediated recycling pathway involving the neonatal Fc receptor (FcRn) (<xref rid="B17" ref-type="bibr">17</xref>). Actually, the plasma half-life of the HSA–Trx fusion protein in normal mice was found to be similar to that of HSA, which is 10 times longer than the plasma half-life of Trx itself. Interestingly, HSA–Trx showed a higher distribution in the lungs than Trx (<xref rid="B16" ref-type="bibr">16</xref>). Therefore, a further attempt will be made to investigate the clinical usefulness of HSA–Trx in treating oxidative stress and inflammation related lung disorders.</p><p>The purpose of this study was to investigate the therapeutic impact of HSA–Trx in the treatment of influenza-induced ALI. Using an influenza-induced ALI mouse model, the results showed that HSA–Trx could prevent this disease via its long acting anti-oxidative and anti-chemotaxis effects.</p></sec><sec sec-type="materials\|methods" id="S2"><title>Materials and Methods</title><sec id="S2-1"><title>Experimental animals and reagents</title><p>Sea-ICR mice (5 weeks, male) were obtained from Kyudo Co., Ltd. (Saga, Japan). Blue Sepharose 6 Fast Flow, HiTrap Phenyl HP column and PD-10 desalting column were obtained from GE Healthcare (Tokyo, Japan). Chloral hydrate was obtained from Sigma (Tokyo, Japan). Diff-Quick reagents were purchased from Kokusai Shiyaku (Kobe, Japan). Coomassie Brilliant Blue solution and HistoVT One were obtained from Nacalai Tesque (Kyoto, Japan). Interferon-γ (IFN-γ) ELISA kit was purchased from Biolegend (San Diego, CA, USA). Mayer’s hematoxylin, a 1% eosin alcohol solution, mounting medium for histological examinations (malinol) were from Muto Pure Chemicals (Tokyo, Japan). 4′,6-Diamidino-2-phenylindole (DAPI) was obtained from Dojindo (Kumamoto, Japan). All other chemicals were of the highest analytical grades available.</p></sec><sec id="S2-2"><title>Production of HSA–Trx fusion protein</title><p>Human serum albumin, Trx, and the HSA–Trx fusion protein were produced following previously reported methods (<xref rid="B18" ref-type="bibr">18</xref>, <xref rid="B19" ref-type="bibr">19</xref>). Transformed <italic>Pichia pastoris</italic> cells were incubated in 1250 ml of BMGY liquid media [1% yeast extract, 2% pepton, 100 mM potassium phosphate (pH 6.0), 1.34% yeast nitrogen base with ammonium sulfate without amino acids, 4 × 10<sup>−5</sup>% biotin, 1% glycerol] (growth phase) for 2 days (OD<sub>600</sub> = 2), and was then cultured in 800 ml of BMMY media that contained a protein expression inducer as well as a carbon source, methanol [1% yeast extract, 2% pepton, 100 mM potassium phosphate (pH 6.0), 1.34% yeast nitrogen base with ammonium sulfate without amino acids, 4 × 10<sup>−5</sup>% biotin, 1% methanol] (protein induction phase) for 3 days at 30°C. Methanol was added at 12 h intervals to permit the concentration of methanol to be maintained at 1% in order to sustain the protein expression induction effect. Purification of the fusion protein was initially carried out by chromatography on a Blue Sepharose 6 Fast Flow column equilibrated with 200 mM sodium acetate buffer (pH 5.5) after dialysis against the same buffer. Using AKTA prime, a 5 ml HiTrap Phenyl HP column for hydrophobic chromatography, the following conditions were employed: Buffer A, 50 mM Tris-HCl/1.5 M ammonium sulfate, pH 7.0; Buffer B, 50 mM Tris-HCl, pH 7.0; Gradient, 0–100% (Buffer B) 100 ml; Flow rate, 3 ml/min. The fusion protein was analyzed by SDS-PAGE on a 10% polyacrylamide gel, with Coomassie blue R250 staining. The purity of the fusion protein was estimated to be in excess of 97%.</p></sec><sec id="S2-3"><title>Production of influenza-induced ALI mice model</title><p>Influenza virus A/Puerto Rico/8/34 (H1N1) was used throughout the experiments. All animal experiments were conducted in accordance with the guidelines of Center for Animal Resources and Development, Kumamoto University for the care and use of laboratory animals. All animal experiments were approved by the experimental animal ethics committee at the Kumamoto University (B 26-058). Influenza-induced ALI model mice were produced by intratracheal injection of influenza virus suspension diluted with LB medium at a dose of 1.5 × LD<sub>50</sub> under anesthesia with chloral hydrate (500 mg/kg) on day 0. Either Trx or HSA–Trx was (but not both) administered intravenously (3.5 nmol protein in 200 μl saline/mouse) via the mice tail vein at 4 and 6 days after virus infection. Tamiful<sup>®</sup> was administrated intraperitoneally (10 mg/kg/day, 2 times/day) for 4 days from 2 days after virus infection. The efficacy of therapeutic agents was evaluated with mice sacrificed at 8 days after virus infection.</p></sec><sec id="S2-4"><title>Analysis of lung lavage samples</title><p>The mice were anesthetized with diethyl ether, the chests were opened, and blood was sampled for measurement of hydroperoxides (described below). BALF was collected by cannulating the trachea and lavaging the lung with 1 ml of sterile PBS containing 50 U/ml heparin (two times). About 1.8 ml of BALF was routinely recovered from each animal. The BALF was centrifuged at 4,100 × <italic>g</italic> for 5 min at 4°C to separate the cells in the BALF from the liquid. Cells were suspended in 0.9% NaCl and the resulting lysate was centrifuged again. From the recovered cells, the total cell number was counted using a hemocytometer. Cells were stained with Diff-Quick reagents, and the ratios of alveolar macrophages, neutrophils, and lymphocytes to total cells were determined. More than 200 cells were counted for each sample. The protein concentration in BALF was measured with the Protein Assay Coomassie Brilliant Blue solution with BSA as a standard. IFN-γ levels in BALF were measured by ELISA, following the manufacturer’s suggested protocol.</p></sec><sec id="S2-5"><title>Histological examination of lung tissue</title><p>For the histological analysis, the recovered lung tissues were fixed in 10% neutral-buffered formalin (Wako, Osaka, Japan), embedded in paraffin (Sakura Finetek Japan, Tokyo, Japan) and sectioned at 4-μm thickness. The lung sections were subjected to hematoxylin and eosin (HE) staining for morphologic analysis, and immunohistochemistry for inducible nitric oxide synthase (iNOS) (N-20) (Santa Cruz Biotechnology, cat#: sc-651, Santa Cruz, CA, USA), 8-hydroxy-2′-deoxygenase (8-OHdG) (15A3) (Santa Cruz Biotechnology, cat#: sc-66036, Santa Cruz, CA, USA), or 3-nitrotyrosine (NO<sub>2</sub>-Tyr) (Millipore, cat#: AB5411, Billerica, MA, USA). For the immunohistochemistry of iNOS, 8-OHdG, or NO<sub>2</sub>-Tyr, antigen retrieval was first performed by means of HistoVT One. A solution containing 50 mM Tris-HCl (pH 7.4) and 0.1% Tween-20 (T-TB) was then used to solubilize the hepatic slices, followed by blocking with Block Ace (Dainippon Pharmaceutical, Osaka, Japan) at 25°C for 15 min. Next, reaction with the primary antibody was carried out overnight at a temperature below 4°C. In addition, the primary antibodies against iNOS, 8-OHdG, or NO<sub>2</sub>-Tyr were diluted (1:50) with 0.5% BSA in PBS before use and conjugated with DAPI solution (10 μg/ml). The hepatic slices were then washed with T-TB, followed by reaction with the secondary antibody at 25°C for 1.5 h. For iNOS and NO<sub>2</sub>-Tyr immunostaining, Alexa Fluor 488 goat anti-rabbit IgG (H + L) (Invitrogen, cat#: A-11034, Tokyo, Japan) and for 8-OHdG immunostaining, Alexa Fluor 555 goat anti-mouse IgG (H + L) (Invitrogen, cat#: A-21424, Tokyo, Japan) were diluted (1:200) with 0.5% BSA in PBS before use. After the reaction, the slide was observed under a microscope (Keyence, BZ-8000, Osaka, Japan). Image analyses of the extent and intensity of iNOS, 8-OHdG and NO<sub>2</sub>-Tyr immunostaining were also performed using the image J software.</p></sec><sec id="S2-6"><title>Measurement of plasma hydroperoxides</title><p>Plasma was prepared from recovered blood sample. The plasma concentration of hydroperoxides (whole oxidant capacity of plasma against <italic>N</italic>,<italic>N</italic>-diethylparaphenylene-diamine in acidic buffer) was measured using the FREE <italic>carpe diem</italic> (Diacron International, Grosseto, Italy) following the manufacturer’s recommended protocols. The measurement unit was CARR U. It has been established that 1 CARR U corresponds to 0.08 mg/dl hydrogen peroxide.</p></sec><sec id="S2-7"><title>Measurement of virus load in lung tissue</title><p>The left lungs from influenza virus-infected mice were removed, weighed, and homogenized in 1 ml of RPMI medium 1640. After centrifugation at 100 × <italic>g</italic> for 10 min at 4°C, the supernatant was recovered. Virus titers were determined by a plaque assay on Madin–Darby canine kidney (MDCK) cells. MDCK cells were seeded on a 24 well culture plate and cultured until reaching confluency. Cells were washed with PBS, and then 0.1-ml amounts of each dilution were inoculated to cultured cell followed by incubation for 1 h at 34°C. DMEM (0.5 ml) containing 100 U/ml penicillin, 100 μg/ml streptomycin, and 0.5 μg/ml fungizone was added to each well and incubated at 37°C for 3 days in a humidified incubator with 5% CO<sub>2</sub>. Cytopathological effect was calculated and 50% tissue culture infectious dose (TCID50) was calculated using the formula developed by Reed and Muench (<xref rid="B20" ref-type="bibr">20</xref>).</p></sec><sec id="S2-8"><title>Preparation of fluorescein isothiocianate-labeled HSA–Trx and evaluation of distribution to BALF of FITC-labeled HSA–Trx in influenza-infected mice</title><p>Human serum albumin–Trx was labeled with fluorescein isothiocianate (FITC). HSA–Trx (4 mg/ml) and FITC (1 mg/ml) were dissolved in 0.15 M K<sub>2</sub>HPO<sub>4</sub> (pH 9.5), followed by mixing for 4 h at room temperature. The resulting solution was desalted by passage through a PD-10 desalting column and eluent was concentrated by Vivapore (Sartorius Stedim Biotech, A.S., France). FITC-labeled HSA–Trx (3.5 nmol/mouse) was administrated intravenously to normal mice or model mice at 4 or 6 days after virus infection. At 2 h after its administration, the mice were anesthetized with diethyl ether, the chests were opened, and blood was drained. BALF was collected by cannulating the trachea and lavaging the lung with 1 ml of sterile PBS containing 50 U/ml heparin. About 0.8 ml of BALF was routinely recovered from each animal. The BALF was centrifuged at 4,100 × <italic>g</italic> for 5 min at 4°C to separate the cells in the BALF from the liquid. The BALF levels of FITC-labeled HSA–Trx were quantified by using a spectral photometer (JASCO Corporation, MODEL: PTL-3965, Tokyo, Japan).</p></sec><sec id="S2-9"><title>Statistical analysis</title><p>All data are expressed as the mean ± SE. Significant differences among each group were examined using a one-way of analysis of variance (ANOVA) followed by Bonferroni multiple comparison. A probability value of <italic>P</italic> < 0.05 was considered to indicate statistical significance.</p></sec></sec><sec id="S3"><title>Results and Discussion</title><sec id="S3-10"><title>Effect of HSA–Trx on body weight and survival of influenza-infected mice</title><p>Acute lung injury was induced in mice by a single intratracheal administration of influenza virus at a dose of 1.5 × LD<sub>50</sub> (day 0). Since Yashiro et al. reported that the multiple administration of Trx (3.5 nmol/body) prevented influenza-induced ALI (<xref rid="B15" ref-type="bibr">15</xref>), we adopted a dose of HSA–Trx equivalent to one Trx treatment (3.5 nmol/body). In Figure S1 in Supplementary Material, the time-dependent variation following the virus infection shows that the infiltration of inflammatory cells, especially neutrophils in BALF, lung edema, and injury was induced from day 4 to 6 after the initial virus infection. In addition, as shown in Figure S1D in Supplementary Material, the virus load after influenza infection reached a peak at day 4, and then decreased. Therefore, HSA–Trx was injected at days 4 and 6 after the virus infection (Figure <xref ref-type="fig" rid="F1">1</xref>).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Experimental protocol for the effective evaluation of HSA–Trx, Trx, or Tamiflu<sup>®</sup> on influenza-induced ALI model mice</bold>. Either HSA–Trx or Trx was administrated intravenously (1.75 nmol/mouse) at 4 and 6 days after virus infection. Tamiflu<sup>®</sup> was administrated intraperitoneally (10 mg/kg/day, 2 times/day) for a total of 4 days, starting 2 days after the virus infection. The efficacy of therapeutic agents was evaluated with mice sacrificed at 8 days after virus infection.</p></caption><graphic xlink:href="fimmu-05-00561-g001"/></fig><p>As shown in Figure <xref ref-type="fig" rid="F2">2</xref>, the body weight of the saline treated group mice decreased from day 6 after the virus infection, with 80% of them dead. On the other hand, HSA–Trx administration significantly suppressed the diminished body weight and increased the survival rate in virus-infected mice. Finally, 70% of the HSA–Trx treated mice were still alive at 14 days after the virus infection.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Effect of HSA–Trx on the body weight (A) and survival (B) of influenza-infected mice</bold>. HSA–Trx (3.5 nmol/mouse) was administrated intravenously at 4 and 6 days after the virus infection. *<italic>P</italic> < 0.01 or <italic>P</italic> < 0.05 as compared with saline (<italic>n</italic> = 10).</p></caption><graphic xlink:href="fimmu-05-00561-g002"/></fig></sec><sec id="S3-11"><title>Relative inhibitory effect of HSA–Trx or Trx on influenza-infected mice</title><p>We compared HSA–Trx vs. Trx for their protective effect at the same administration schedule against influenza-induced ALI mice (Figure <xref ref-type="fig" rid="F1">1</xref>). The influenza infection caused a significant induction in the infiltration of inflammatory cells, especially neutrophils in the BALF of infected mice on day 8 (Supporting Figure in Supplementary Material). The number of total cells and neutrophils in the BALF remained significantly lower in HSA–Trx treated mice compared with saline or Trx treated mice (Figures <xref ref-type="fig" rid="F3">3</xref>A,B).</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Relative inhibitory effect of HSA–Trx or Trx on influenza-infected mice</bold>. The same injected dose (3.5 nmol/mouse) of HSA–Trx or HSA was administered intravenously at 4 and 6 days after the virus infection. The numbers of <bold>(A)</bold> total cells and <bold>(B)</bold> neutrophils, or <bold>(C)</bold> protein concentration in BALF were determined at 8 days after virus infection. <bold>(D)</bold> Sections of pulmonary tissue were prepared at 8 days after virus infection, and subjected to histopathological examination (HE staining). Magnifications: ×40 in <bold>(D)</bold>; ×200 in <bold>(E)</bold>. Each value represents the mean ± SE (<italic>n</italic> = 5).</p></caption><graphic xlink:href="fimmu-05-00561-g003"/></fig><p>The extent of acute lung damage elicited by the influenza virus infection was assessed based on the increase in total protein in the BALF. We compared the levels of protein in the BALF from the saline, HSA–Trx, or Trx treated mice. As shown in Figure <xref ref-type="fig" rid="F3">3</xref>C, HSA–Trx significantly suppressed the elevation in the BALF protein that was induced by the influenza virus. The Trx treatment failed to result in a significant reduction in the virus-induced elevation of protein concentration in the BALF (Figure <xref ref-type="fig" rid="F3">3</xref>C).</p><p>Hematoxylin and eosin staining of lung sections indicated that the virus-inoculated mice presented diffuse edema and inflammatory infiltration in the alveoli and interstitium of the lung, hemorrhaging, and thickened airways on day 8 (Supporting Figure in Supplementary Material). However, in lung sections from the HSA–Trx administration group, the extent of damage was attenuated, compared with the saline treated group. Histopathological findings for lung sections from the Trx administration group were similar to that for the saline administration group (Figures <xref ref-type="fig" rid="F3">3</xref>D,E). This tendency was similar to the previous findings for OVA or BLM-induced lung injuries, in that HSA–Trx ameliorated these lung injuries in an administration schedule in which Trx failed to show any efficacy against these injuries (<xref rid="B21" ref-type="bibr">21</xref>, <xref rid="B22" ref-type="bibr">22</xref>). Therefore, the suppressive effect of HSA–Trx may be due to the extended retention of the fusion protein in the blood, compared to HSA via FcRn.</p></sec><sec id="S3-12"><title>Effect of HSA–Trx on BALF IFN-γ and pulmonary iNOS expression</title><p>A previous report indicated that a large amount of IFN-γ was induced in BALF obtained from influenza virus infections, and IFN-γ up-regulated the expression of iNOS in alveolar macrophages, bronchial, and alveolar epithelial cells (<xref rid="B9" ref-type="bibr">9</xref>, <xref rid="B23" ref-type="bibr">23</xref>). To reveal the mechanism underlying the suppressive effect of HSA–Trx on influenza-induced ALI, the levels of IFN-γ in BALF or the expression of iNOS in lung tissue was confirmed by ELISA or immunostaining, respectively. As shown in Figure <xref ref-type="fig" rid="F4">4</xref>, the administration of HSA–Trx failed to reduce the increased levels of expressed IFN-γ and iNOS that were produced as the result of a viral infection.</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>Effect of HSA–Trx on BALF IFN-γ and pulmonary iNOS expression in influenza-infected mice</bold>. <bold>(A)</bold> IFN-γ levels in BALF were determined, and <bold>(B)</bold> immunostaining of lung slice for iNOS/DAPI was performed at 8 days after the influenza infection. <bold>(C)</bold> Image analysis of the extent and intensity of iNOS/DAPI staining was performed. Each bar represents the mean ± SE [<bold>(A)</bold> <italic>n</italic> = 5, <bold>(C)</bold> <italic>n</italic> = 4]. NS; non-significant vs. saline.</p></caption><graphic xlink:href="fimmu-05-00561-g004"/></fig></sec><sec id="S3-13"><title>Effect of HSA–Trx on oxidative stress in lung tissue</title><p>Previous reported findings suggest that ROS released from activated leukocytes, especially alveolar macrophages and neutrophils, are associated with the development of influenza-induced ALI (<xref rid="B6" ref-type="bibr">6</xref>). In fact, the genetic knockout of NADPH-oxidase in which neutrophils are produced, resulted in the suppression of influenza-induced ALI (<xref rid="B24" ref-type="bibr">24</xref>). In addition, NO released by iNOS reacts with <inline-formula><mml:math id="M2"><mml:msubsup><mml:mrow><mml:mi mathvariant="normal">O</mml:mi></mml:mrow><mml:mrow><mml:mn>2</mml:mn></mml:mrow><mml:mrow><mml:mi>•</mml:mi><mml:mo class="MathClass-bin">−</mml:mo></mml:mrow></mml:msubsup></mml:math></inline-formula> derived from activated leukocytes to produce peroxynitrite (ONOO<sup>−</sup>) that has a strong injurious activity. ONOO<sup>−</sup> can nitrate a wide range of biological targets, such as proteins and fatty acids to produce NO<sub>2</sub>-Tyr adducts (<xref rid="B25" ref-type="bibr">25</xref>).</p><p>To evaluate the effect of HSA–Trx on the oxidative stress induced by a virus infection, lung tissue was immunostained for 8-OHdG and NO<sub>2</sub>-Tyr, an oxidized product of nucleic acids and proteins, respectively, on day 8 after the influenza infection. Additionally, to determine the levels of hydroperoxide in plasma, dROMs tests were performed. As shown in Figure <xref ref-type="fig" rid="F5">5</xref>, the expression of 8-OHdG and NO<sub>2</sub>-Tyr in lung tissue and plasma hydroperoxides levels were increased in influenza-infected mice compared to normal mice, while HSA–Trx clearly suppressed the levels of these oxidative stress markers. Our previous report suggested that HSA–Trx has direct scavenging activities against <inline-formula><mml:math id="M3"><mml:mrow><mml:mo class="MathClass-open">(</mml:mo><mml:mrow><mml:msubsup><mml:mrow><mml:mi mathvariant="normal">O</mml:mi></mml:mrow><mml:mrow><mml:mn>2</mml:mn></mml:mrow><mml:mrow><mml:mi>•</mml:mi><mml:mo class="MathClass-bin">−</mml:mo></mml:mrow></mml:msubsup></mml:mrow><mml:mo class="MathClass-close">)</mml:mo></mml:mrow></mml:math></inline-formula> derived from activated neutrophils and that this effect is concentration dependent (<xref rid="B22" ref-type="bibr">22</xref>). Therefore, the inhibitory effect of oxidized products that accumulated was thought to involve the same mechanism as proposed in a previous report. Furthermore, Trx has anti-chemotaxis properties for neutrophils by inhibiting the activation of p38 mitogen-activated protein kinase, the downregulation of L-selectin (CD62L) and adhesion to endothelial cells (<xref rid="B13" ref-type="bibr">13</xref>). In fact, as shown in Figure <xref ref-type="fig" rid="F3">3</xref>B, HSA–Trx significantly lowered the number of neutrophils in the BALF of influenza-infected mice. Taken together, the mechanism underlying the improving effect of HSA–Trx on influenza-induced ALI could be not only attributed exclusively to an anti-oxidative effect but also anti-chemotaxis appears to be involved.</p><fig id="F5" position="float"><label>Figure 5</label><caption><p><bold>Effect of HSA–Trx on pulmonary and plasma oxidative stress in influenza-infected mice</bold>. Immunostaining of lung slice for <bold>(A)</bold> 8-OHdG/DAPI and <bold>(B)</bold> NO<sub>2</sub>-Tyr/DAPI was performed at 8 days after the influenza infection. Image analysis of the extent and intensity of <bold>(C)</bold> 8-OHdG/DAPI and <bold>(D)</bold> NO<sub>2</sub>-Tyr/DAPI staining was performed. <bold>(E)</bold> Plasma hydroperoxide level was determined at 8 days after influenza infection. Each bar represents the mean ± SE (<italic>n</italic> = 4).</p></caption><graphic xlink:href="fimmu-05-00561-g005"/></fig></sec><sec id="S3-14"><title>Relative inhibitory effect of HSA–Trx or Tamiflu<sup>®</sup> on influenza-infected mice</title><p>For future clinical applications, the inhibitory effect of HSA–Trx against influenza-induced ALI was examined. A previous study demonstrated that the intraperitoneal administration of Tamiflu<sup>®</sup> (5 mg/kg) daily at two times per day showed a protective effect for influenza-induced ALI (<xref rid="B26" ref-type="bibr">26</xref>). Hence, the inhibitory effect of HSA–Trx (3.5 nmol/mouse, i.v.) at 4 and 6 days against influenza-induced ALI was compared with that of Tamiflu<sup>®</sup> (5 mg/kg), which was administered every day at 2 times per day for 4 days from 2 day after the influenza infection (Figure <xref ref-type="fig" rid="F1">1</xref>). The evaluation items were the same as Figure <xref ref-type="fig" rid="F3">3</xref>. As shown in Figure <xref ref-type="fig" rid="F6">6</xref>, Tamiflu<sup>®</sup> also significantly suppressed the derivation of inflammation cells and neutrophils infiltration, protein elevation in BALF and lung histopathological alteration caused by the influenza virus. The suppressive effect of Tamiflu<sup>®</sup> treatment group was slightly stronger than that of HSA–Trx.</p><fig id="F6" position="float"><label>Figure 6</label><caption><p><bold>Relative inhibitory effect of HSA–Trx or Tamiflu<sup>®</sup> on influenza-infected mice</bold>. HSA–Trx (3.5 nmol/mouse) was administered intravenously at 4 and 6 days after virus infection. Tamiflu<sup>®</sup> (5 mg/kg) was administrated intraperitonealy every day at 2 times per day for a total of 4 days starting on 2 day after the influenza infection. The numbers of <bold>(A)</bold> total cells and <bold>(B)</bold> neutrophils, or <bold>(C)</bold> protein concentration in BALF were determined at 8 days after virus infection. <bold>(D)</bold> Sections of pulmonary tissue were prepared at 8 days after virus infection, and subjected to histopathological examination (HE staining). Magnifications: ×40 in <bold>(D)</bold>; ×200 in <bold>(E)</bold>. Each value represents the mean ± SE (<italic>n</italic> = 5). <italic>P</italic> < 0.05 and <italic>P</italic> < 0.01 as compared with saline. <sup>#</sup><italic>P</italic> < 0.05 as compared with HSA–Trx.</p></caption><graphic xlink:href="fimmu-05-00561-g006"/></fig></sec><sec id="S3-15"><title>Effect of HSA–Trx or Tamiflu<sup>®</sup> on viral titers</title><p>To validate that HSA–Trx treatment did not inhibit viral replication, we performed a plaque-forming assay in MDCK cells using the lung supernatant recovered from mice at day 4 or 6 after the virus infection. As shown in Figure <xref ref-type="fig" rid="F7">7</xref>, although the Tamiflu<sup>®</sup> treatment markedly inhibited virus proliferation, no significant difference between the lung viral titers in the Saline and HSA–Trx administration mice was found. These results indicate that HSA–Trx, unlike a Tamiflu<sup>®</sup>, treatment ameliorates the acute lung damage induced by the virus infection without affecting viral propagation in the lungs of these mice.</p><fig id="F7" position="float"><label>Figure 7</label><caption><p><bold>Effect of HSA–Trx or Tamiflu<sup>®</sup> on pulmonary virus replication in influenza-infected mice</bold>. HSA–Trx (3.5 nmol/mouse) was administered intravenously at 4 and 6 days after the virus infection. Tamiflu<sup>®</sup> (5 mg/kg) was administrated intraperitonealy every day at 2 times per day for a total of 4 days starting on 2 day after the influenza infection. The virus titles in lung tissue were determined at 4 and 6 days after the virus injection. Each value represents the mean ± SE (<italic>n</italic> = 5).</p></caption><graphic xlink:href="fimmu-05-00561-g007"/></fig></sec><sec id="S3-16"><title>BALF distribution of FITC-labeled HSA–Trx in influenza-infected mice</title><p>The concentration of albumin in alveolar fluid is usually much lower than that in the blood, whereas the concentration can increase to 75–95% of the plasma level in the case of a lung injury due to vascular hyperpermeability caused by a marked elevation in the surfactant’s surface tension (<xref rid="B27" ref-type="bibr">27</xref>). In fact, the albumin concentration in BALF is used as a bio-marker of lung injuries in clinical settings. Thus, it would be predicted that the BALF distribution of exogenous HSA–Trx in influenza-infected mice would be higher than that in influenza-non-infected mice. Hence, FITC-labeled HSA–Trx was prepared, and the distribution to BALF at 2 h after its administration in control mice (Day 0) or model mice on day 4 (Day 4) or 6 (Day 6) after the virus infection was evaluated. As shown in Figure <xref ref-type="fig" rid="F8">8</xref>, the BALF distribution of FITC-labeled HSA–Trx at Day 6 was approximately 4 times higher than that at Day 0. This result suggests that the distribution of HSA–Trx to the site of the lesion, the interstitium in influenza-infected patients may be higher than that in healthy subjects.</p><fig id="F8" position="float"><label>Figure 8</label><caption><p><bold>BALF distribution of FITC-labeled HSA–Trx in influenza-infected mice</bold>. BALF levels of FITC-labeled HSA–Trx after intravenous administration into the tail vein of normal (Day 0) or influenza-infected (Day 4 or 6) mice were determined using fluorescence spectroscopy. Each point represents the mean ± SE (<italic>n</italic> = 4). *<italic>P</italic> < 0.01 vs. Day 0.</p></caption><graphic xlink:href="fimmu-05-00561-g008"/></fig></sec></sec><sec id="S4"><title>Conclusion</title><p>We evaluated the therapeutic effects of HSA–Trx, a long-lasting anti-oxidant and anti-inflammatory modulator that acts via the efficient receptor-mediated recycling pathway involving FcRn, on influenza-induced ALI model mice, and reached four major findings; (1) The post-administration of HSA–Trx suppressed influenza-induced ALI, (2) The mechanism by which HSA–Trx inhibited influenza-induced ALI can be attributed to a combination of anti-oxidative and anti-chemotaxis effects, (3) Unlike Tamiflu<sup>®</sup> the HSA–Trx treatment ameliorated acute lung damage induced by the virus infection without affecting viral propagation in the lung of these mice, (4) The distribution of HSA–Trx to the lesion site, namely the interstitium in influenza-infected patients may be higher than that in healthy subjects. Thus, we conclude that HSA–Trx has considerable potential for use as a therapeutic agent for the treatment of various acute inflammatory disorders such as influenza-virus-induced pneumonia.</p></sec><sec id="S5"><title>Conflict of Interest Statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec><sec sec-type="supplementary-material" id="S6"><title>Supplementary Material</title><p>The Supplementary Material for this article can be found online at <uri xlink:type="simple" xlink:href="http://www.frontiersin.org/Journal/10.3389/fimmu.2014.00561/abstract">http://www.frontiersin.org/Journal/10.3389/fimmu.2014.00561/abstract</uri></p><supplementary-material content-type="local-data" id="SM1"><media xlink:href="Data_Sheet_1.DOCX"><caption><p>Click here for additional data file.</p></caption></media></supplementary-material></sec>
Regulatory T Cell-Derived Exosomes: Possible Therapeutic and Diagnostic Tools in Transplantation	<p>Exosomes are extracellular vesicles released by many cells of the body. These small vesicles play an important part in intercellular communication both in the local environment and systemically, facilitating in the transfer of proteins, cytokines as well as miRNA between cells. The observation that exosomes isolated from immune cells such as dendritic cells (DCs) modulate the immune response has paved the way for these structures to be considered as potential immunotherapeutic reagents. Indeed, clinical trials using DC derived exosomes to facilitate immune responses to specific cancer antigens are now underway. Exosomes can also have a negative effect on the immune response and exosomes isolated from regulatory T cells (Tregs) and other subsets of T cells have been shown to have immune suppressive capacities. Here, we review what is currently known about Treg derived exosomes and their contribution to immune regulation, as well as highlighting their possible therapeutic potential for preventing graft rejection, and use as diagnostic tools to assess transplant outcome.</p>	<contrib contrib-type="author"><name><surname>Agarwal</surname><given-names>Akansha</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="author-notes" rid="fn001"><sup>†</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/174582"/></contrib><contrib contrib-type="author"><name><surname>Fanelli</surname><given-names>Giorgia</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="author-notes" rid="fn001"><sup>†</sup></xref></contrib><contrib contrib-type="author"><name><surname>Letizia</surname><given-names>Marilena</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="author-notes" rid="fn001"><sup>†</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/190233"/></contrib><contrib contrib-type="author"><name><surname>Tung</surname><given-names>Sim Lai</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref></contrib><contrib contrib-type="author"><name><surname>Boardman</surname><given-names>Dominic</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/190097"/></contrib><contrib contrib-type="author"><name><surname>Lechler</surname><given-names>Robert</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref></contrib><contrib contrib-type="author"><name><surname>Lombardi</surname><given-names>Giovanna</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/23690"/></contrib><contrib contrib-type="author"><name><surname>Smyth</surname><given-names>Lesley A.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="corresp" rid="cor1">*</xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/41483"/></contrib>	Frontiers in Immunology	<sec id="S1"><title>Treg Exosomes – Immune Modulators</title><p>Exosomes are small, cup-shaped, secreted membrane vesicles (approximately 50–100 nM in diameter) that are formed by the inward budding of endosomal membranes (<xref rid="B1" ref-type="bibr">1</xref>–<xref rid="B6" ref-type="bibr">6</xref>). Exosomes are released into the extracellular environment following the fusion of multivesicular endosomes with the plasma membrane (<xref rid="B7" ref-type="bibr">7</xref>). Several proteins involved in their biogenesis and release have been described and have recently been reviewed by Colombo et al. (<xref rid="B7" ref-type="bibr">7</xref>). Exosomes released by many immune and non-immune cells have been shown to have a range of physiological properties within the immune system. These include antigen presentation, immune regulation, and programed cell death, each of which is linked to the cell from which they are released (<xref rid="B6" ref-type="bibr">6</xref>, <xref rid="B7" ref-type="bibr">7</xref>). They play an important role in intercellular communication and can act as shuttles for transferring proteins, miRNA, mRNA, and cytokines from one cell to another (<xref rid="B8" ref-type="bibr">8</xref>).</p><p>Many cells of the body produce these extracellular vesicles (EVs) including those of the immune system such as CD4<sup>+</sup> and CD8<sup>+</sup> T cells, B cells, and dendritic cells (DCs). Exosomes from these cells have been shown to mediate either immune stimulation (DCs) or immune modulation (T cells) (<xref rid="B9" ref-type="bibr">9</xref>–<xref rid="B14" ref-type="bibr">14</xref>). Recently, the release of exosomes by murine CD4<sup>+</sup>CD25<sup>+</sup>Foxp3<sup>+</sup> regulatory T cells (Tregs), following TCR activation, was shown, initially by Smyth et al. (<xref rid="B15" ref-type="bibr">15</xref>) and later by Okoye et al. (<xref rid="B16" ref-type="bibr">16</xref>). In addition to CD4<sup>+</sup>CD25<sup>+</sup>Foxp3<sup>+</sup> cells, other murine T cells with regulatory capacities were found to also release exosomes following activation. Bryniarski et al. observed that “exosome like” particles were present in the supernatants of cultured CD8<sup>+</sup> T cells with suppressive capacity (<xref rid="B17" ref-type="bibr">17</xref>), whilst Xie et al. observed that CD8<sup>+</sup>CD25<sup>+</sup>Foxp3<sup>+</sup> T cells secreted exosomes capable of inhibiting DC induced CD8<sup>+</sup> CTL responses (<xref rid="B18" ref-type="bibr">18</xref>).</p><p>Exosome production by murine CD4<sup>+</sup>CD25<sup>+</sup>Foxp3<sup>+</sup> Tregs appears to be quantitatively greater than other murine T cells, including naïve CD4<sup>+</sup> and CD8<sup>+</sup> T cells, T helper 1 (Th1), and Th17 cells, and is regulated by changes in intracellular calcium, hypoxia, and sphingolipids ceramide synthesis, as well as in the presence of IL-2 (<xref rid="B16" ref-type="bibr">16</xref>). Exosomes contribute significantly to the function of murine CD4<sup>+</sup>CD25<sup>+</sup>FoxP3<sup>+</sup> Tregs, inhibiting the release of exosomes reversed these cells suppressive capabilities (<xref rid="B16" ref-type="bibr">16</xref>). In parallel, murine Tregs exosomes were found to be immune modulatory. Reduced CD4<sup>+</sup> T cell proliferation and cytokine (IL-2 and IFNγ) release was observed in their presence <italic>in vitro</italic> (<xref rid="B15" ref-type="bibr">15</xref>). The suppressive nature of Treg exosomes, in one study, has been attributed to the ectoenzyme CD73 (<xref rid="B15" ref-type="bibr">15</xref>). The loss of CD73 on Treg exosomes reversed their suppressive nature. Expression of both CD39 and CD73 on Tregs contributes to immune suppression through the production of the anti-inflammatory mediator adenosine (<xref rid="B19" ref-type="bibr">19</xref>–<xref rid="B21" ref-type="bibr">21</xref>). Binding of this molecule to adenosine receptors A2aR, expressed by activated T effector cells (Teffs) triggers intracellular cAMP leading to the inhibition of cytokine production, thereby limiting T cell responses (<xref rid="B22" ref-type="bibr">22</xref>). Given that adenosine was produced following incubation of CD73 expressing Treg exosomes with exogenous 5′AMP it is feasible that the release of exosomes expressing CD73 within the local environment increases the surface area by which this membrane-associated enzyme, and ultimately Treg suppression, can function (<xref rid="B15" ref-type="bibr">15</xref>).</p><p>Several molecules associated with immune modulation including CD25 and CTLA-4, were also found on CD4<sup>+</sup>CD25<sup>+</sup>Foxp3<sup>+</sup> Treg exosomes (<xref rid="B15" ref-type="bibr">15</xref>). Nolte-’t Hoen et al. have previously shown that exosomes, derived from anergic rat T cells, inhibited Teffs responses following co-culture with B cells and DCs <italic>in vitro</italic> (<xref rid="B23" ref-type="bibr">23</xref>). These T cell-derived exosomes expressed high levels of CD25 and the authors suggested that CD25 expressing exosomes, binding to the surface of an antigen presenting cells (APC), bestows that cell with the ability to bind free IL-2 in the local environment leading to depletion of available cytokines and apoptosis of Teffs (<xref rid="B23" ref-type="bibr">23</xref>). Although CD25 expression was observed on Treg exosomes, this molecule may not play a role in their suppressive function given the observation that exosomes isolated from a T cell line, incapable of suppressing proliferation or cytokine production of CD4<sup>+</sup> T cells, in the presence of B cells, expressed similar levels of CD25 to Treg exosomes with regulatory function (<xref rid="B15" ref-type="bibr">15</xref>). A redundant role for CTLA-4 molecules has also been reported. Although present on Treg exosomes, blocking CTLA-4 did not modulate their suppressive function (<xref rid="B15" ref-type="bibr">15</xref>). So far, no molecules have been associated with the regulatory capacity of CD8<sup>+</sup>25<sup>+</sup>FoxP3<sup>+</sup> exosomes (<xref rid="B18" ref-type="bibr">18</xref>).</p><p>Recently, the transfer of miRNAs contained in T cell exosomes has been shown to affect the function of recipient APCs by inhibiting translation of target mRNA molecules (<xref rid="B14" ref-type="bibr">14</xref>, <xref rid="B24" ref-type="bibr">24</xref>). Likewise, the transfer of miRNAs, including Let-7d, miR-155, and Let-7b, to Teffs through the acquisition of CD4<sup>+</sup>CD25<sup>+</sup>Foxp3<sup>+</sup> Treg exosomes has been shown (<xref rid="B16" ref-type="bibr">16</xref>). Inhibiting Let-7d expression in Treg exosomes reversed the suppressive nature of these vesicles suggesting that miRNAs present in Treg exosomes may also play a role in their suppressive capacity (<xref rid="B16" ref-type="bibr">16</xref>). These findings confirm those of Bryniarski et al. (<xref rid="B17" ref-type="bibr">17</xref>) who observed the targeted delivery of an inhibitory miRNA, miR-150, to Teffs using exosomes isolated from CD8<sup>+</sup> T cells with suppressive capacity.</p><p>Several molecules present on exosomes isolated from Teffs, DCs, and B cells have been shown to have immune modulatory properties. Whether they also contribute to the suppressive nature of Treg exosomes has yet to be validated. For example, expression of FasL on murine CD8<sup>+</sup> T cell exosomes induced death of APCs (<xref rid="B12" ref-type="bibr">12</xref>, <xref rid="B25" ref-type="bibr">25</xref>), in addition, FasL-expressing exosomes isolated from DCs, genetically modified to express FasL, suppressed antigen-specific immune responses <italic>in vivo</italic> (<xref rid="B26" ref-type="bibr">26</xref>) and lastly, MHCII<sup>+</sup>FasL<sup>+</sup> exosomes constitutively produced by a human B cell-derived lymphoblastoid cell lines induced apoptosis in CD4<sup>+</sup> T cells (<xref rid="B27" ref-type="bibr">27</xref>). Murine and human CD4<sup>+</sup>25<sup>+</sup> Tregs express FasL (<xref rid="B28" ref-type="bibr">28</xref>). Whether FasL is expressed on Treg exosomes and contributes to the death of Teffs is yet to be tested. Other molecules, present on Tregs such as the inhibitory cell surface ligand programed cell death 1 ligand 1 (PDL-1) and Galectin-1 (<xref rid="B29" ref-type="bibr">29</xref>–<xref rid="B31" ref-type="bibr">31</xref>) may also be present on Treg exosomes. PDL-1 was found on mesenchymal stem cell EVs (<xref rid="B32" ref-type="bibr">32</xref>) and exosomes have been identified as transport vehicles for the secretion of molecules that lack a signal sequence such as Galectin-1 (<xref rid="B33" ref-type="bibr">33</xref>). Not only is this molecule highly expressed on Tregs it is essential for their function (<xref rid="B34" ref-type="bibr">34</xref>).</p><p>Regulatory T cells produce immune modulating cytokines such as IL-10, IL-35, and TGFβ (<xref rid="B35" ref-type="bibr">35</xref>). Presently, it is unknown whether these cytokines are contained in Treg exosomes however, expression of IL-10 and TGFβ in exosomes isolated from DCs, transduced to express these cytokines, has been shown (<xref rid="B36" ref-type="bibr">36</xref>, <xref rid="B37" ref-type="bibr">37</xref>) as has surface TGFβ on MSC derived EVs (<xref rid="B32" ref-type="bibr">32</xref>). Given the aforementioned it is a theoretical possibility, that Treg exosomes may contain one or more of these cytokines.</p></sec><sec id="S2"><title>Role of Treg Exosomes in Transplantation</title><sec id="S2-1"><title>Possible therapy?</title><p>In 1990, Hall et al. observed that the adoptive transfer of CD4<sup>+</sup>CD25<sup>+</sup> T cells resulted in long-term cardiac allograft survival in cyclosporine-treated rats (<xref rid="B38" ref-type="bibr">38</xref>). Since then this field of immunotherapy has been intensely studied in mouse (<xref rid="B39" ref-type="bibr">39</xref>–<xref rid="B41" ref-type="bibr">41</xref>), and recently in preclinical humanized mouse models (mice reconstituted with a human immune system and transplanted with human skin or human pancreatic islets of Langerhans) (<xref rid="B42" ref-type="bibr">42</xref>, <xref rid="B43" ref-type="bibr">43</xref>). In the latter, human CD4<sup>+</sup>25<sup>+</sup>Tregs, expanded with anti-CD3/28 antibody coated beads, have been found to prolong islet transplant survival and function (<xref rid="B42" ref-type="bibr">42</xref>, <xref rid="B44" ref-type="bibr">44</xref>). These positive outcomes have led to the application of humans Tregs for the prevention of graft versus host disease (GvHD) and to promote transplant tolerance (<xref rid="B45" ref-type="bibr">45</xref>–<xref rid="B48" ref-type="bibr">48</xref>). Currently, several organizations around the world are investigating the use of CD4<sup>+</sup>CD25<sup>+</sup> Tregs to promote “tolerance” to transplanted organs. At King’s College London, UK, phase I/II clinical trials are currently under way to test the safety and efficacy of using these cells in human kidney (One Study) and liver (ThRIL) transplant patients. Other clinical trials using human Tregs are also underway and are described elsewhere (<xref rid="B49" ref-type="bibr">49</xref>). Presently, we do not know the efficacy and efficiency of Tregs in these trials. Although Tregs are now being used in patients how they function <italic>in vivo</italic> is still unknown.</p><p>Given their immune modulatory capacity, the question arises, what is the contribution of Treg exosomes to transplant tolerance seen in the preclinical mouse models and can Treg exosomes be used <italic>in vivo</italic> as an alternative/or complementary therapy? At present, we are a long way away from using Treg exosomes in man given that the optimal Treg subset required to induce transplant tolerance is as yet unknown, as is whether they prolong graft survival in a patient setting. So why should we consider these EVs as a therapy? Several studies have suggested that inflammatory environments can subvert human Foxp3<sup>+</sup>Treg cell function by converting them to Teffs <italic>in vivo</italic> (<xref rid="B50" ref-type="bibr">50</xref>, <xref rid="B51" ref-type="bibr">51</xref>). However, unlike Foxp3<sup>+</sup> Tregs, adoptive transfer of human Treg exosomes are unlikely to be modified during inflammatory conditions <italic>in vivo</italic> (<xref rid="B1" ref-type="bibr">1</xref>) making them an ideal immune modulatory reagent (Figure <xref ref-type="fig" rid="F1">1</xref>A).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>A possible role for Treg exosome in transplantation</bold>. <bold>(A)</bold> Exosomes isolated from <italic>ex vivo</italic> expanded polyclonal or antigen-specific Tregs represent a potential adoptive therapy tool to promote transplant tolerance. Exosomes isolated from activated Tregs either alone or modified to express specific inhibitory miRNAs, chemicals, or cell surface molecules could be used in conjunction with Tregs to promote transplant tolerance. <bold>(B)</bold> Following transplantation, exosomes released from Tregs maybe used as a diagnostic tool to monitor activation and survival of Tregs <italic>in vivo</italic>. As Tregs release exosomes following activation, interaction with APCs expressing alloantigen on grafted tissue will result in exosome release. Identifying specific miRNAs expressed in Treg exosomes will help in their identification in blood or urine.</p></caption><graphic xlink:href="fimmu-05-00555-g001"/></fig><p>Several lines of evidence exist, some preliminary, some not, suggesting that studying these vesicles for this purpose is worthwhile, albeit challenging. So far, Yu et al. are the only group that have investigated the use of Treg exosomes as a therapy in a transplantation setting (<xref rid="B52" ref-type="bibr">52</xref>). These authors observed that the adoptive transfer of autologous rat Treg exosomes, post transplant, prolonged both survival and function of kidney allografts (<xref rid="B52" ref-type="bibr">52</xref>). Suggesting that Treg exosomes may represent an exciting new therapy for the induction of transplant tolerance.</p><p>Can this observation be translated into a human setting? Using preclinical methods to isolate and expand human Tregs, from peripheral blood of health individuals (<xref rid="B53" ref-type="bibr">53</xref>), we have successfully identified the release of CD63 and CD81 expressing exosomes from CD4<sup>+</sup>CD25<sup>hi</sup>Foxp3<sup>+</sup> suppressive human Tregs, following TCR activation (Agarwal et al., personal communication). Whether human Treg exosomes display molecules that can modulate the immune response <italic>in vivo</italic> is still being assessed. However, given that Jurkat CD4<sup>+</sup> T cells (a human T cell line) as well as human CD3<sup>+</sup> T cells, isolated from PBMCs, produce exosomes (<xref rid="B54" ref-type="bibr">54</xref>–<xref rid="B56" ref-type="bibr">56</xref>) containing molecules with potential immune regulatory effects, such as TCRs (<xref rid="B54" ref-type="bibr">54</xref>) and CTLA-4 (<xref rid="B56" ref-type="bibr">56</xref>) the possibility that human Treg exosomes contain immune regulatory molecules is very high.</p><p>Two phase I clinical trials using exosomes isolated from immature DCs have been conducted in advanced stage melanoma and MAGE-expressing non-small cell lung cancer patients (<xref rid="B57" ref-type="bibr">57</xref>–<xref rid="B59" ref-type="bibr">59</xref>). Despite a lack of antigen-specific T cell responses, stable disease was observed in some patients with tumor regression reported in one patient following treatment (<xref rid="B60" ref-type="bibr">60</xref>–<xref rid="B62" ref-type="bibr">62</xref>). These positive outcomes have paved the way for Phase II clinical trials using exosomes isolated from LPS or IFNγ activated DCs in non-small cell lung cancers. These studies have validated the efficacy and safety of exosomes as a therapy in man. In spite of these encouraging findings, several key limitations pertaining to the use of exosomes cannot be ignored. Firstly, at present there is no standardized protocol for isolating and analyzing “pure” exosomes (<xref rid="B7" ref-type="bibr">7</xref>). Contamination from other EVs as well as membrane free aggregates may be an issue depending on the isolation method used. Therefore, careful analyses of the purified exosomes will be required before administration. This will require the use of expensive equipment such as EM and Nanosight, which are not always readily available (<xref rid="B7" ref-type="bibr">7</xref>, <xref rid="B63" ref-type="bibr">63</xref>). Secondly, given that exosome release by Tregs is not constitutive and requires activation using anti-CD3/CD28 antibodies, the possibility that these antibodies contaminate Treg exosome preparations is as yet untested. Additionally contaminating molecules, for example, proteins/cytokines present in media, may pose a potential problem especially as exosomes will be isolated from culture supernatants. Thirdly, the quantity of exosomes isolated and the amount required for therapy purposes are at present unknown, as is whether large-scale production of Treg exosomes is actually possible. Lamparski et al. published that 1.8–5.8 mg of exosomes could be isolated from human monocyte derived DCs, expanded from peripheral blood leukopacks (originally containing 12–25 × 10<sup>9</sup> cells) higlighting the feasibility of large-scale production of DC exosomes (<xref rid="B64" ref-type="bibr">64</xref>). However, DCs produce these EVs vesicles constitutively making their production easier than those from Tregs, which are isolated only after activation (<xref rid="B65" ref-type="bibr">65</xref>). Yu et al. obtained 117 μg of exosomes from 4 × 10<sup>9</sup> freshly isolated rat Tregs, following activation, and the administration of 33 μg of exosomes, given over 3 time points, was sufficient to prolong the lifespan of a kidney transplant (<xref rid="B52" ref-type="bibr">52</xref>). Whether large quantities of pure exosomes can be isolated from human Tregs grown under GMP conditions is as yet unknown. Lastly, what happens to Treg exosomes <italic>in vivo</italic>, which cells acquire them and whether is it receptor driven is poorly understood. Recently, Teffs were shown to acquire Treg exosomes (<xref rid="B16" ref-type="bibr">16</xref>) whilst exosomes from EL4, a T cell lymphoma, have been shown to be preferential acquired by macrophages (<xref rid="B66" ref-type="bibr">66</xref>), perhaps via the CD169 pathway (<xref rid="B67" ref-type="bibr">67</xref>). Therefore, <italic>in vivo</italic> analysis of Treg exosomes is essential before they can be used in a clinical setting. Until all of these factors are addressed, using Treg exosomes in a transplant setting remains challenging and potential advantages remain at present theoretical.</p></sec><sec id="S2-2"><title>Diagnostic tool?</title><p>Biomarkers are quantitatively, measurable biological parameters that help indicate health and disease. The use of exosomes as biomarkers is a relatively new concept. Although it has not yet reached clinical practice, it is one area of exosome research that is rapidly expanding, with many clinical trials focusing on their use as a diagnostic tool, particularly for cancer (Table <xref ref-type="table" rid="T1">1</xref>). Several factors make exosomes suitable for this purpose, firstly, they travel through the bloodstream and can be isolated from plasma, serum, and urine (<xref rid="B68" ref-type="bibr">68</xref>, <xref rid="B69" ref-type="bibr">69</xref>). Secondly they receive surface markers from the cell from which they are derived, such that they can be identified and isolated. Lastly, they express unique miRNA and mRNA (Table <xref ref-type="table" rid="T1">1</xref>).</p><table-wrap id="T1" position="float"><label>Table 1</label><caption><p><bold>miRNAs present in exosomes isolated from the sera of patients with specific cancers or following immunization are being used as diagnostic biomarkers</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="left" rowspan="1" colspan="1">miRNA identified in exosomes</th><th align="left" rowspan="1" colspan="1">Cells origin</th><th align="center" rowspan="1" colspan="1">Reference</th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">miR-150</td><td align="left" rowspan="1" colspan="1">CD4<sup>+</sup> T cells</td><td align="center" rowspan="1" colspan="1">(<xref rid="B70" ref-type="bibr">70</xref>)</td></tr><tr><td align="left" rowspan="1" colspan="1">miR-21, miR-141, miR-200a, miR-200b, miR-200c, miR-203, miR-205, and miR-214</td><td align="left" rowspan="1" colspan="1">Ovarian cancer</td><td align="center" rowspan="1" colspan="1">(<xref rid="B71" ref-type="bibr">71</xref>)</td></tr><tr><td align="left" rowspan="1" colspan="1">miR-205, miR-19a, miR-19b, miR-30b, and miR-20a</td><td align="left" rowspan="1" colspan="1">Lung squamous cell carcinoma</td><td align="center" rowspan="1" colspan="1">(<xref rid="B72" ref-type="bibr">72</xref>)</td></tr><tr><td align="left" rowspan="1" colspan="1">let-7a, miR-1229, miR-1246, miR-150, miR-21, miR-223, and miR-23a</td><td align="left" rowspan="1" colspan="1">Colon cancer</td><td align="center" rowspan="1" colspan="1">(<xref rid="B73" ref-type="bibr">73</xref>)</td></tr><tr><td align="left" rowspan="1" colspan="1">hsa-miR-31, miR-185, and miR-34b</td><td align="left" rowspan="1" colspan="1">Melanoma</td><td align="center" rowspan="1" colspan="1">(<xref rid="B44" ref-type="bibr">44</xref>)</td></tr></tbody></table></table-wrap><p>Valadi et al. were the first group to publish that exosomes contained RNA (<xref rid="B8" ref-type="bibr">8</xref>). Exosome RNA is small, typically of about 200 bases in length and lacks the 18S and 28S RNA found in cells (<xref rid="B74" ref-type="bibr">74</xref>). Different RNA species including small ribosomal RNA, specific tRNA fragments, long interspersed elements, and long terminal repeats, have all been found in exosomes (<xref rid="B75" ref-type="bibr">75</xref>). Additionally, and as discussed earlier, there is also a selective enrichment of specific miRNAs into exosomes (<xref rid="B24" ref-type="bibr">24</xref>, <xref rid="B76" ref-type="bibr">76</xref>). The miRNA repertoire of an exosome is generally different to that of the parent cell, suggesting that exosome packaging is an active process (<xref rid="B14" ref-type="bibr">14</xref>). In T cells, for example, Rossi et al. identified a set of 20 miRNAs of which only 2 were differentially expressed in T<sub>H</sub> cell-derived exosomes (<xref rid="B77" ref-type="bibr">77</xref>). Upon activation primary CD4<sup>+</sup> T cells down-regulate their miRNA content. Some of these miRNAs accumulate in exosomes, for example, miR-150, suggesting that the cell may be shedding miRNA as part of a regulation step (<xref rid="B70" ref-type="bibr">70</xref>). de Candia et al. quantified the amount of miR-150 present in sera isolated from mice immunized with OVA plus an adjuvant, and reported an increased level of this miRNA in immunized mice as compared to non-immunized mice (<xref rid="B70" ref-type="bibr">70</xref>, <xref rid="B78" ref-type="bibr">78</xref>). When they removed CD4<sup>+</sup> T cells no elevated miR-150 levels were observed. They next validated this observation using sera collected from adults and children vaccinated with the 2009 pandemic flu (H1N1) vaccine. Similar to the mouse model, they observed that miR-150 was evident in the sera following vaccination, and that this miRNA was associated with lymphocyte derived exosomes. In addition, increased levels of miR-150 correlated with high antibody levels post vaccine, suggesting a link between activation of the adaptive immune responses and expression of a specific miRNAs in exosomes (<xref rid="B70" ref-type="bibr">70</xref>, <xref rid="B78" ref-type="bibr">78</xref>). From the adoptive cellular therapy point of view, this data is very exciting as it highlights the possibility of using exosomes to monitor cellular therapies such as Tregs <italic>in vivo</italic>. Given that Tregs produce exosomes only following activation, and in the case of transplantation this will be following recognition of alloantigen presented by donor and recipient DCs, it may be possible to assess Treg viability and function <italic>in vivo</italic> by monitoring Treg exosomes in the blood of transplant recipients. If this is possible Treg exosomes may be unique biomarkers for immune suppression (Figure <xref ref-type="fig" rid="F1">1</xref>B).</p><p>As mentioned earlier in addition to miRNA, mRNA, and proteins associated with exosomes can also act as diagnostic tools. For example, in patients with kidney disease CD2AP mRNA was associated with urinary exosomes (<xref rid="B79" ref-type="bibr">79</xref>). Several specific proteins have been identified in exosomes isolated from: (1) the urine of healthy individuals (CD24 and Aquaporin 2) (<xref rid="B80" ref-type="bibr">80</xref>), (2) sera from cancer patients (MUC1, LRG1, Hsp90a, and RAD21) (<xref rid="B81" ref-type="bibr">81</xref>), (3) the placenta (syncytin-1) (<xref rid="B82" ref-type="bibr">82</xref>), and 4) from patients with multiple sclerosis (IB4) (<xref rid="B83" ref-type="bibr">83</xref>). Taken together, these studies suggest the importance of validating the expression of mRNA and proteins, in addition to miRNAs, in Treg exosomes if unique biomarkers are to be identified.</p><p>In conclusion, at present Treg exosomes are still in their infancy with regard to transplantation, either as a therapy or a diagnostic tool. As outlined in this review, several key questions regarding their composition and function need to be addressed. In addition, better isolation and analysis protocols, as well as preclinical models are required before Treg exosomes can make the transition from the lab to the clinic, even for diagnostic purposes. Although some of the ideas presented here are speculative, pursuing the use of Treg exosomes for immune modulation and diagnostic purposes within a transplantation setting is timely given that clinical trials are now underway using Treg cells themselves.</p></sec></sec><sec id="S3"><title>Conflict of Interest Statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec>
Feasibility of the Medial Temporal lobe Atrophy index (MTAi) and derived methods for measuring atrophy of the medial temporal lobe	<p><bold>Introduction</bold>: The Medial Temporal-lobe Atrophy index (MTAi), 2D-Medial Temporal Atrophy (2D-MTA), yearly rate of MTA (yrRMTA) and yearly rate of relative MTA (yrRMTA) are simple protocols for measuring the relative extent of atrophy in the medial temporal lobe (MTL) in relation to the global brain atrophy. Albeit preliminary studies showed interest of these methods in the diagnosis of Alzheimer’s disease (AD), frontotemporal lobe degeneration (FTLD) and correlation with cognitive impairment in Parkinson’s disease (PD), formal feasibility and validity studies remained pending. As a first step, we aimed to assess the feasibility. Mainly, we aimed to assess the reproducibility of measuring the areas needed to compute these indices. We also aimed to assess the efforts needed to start using these methods correctly.</p><p><bold>Methods</bold>: A series of 290 1.5T-MRI studies from 230 subjects ranging 65–85 years old who had been studied for cognitive impairment were used in this study. Six inexperienced tracers (IT) plus one experienced tracer (ET) traced the three areas needed to compute the indices. Finally, tracers underwent a short survey on their experience learning to compute the MTAi and experience of usage, including items relative to training time needed to understand and apply the MTAi, time to perform a study after training and overall satisfaction.</p><p><bold>Results</bold>: Learning to trace the areas needed to compute the MTAi and derived methods is quick and easy. Results indicate very good intrarater Intraclass Correlation Coefficient (ICC) for the MTAi, good intrarater ICC for the 2D-MTA, yrMTA and yrRMTA and also good interrater ICC for the MTAi, 2D-MTA, yrMTA and yrRMTA.</p><p><bold>Conclusion</bold>: Our data support that MTAi and derived methods (2D-MTA, yrMTA and yrRTMA) have good to very good intrarater and interrater reproducibility and may be easily implemented in clinical practice even if new users have no experience tracing the area of regions of interest.</p>	<contrib contrib-type="author"><name><surname>Conejo Bayón</surname><given-names>Francisco</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/185241"/></contrib><contrib contrib-type="author"><name><surname>Maese</surname><given-names>Jesús</given-names></name><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/185276"/></contrib><contrib contrib-type="author"><name><surname>Fernandez Oliveira</surname><given-names>Aníbal</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/185237"/></contrib><contrib contrib-type="author"><name><surname>Mesas</surname><given-names>Tamara</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref></contrib><contrib contrib-type="author"><name><surname>Herrera de la Llave</surname><given-names>Estibaliz</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/190120"/></contrib><contrib contrib-type="author"><name><surname>Álvarez Avellón</surname><given-names>Tania</given-names></name><xref ref-type="aff" rid="aff3"><sup>3</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/125759"/></contrib><contrib contrib-type="author"><name><surname>Menéndez-González</surname><given-names>Manuel</given-names></name><xref ref-type="aff" rid="aff4"><sup>4</sup></xref><xref ref-type="aff" rid="aff5"><sup>5</sup></xref><xref ref-type="aff" rid="aff6"><sup>6</sup></xref><xref ref-type="author-notes" rid="fn001"><sup>*</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/52592"/></contrib>	Frontiers in Aging Neuroscience	<sec sec-type="introduction" id="s1"><title>Introduction</title><p>The recent focus on biomarkers in the diagnosis of Alzheimer’s disease (AD) have created a need to translate research findings into tools for use in everyday clinical practice. Although AD and mild cognitive impairment (MCI) are commonly diagnosed using criteria based in clinical findings, MRI findings may aid the clinical diagnosis, and may predict clinical progression. New research criteria have recently been proposed for AD, and MCI that incorporate (disproportionate) medial temporal lobe (MTL) or hippocampal atrophy on MRI as one of the supportive features.</p><p>Age-associated differences are detected in the MTL with an acceleration of Medial Temporal Lobe Atrophy (MTA) starting around 72 years of age in healthy people (Jack et al., <xref rid="B5" ref-type="bibr">2004</xref>; Salk et al., <xref rid="B12" ref-type="bibr">2014</xref>). However, these changes are modest and their rate of progression over time is relatively slow with a mean rate of about 1.6% per year. Accelerated MTA is a consistent finding in AD and MCI with rates of about 2.8% in stable MCI, 3.7% in MCI transitioning to AD (MCI progressors), and up to 4.0% in AD. Frontotemporal dementia may also lead to MTA, but in a different pattern: behavioral frontotemporal dementia and semantic dementia show atrophy in the anterior portion of the hippocampus, and in semantic dementia the atrophy is asymmetrical, with the left hippocampus being affected more severely. No significant hippocampal atrophy is detected in non-fluent progressive aphasia (Mesulam et al., <xref rid="B10" ref-type="bibr">2014</xref>). Other diseases such as dementia with Lewy bodies do not show MTA or it is much milder (Menéndez-González et al., <xref rid="B9" ref-type="bibr">2014</xref>).</p><p>All these changes can be measured on brain MRI using different approaches on structural MRI. Volumetric techniques quantify the volume of regions of interest on <italic>ad-hoc</italic> MRI studies that are quantified by using well manual well automatic image analysis. Planimetry methods are conceived to be used on standard MRI studies in routine clinical practice by measuring the area of regions of interest manually. Linear methods use measures of the width or height of brain structures, including the ventricular system or spaces around the brain cortex. Visual assessment rating scales are a quick way of assessing atrophy of the MTL on one single coronal MRI slice straightforward, and can be performed easily in the clinical setting; the disadvantage is that these scales are totally subjective and there is a loss of accuracy compared with objective analysis (Scheltens et al., <xref rid="B13" ref-type="bibr">1992</xref>).</p><p>In contrast to MTA, ventricular enlargement in old people lacks specificity, representing a measure of global brain atrophy and is strongly associated with aging both in healthy and diseased people. In addition, almost any neurodegenerative disorder affecting the brain hemispheres leads to some degree of ventricular enlargement, and so do some psychiatric conditions. Thus, it is interesting to compare measures indicative of atrophy in the MTL with measures indicative of global brain atrophy. This can be done using volumetry (3D) or planimetry (2D).</p><p>Shortly, the “Medial Temporal-lobe Atrophy index” (MTAi), is a simple method for measuring the relative extent of atrophy in the MTL in relation to the global brain atrophy (Menéndez-González, <xref rid="B8" ref-type="bibr">2014b</xref>). This 2D-method consists on calculating a ratio using the area of three regions traced manually on one single coronal MRI slide. High values are suggestive of atrophy in the MTL out of proportion to other brain structures, and therefore the pattern of atrophy matches the expected in typical AD (Galton et al., <xref rid="B3" ref-type="bibr">2001</xref>; van de Pol et al., <xref rid="B15" ref-type="bibr">2006</xref>).</p><p>Albeit preliminary studies showed interest of planimetric methods for diagnosing AD, frontotemporal lobe degeneration (FTLD) and cognitive impairment in Parkinson’s disease (PD)<xref ref-type="fn" rid="fn0001"><sup>1</sup></xref>, formal feasibility and validity studies remained pending. As a first step, we aimed to assess the feasibility of the MTAi and derived methods: 2D-Medial Temporal Atrophy (2D-MTA), yearly rate of MTA (yrRMTA) and tyearly rate of relative MTA (yrRMTA). Mainly, we aimed to assess the reproducibility of measuring the areas needed to compute these indices. Reproducibility refers to the degree of agreement between measurements or observations conducted on replicate specimens by different people, as part of the precision of a test method. Test–retest variability can be caused by intra-individual variability and intra-observer variability. These parameters have paramount importance when validating a new diagnostic test (Bossuyt et al., <xref rid="B1" ref-type="bibr">2003</xref>); and some recommendations on the validation of biomarkers for diagnosing dementias have also remarked the importance of assessing them in first term (McGhee et al., <xref rid="B6" ref-type="bibr">2014</xref>; Noel-Storr et al., <xref rid="B11" ref-type="bibr">2014</xref>). We also aimed to assess the efforts needed to start using these methods correctly.</p></sec><sec sec-type="methods" id="s2"><title>Methods</title><p>A series of 230 1.5T-MRI studies of subjects ranging 65–85 years old who had been studied for cognitive impairment were used in this study. Six inexperienced tracers (IT) plus one experienced tracer (ET) took part in this study. IT had different backgrounds in life sciences, basic neuroanatomical knowledge, and no previous experience in tracing areas of brain regions at all. The ET is one of the researchers who described the methods and has extensive experience using them.</p><p>First, the IT read the protocol of the MTAi (Menéndez-González, <xref rid="B7" ref-type="bibr">2014a</xref>). They were in charge of installing the DICOM software, loading MRI studies, selecting the appropriate slide and tracing the three areas needed to compute the MTAi on each hemisphere (right and left A, B and C) according to the original protocol, on a number of MRI studies ranging from 20 to 120 cases. The IT used different DICOM viewers depending on the operating system installed in their computers: 3 IOS users used Osirix™ and 3 Windows users used ONIS™. The areas traced by the IT were corrected by the ET in all cases. The MTAi consists on calculating a ratio using the area of three regions traced manually on one single coronal MRI slide at the level of the interpeduncular fossa: (1) the MTL region (A), defined in a coronal brain slide as the four-sided space bordered in its inferior side by the tentorium cerebelli, in its medial side by the cerebral peduncles, in its upper side by the roof of the temporal horn of the lateral ventricle and in its lateral side by the collateral sulcus and a straight-line linking the collateral sulcus with the lateral edge of the temporal horn of the lateral ventricle; (2) the parenchima within the medial temporal region, that includes the hippocampus and the parahippocampal gyrus—the fimbria taenia and plexus choroideus are excluded—(B); and (3) the body of the ipsilateral lateral ventricle (C) (Figure <xref ref-type="fig" rid="F1">1</xref>). Therefrom we can compute the “2D-Medial Temporal Atrophy” <italic>(2D-MTA = A−B)</italic> that represents absolute atrophy of the MTL; and the ratio “Medial Temporal Atrophy index” <italic> (MTAi = (A−B) × 10/C)</italic> that represents the relative atrophy of the MTL in comparison with the enlargement of the lateral ventricles, that represent global brain atrophy. The MTAi is suitable to assess the asymmetry of relative MTA within a subject. High asymmetry is typical of some types of FTLD. However, as there is important inter-individual variability in the size of the lateral ventricles, this index is not recommended for comparing subjects but to track the progression in a given subject over time. Indeed, if we have 2 MRI studies from different times (1 = first one, 2 = second one), we can also compute the yrMTA as follows: <italic>yrMTA = (A2−B2)−(A1−B1) × 1200/(#months between MRI studies)</italic> and the yearly rate of relative MTA (yrRMTA) as follows: <italic>yrRMTA=(A2−B2)−(A1−B1) × 1200/(C2−C1) × (# months between MRI studies)</italic>.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>MRI coronal section passing through the interpeduncular fosae</bold>. The boundaries of the three areas needed for calculating the Medial Temporal Atrophy index (MTAi) and derived methods have been drawn in three different colors: (1) the medial temporal lobe region (A, red), defined in a coronal brain slide as the space bordered in its inferior side by the tentorium cerebelli, in its medial side by the cerebral peduncles, in its upper side by the roof of the temporal horn of the lateral ventricle and in its lateral side by the colateral sulcus and a straight-line linking the colateral sulcus with the lateral edge of the temporal horn of the lateral ventricle; (2) the parenchima within the medial temporal region, that includes the hippocampus and the parahippocampal girus (B, blue); and (3) the body of the ipsilateral lateral ventricle (C, green).</p></caption><graphic xlink:href="fnagi-06-00305-g0001"/></fig><p>Finally, tracers underwent a short survey on their experience learning to compute the MTAi and experience of usage, including items relative to training time needed to understand and apply the MTAi (number of attempts needed until the ET verified the tracing was correct), time needed to perform a study after training (timed by the own tracers, in minutes) and overall satisfaction (measured using a simple qualitative scale: easy/normal/hard).</p><p>Statistical analyses were performed with the softwares R™ and Epidat™. Computation of the MTA, 2D-MTA, yrMTA and yrRMTA were made following the formulas explained above (Menéndez-González et al., <xref rid="B9" ref-type="bibr">2014</xref>). The reproducibility was assessed using the Intraclass Correlation Coefficient (ICC)-Model 2, one by one for every of the three areas needed to compute the MTAi on each hemisphere and for the final MTAi results (right and left). We used a qualitative scale to qualify the strength of concordance as very good (>0.90), good (0.80–0.90), moderate (0.60–0.80) and poor (<0.60).</p></sec><sec sec-type="results" id="s3"><title>Results</title><sec id="s3-1"><title>Reproducibility: results from the test-retest studies</title><p>In total, IT traced areas from 290 MRI studies from 230 cases, 60 of which also had a 1-year follow-up MRI study. Ninety studies were traced twice by the same IT (180 sets of results), and 200 were traced twice by 2 different IT (400 sets of results). We used the set of 180 double-traced by same-tracer results to assess the intrarater reproducibility and the set of 400 double-traced by different-tracer results to assess the interrater reproducibility. Thirty cases of the each set corresponded to the 1 year follow-up MRI. Thus, we computed the yrMTA and the yrRMTA in 60 cases and the MTAi and 2D-MTA in 230 cases. Results of intrarater and interrater reproducibility are shown in Tables <xref ref-type="table" rid="T1">1</xref>, <xref ref-type="table" rid="T2">2</xref> respectively.</p><table-wrap id="T1" position="float"><label>Table 1</label><caption><p><bold>Intrarater Intraclass Correlation Coefficient and strength of concordance among the different areas needed to compute the MTAi, 2D-MTA, yrMTA and yrRMTA</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th rowspan="1" colspan="1"/><th align="center" rowspan="1" colspan="1">CCI</th><th align="center" colspan="2" rowspan="1">CI 95%</th><th align="center" rowspan="1" colspan="1">Strength</th></tr><tr><td rowspan="1" colspan="1"/><th align="center" rowspan="1" colspan="1">Mean</th><th align="center" rowspan="1" colspan="1">Inferior</th><th align="center" rowspan="1" colspan="1">Superior</th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">rA</td><td align="char" char="." rowspan="1" colspan="1">0.87</td><td align="char" char="." rowspan="1" colspan="1">0.68</td><td align="char" char="." rowspan="1" colspan="1">0.91</td><td align="left" rowspan="1" colspan="1">Good</td></tr><tr><td align="left" rowspan="1" colspan="1">lA</td><td align="char" char="." rowspan="1" colspan="1">0.86</td><td align="char" char="." rowspan="1" colspan="1">0.66</td><td align="char" char="." rowspan="1" colspan="1">0.90</td><td align="left" rowspan="1" colspan="1">Good</td></tr><tr><td align="left" rowspan="1" colspan="1">rB</td><td align="char" char="." rowspan="1" colspan="1">0.83</td><td align="char" char="." rowspan="1" colspan="1">0.71</td><td align="char" char="." rowspan="1" colspan="1">0.89</td><td align="left" rowspan="1" colspan="1">Good</td></tr><tr><td align="left" rowspan="1" colspan="1">lB</td><td align="char" char="." rowspan="1" colspan="1">0.84</td><td align="char" char="." rowspan="1" colspan="1">0.76</td><td align="char" char="." rowspan="1" colspan="1">0.87</td><td align="left" rowspan="1" colspan="1">Good</td></tr><tr><td align="left" rowspan="1" colspan="1">rC</td><td align="char" char="." rowspan="1" colspan="1">0.95</td><td align="char" char="." rowspan="1" colspan="1">0.92</td><td align="char" char="." rowspan="1" colspan="1">0.94</td><td align="left" rowspan="1" colspan="1">Very good</td></tr><tr><td align="left" rowspan="1" colspan="1">lC</td><td align="char" char="." rowspan="1" colspan="1">0.94</td><td align="char" char="." rowspan="1" colspan="1">0.91</td><td align="char" char="." rowspan="1" colspan="1">0.96</td><td align="left" rowspan="1" colspan="1">Very good</td></tr><tr><td align="left" rowspan="1" colspan="1">rMTAi</td><td align="char" char="." rowspan="1" colspan="1">0.91</td><td align="char" char="." rowspan="1" colspan="1">0.90</td><td align="char" char="." rowspan="1" colspan="1">0.92</td><td align="left" rowspan="1" colspan="1">Very good</td></tr><tr><td align="left" rowspan="1" colspan="1">lMTAi</td><td align="char" char="." rowspan="1" colspan="1">0.92</td><td align="char" char="." rowspan="1" colspan="1">0.91</td><td align="char" char="." rowspan="1" colspan="1">0.93</td><td align="left" rowspan="1" colspan="1">Very good</td></tr><tr><td align="left" rowspan="1" colspan="1">r2D-MTA</td><td align="char" char="." rowspan="1" colspan="1">0.88</td><td align="char" char="." rowspan="1" colspan="1">0.80</td><td align="char" char="." rowspan="1" colspan="1">0.95</td><td align="left" rowspan="1" colspan="1">Good</td></tr><tr><td align="left" rowspan="1" colspan="1">l2D-MTA</td><td align="char" char="." rowspan="1" colspan="1">0.85</td><td align="char" char="." rowspan="1" colspan="1">0.81</td><td align="char" char="." rowspan="1" colspan="1">0.89</td><td align="left" rowspan="1" colspan="1">Good</td></tr><tr><td align="left" rowspan="1" colspan="1">ryrMTA</td><td align="char" char="." rowspan="1" colspan="1">0.82</td><td align="char" char="." rowspan="1" colspan="1">0.76</td><td align="char" char="." rowspan="1" colspan="1">0.89</td><td align="left" rowspan="1" colspan="1">Good</td></tr><tr><td align="left" rowspan="1" colspan="1">lyrMTA</td><td align="char" char="." rowspan="1" colspan="1">0.84</td><td align="char" char="." rowspan="1" colspan="1">0.77</td><td align="char" char="." rowspan="1" colspan="1">0.90</td><td align="left" rowspan="1" colspan="1">Good</td></tr><tr><td align="left" rowspan="1" colspan="1">ryrRMTA</td><td align="char" char="." rowspan="1" colspan="1">0.81</td><td align="char" char="." rowspan="1" colspan="1">0.77</td><td align="char" char="." rowspan="1" colspan="1">0.87</td><td align="left" rowspan="1" colspan="1">Good</td></tr><tr><td align="left" rowspan="1" colspan="1">lyrRMTA</td><td align="char" char="." rowspan="1" colspan="1">0.83</td><td align="char" char="." rowspan="1" colspan="1">0.75</td><td align="char" char="." rowspan="1" colspan="1">0.89</td><td align="left" rowspan="1" colspan="1">Good</td></tr></tbody></table><table-wrap-foot><p><italic>r: right hemisphere; l: left hemisphere</italic>.</p></table-wrap-foot></table-wrap><table-wrap id="T2" position="float"><label>Table 2</label><caption><p><bold>Interrater Intraclass Correlation Coefficient and strength of concordance among the different areas needed to compute the MTAi, 2D-MTA, yrMTA and yrRMTA</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th rowspan="1" colspan="1"/><th align="center" rowspan="1" colspan="1">CCI</th><th align="center" colspan="2" rowspan="1">CI 95%</th><th align="center" rowspan="1" colspan="1">Strength</th></tr><tr><td rowspan="1" colspan="1"/><th align="center" rowspan="1" colspan="1">Mean</th><th align="center" rowspan="1" colspan="1">Inferior</th><th align="center" rowspan="1" colspan="1">Superior</th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">rA</td><td align="char" char="." rowspan="1" colspan="1">0.87</td><td align="char" char="." rowspan="1" colspan="1">0.69</td><td align="char" char="." rowspan="1" colspan="1">0.92</td><td align="left" rowspan="1" colspan="1">Good</td></tr><tr><td align="left" rowspan="1" colspan="1">lA</td><td align="char" char="." rowspan="1" colspan="1">0.86</td><td align="char" char="." rowspan="1" colspan="1">0.72</td><td align="char" char="." rowspan="1" colspan="1">0.91</td><td align="left" rowspan="1" colspan="1">Good</td></tr><tr><td align="left" rowspan="1" colspan="1">rB</td><td align="char" char="." rowspan="1" colspan="1">0.83</td><td align="char" char="." rowspan="1" colspan="1">0.69</td><td align="char" char="." rowspan="1" colspan="1">0.89</td><td align="left" rowspan="1" colspan="1">Good</td></tr><tr><td align="left" rowspan="1" colspan="1">lB</td><td align="char" char="." rowspan="1" colspan="1">0.84</td><td align="char" char="." rowspan="1" colspan="1">0.71</td><td align="char" char="." rowspan="1" colspan="1">0.89</td><td align="left" rowspan="1" colspan="1">Good</td></tr><tr><td align="left" rowspan="1" colspan="1">rC</td><td align="char" char="." rowspan="1" colspan="1">0.88</td><td align="char" char="." rowspan="1" colspan="1">0.73</td><td align="char" char="." rowspan="1" colspan="1">0.91</td><td align="left" rowspan="1" colspan="1">Very good</td></tr><tr><td align="left" rowspan="1" colspan="1">lC</td><td align="char" char="." rowspan="1" colspan="1">0.90</td><td align="char" char="." rowspan="1" colspan="1">0.87</td><td align="char" char="." rowspan="1" colspan="1">0.92</td><td align="left" rowspan="1" colspan="1">Very good</td></tr><tr><td align="left" rowspan="1" colspan="1">rMTAi</td><td align="char" char="." rowspan="1" colspan="1">0.88</td><td align="char" char="." rowspan="1" colspan="1">0.82</td><td align="char" char="." rowspan="1" colspan="1">0.92</td><td align="left" rowspan="1" colspan="1">Good</td></tr><tr><td align="left" rowspan="1" colspan="1">lMTAi</td><td align="char" char="." rowspan="1" colspan="1">0.87</td><td align="char" char="." rowspan="1" colspan="1">0.83</td><td align="char" char="." rowspan="1" colspan="1">0.91</td><td align="left" rowspan="1" colspan="1">Good</td></tr><tr><td align="left" rowspan="1" colspan="1">r2D-MTA</td><td align="char" char="." rowspan="1" colspan="1">0.84</td><td align="char" char="." rowspan="1" colspan="1">0.79</td><td align="char" char="." rowspan="1" colspan="1">0.90</td><td align="left" rowspan="1" colspan="1">Good</td></tr><tr><td align="left" rowspan="1" colspan="1">l2D-MTA</td><td align="char" char="." rowspan="1" colspan="1">0.85</td><td align="char" char="." rowspan="1" colspan="1">0.81</td><td align="char" char="." rowspan="1" colspan="1">0.91</td><td align="left" rowspan="1" colspan="1">Good</td></tr><tr><td align="left" rowspan="1" colspan="1">ryrMTA</td><td align="char" char="." rowspan="1" colspan="1">0.83</td><td align="char" char="." rowspan="1" colspan="1">0.74</td><td align="char" char="." rowspan="1" colspan="1">0.89</td><td align="left" rowspan="1" colspan="1">Good</td></tr><tr><td align="left" rowspan="1" colspan="1">lyrMTA</td><td align="char" char="." rowspan="1" colspan="1">0.83</td><td align="char" char="." rowspan="1" colspan="1">0.75</td><td align="char" char="." rowspan="1" colspan="1">0.90</td><td align="left" rowspan="1" colspan="1">Good</td></tr><tr><td align="left" rowspan="1" colspan="1">ryrRMTA</td><td align="char" char="." rowspan="1" colspan="1">0.82</td><td align="char" char="." rowspan="1" colspan="1">0.75</td><td align="char" char="." rowspan="1" colspan="1">0.88</td><td align="left" rowspan="1" colspan="1">Good</td></tr><tr><td align="left" rowspan="1" colspan="1">lyrRMTA</td><td align="char" char="." rowspan="1" colspan="1">0.81</td><td align="char" char="." rowspan="1" colspan="1">0.76</td><td align="char" char="." rowspan="1" colspan="1">0.86</td><td align="left" rowspan="1" colspan="1">Good</td></tr></tbody></table><table-wrap-foot><p><italic>r: right hemisphere; l: left hemisphere</italic>.</p></table-wrap-foot></table-wrap></sec><sec id="s3-2"><title>Satisfaction: results from the survey</title><p>IT needed to train with 2–5 cases (mean 3 cases) before being able to compute the MTAi on their own correctly. After training, IT needed between 4 and 7 min (mean 5 min) to examine a new case. All tracers rated the method as “easy to learn” and “easy to apply”.</p></sec></sec><sec sec-type="discussion" id="s4"><title>Discussion</title><p>One of the strengths of planimetry methods is that can be measured using almost any of the DICOM softwares commonly used by clinicians or radiologists to visualize medical images worldwide. Most of these softwares are intuitive and require little or no training at all. Learning to trace the areas needed to compute the MTAi and derived methods is quick and easy even for naive tracers. Even more importantly, tracing these areas have good intra- and interrater reproducibility. As expected, the area with the best intra- and interrater ICC was area C—the lateral ventricle—since it has the easiest anatomical limits. Areas A and B had poorer intrarater and interrater ICC since anatomical limits are somewhat more complicated. However, the intrarater and interrater ICC for areas A and B is still good enough to yield very good intrarater ICC for the MTAi, good intrarater ICC for the 2D-MTA, yrMTA and yrRMTA and also good interrater ICC for the MTAi, 2D-MTA, yrMTA and yrRMTA. These results are comparable to those from automatic volumetry (Hsu et al., <xref rid="B4" ref-type="bibr">2002</xref>; Wolz et al., <xref rid="B16" ref-type="bibr">2014</xref>) and much better than those from manual volumetry and visual scales (Scheltens et al., <xref rid="B14" ref-type="bibr">1997</xref>; Hsu et al., <xref rid="B4" ref-type="bibr">2002</xref>). Particularly, assessment of cerebral atrophy using visual rating scales is totally subjective and has moderate intrarater and poor interrater reproducibility (Scheltens et al., <xref rid="B14" ref-type="bibr">1997</xref>).</p><p>Accurate manual volumetric assessment requires standard operating procedures that include the know-how specific for the modality, acquisition parameters, and extensive learning and <italic>ad-hoc</italic> softwares that also require training to be used correctly (Frisoni et al., <xref rid="B17" ref-type="bibr">2014</xref>). In addition, harmonization of manually segmented hippocampus is still in progress (Frisoni and Jack, <xref rid="B2" ref-type="bibr">2011</xref>). Thus, manual volumetric methods have a steep learning curve while the MTAi while derived 2D methods have a learning curve with a quick start. The MTAi and derived methods may be easily implemented for estimating MTA in clinical practice, even if new users have no experience tracing the area of regions of interest. Indeed, any health professional with basic neuroanatomical knowledge can take these measures after short training. The MTAi and derived methods can also readily be incorporated into a standardized radiological report and may also be useful in clinical trials.</p></sec><sec sec-type="conclusions" id="s5"><title>Conclusions</title><p>In conclusion, results from this feasibility study support that the MTAi and derived methods (2D-MTA, yrMTA and yrRMTA) have good to very good intrarater and interrater reproducibility and may be easily implemented for estimating MTA in clinical practice, even if new users have no experience tracing the area of regions of interest.</p></sec><sec id="s6"><title>Conflict of interest statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec>
Comparative Functional Genomics and the Bovine Macrophage Response to Strains of the <italic>Mycobacterium</italic> Genus	<p>Mycobacterial infections are major causes of morbidity and mortality in cattle and are also potential zoonotic agents with implications for human health. Despite the implementation of comprehensive animal surveillance programs, many mycobacterial diseases have remained recalcitrant to eradication in several industrialized countries. Two major mycobacterial pathogens of cattle are <italic>Mycobacterium bovis</italic> and <italic>Mycobacterium avium</italic> subspecies <italic>paratuberculosis</italic> (MAP), the causative agents of bovine tuberculosis (BTB) and Johne’s disease (JD), respectively. BTB is a chronic, granulomatous disease of the respiratory tract that is spread via aerosol transmission, while JD is a chronic granulomatous disease of the intestines that is transmitted via the fecal-oral route. Although these diseases exhibit differential tissue tropism and distinct complex etiologies, both <italic>M. bovis</italic> and MAP infect, reside, and replicate in host macrophages – the key host innate immune cell that encounters mycobacterial pathogens after initial exposure and mediates the subsequent immune response. The persistence of <italic>M. bovis</italic> and MAP in macrophages relies on a diverse series of immunomodulatory mechanisms, including the inhibition of phagosome maturation and apoptosis, generation of cytokine-induced necrosis enabling dissemination of infection through the host, local pathology, and ultimately shedding of the pathogen. Here, we review the bovine macrophage response to infection with <italic>M. bovis</italic> and MAP. In particular, we describe how recent advances in functional genomics are shedding light on the host macrophage–pathogen interactions that underlie different mycobacterial diseases. To illustrate this, we present new analyses of previously published bovine macrophage transcriptomics data following <italic>in vitro</italic> infection with virulent <italic>M. bovis</italic>, the attenuated vaccine strain <italic>M. bovis</italic> BCG, and MAP, and discuss our findings with respect to the differing etiologies of BTB and JD.</p>	<contrib contrib-type="author"><name><surname>Rue-Albrecht</surname><given-names>Kévin</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/132796"/></contrib><contrib contrib-type="author"><name><surname>Magee</surname><given-names>David A.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="author-notes" rid="fn001"><sup>†</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/25243"/></contrib><contrib contrib-type="author"><name><surname>Killick</surname><given-names>Kate E.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/153201"/></contrib><contrib contrib-type="author"><name><surname>Nalpas</surname><given-names>Nicolas C.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/180724"/></contrib><contrib contrib-type="author"><name><surname>Gordon</surname><given-names>Stephen V.</given-names></name><xref ref-type="aff" rid="aff3"><sup>3</sup></xref><xref ref-type="aff" rid="aff4"><sup>4</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/180648"/></contrib><contrib contrib-type="author"><name><surname>MacHugh</surname><given-names>David E.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="aff" rid="aff4"><sup>4</sup></xref><xref ref-type="corresp" rid="cor1">*</xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/23873"/></contrib>	Frontiers in Immunology	<sec sec-type="introduction" id="S1"><title>Introduction</title><p><italic>Mycobacterium</italic> is a Gram-positive genus of Actinobacteria that includes more than 120 species (<xref rid="B1" ref-type="bibr">1</xref>, <xref rid="B2" ref-type="bibr">2</xref>). Although the majority of species in this genus are non-pathogenic environmental bacteria, a few species are highly successful intracellular pathogens of human beings and other mammals including <italic>Mycobacterium tuberculosis</italic> and <italic>Mycobacterium bovis –</italic> the causative agents of human being and bovine tuberculosis (BTB), respectively – and <italic>Mycobacterium avium</italic> subspecies <italic>paratuberculosis</italic> (MAP), the causative agent of Johne’s disease (JD) in cattle (<xref rid="B3" ref-type="bibr">3</xref>, <xref rid="B4" ref-type="bibr">4</xref>). The success of these pathogenic mycobacteria is partly due to their ability to infect, reside, and proliferate inside host macrophages, despite the antimicrobial properties of these cells. Macrophages serve as key effector innate immune cells that mediate the initial host response to infection via the activity of inflammatory cytokines and chemokines; this initial interaction leads to either the eradication of intracellular bacilli or the formation of organized collections of immune cells, termed granulomas, which contain infection (<xref rid="B5" ref-type="bibr">5</xref>).</p><p>Infections with pathogenic mycobacteria can manifest as acute or chronic disease or involve lengthy subclinical phases of infection with the potential to reactivate later. It is also understood that the establishment of successful infection is underpinned by subversion and modulation of host macrophage antimicrobial mechanisms, including the prevention of macrophage phagosome–lysosome fusion, inhibition of macrophage apoptosis, and suppression of antigen presentation and signaling mechanisms within the macrophage (<xref rid="B6" ref-type="bibr">6</xref>–<xref rid="B8" ref-type="bibr">8</xref>). Furthermore, it has been proposed that virulent mycobacteria exploit host defense mechanisms, such as the induction of cytokine-induced necrosis, which results in immunopathology, the dissemination of infection through the host and ultimately pathology that leads to shedding of the pathogen from the host, thereby maintaining the cycle of infection (<xref rid="B9" ref-type="bibr">9</xref>). Consequently, investigating the complex interplay between mycobacterial pathogens and the host macrophage is critical to our understanding of the immuno-pathogenesis of mycobacterial diseases.</p></sec><sec id="S2"><title>The <italic>Mycobacterium tuberculosis</italic> Complex</title><p>The genus <italic>Mycobacterium</italic> contains the <italic>Mycobacterium tuberculosis</italic> complex (MTBC) that includes seven major pathogenic species and subspecies that cause tuberculosis in a range of mammalian hosts, the most well-studied member of which is <italic>M. tuberculosis</italic> – the causative agent of human tuberculosis. Typically, the members of the MTBC display greater than 99.95% nucleotide sequence identity at the genome level, with little or no evidence for the exchange of genetic material between species and strains (<xref rid="B10" ref-type="bibr">10</xref>). Despite this high level of genome similarity, the members of the MTBC differ with respect to host range and pathogenicity: <italic>M. tuberculosis</italic> and <italic>Mycobacterium africanum</italic> are almost exclusively human pathogens; <italic>Mycobacterium microti</italic> causes disease in rodents including voles; <italic>Mycobacterium pinnipedii</italic> causes tuberculosis in marine mammals including seals and sea lions; and <italic>Mycobacterium caprae</italic> is very closely related to <italic>M. bovis</italic> and infects both goats and deer. The species with the largest host range is <italic>M. bovis</italic>, which is mainly isolated from cattle, but can also be responsible for outbreaks in wild animals. Furthermore, <italic>M. bovis</italic> can cause disease in human beings yet rarely transmits between immunocompetent hosts. A closely related mycobacterial species, <italic>Mycobacterium canettii</italic>, causes pathology in human beings, but differs from the other members of the MTBC in that it displays smooth colony morphology rather than the characteristic rough morphology of the other MTBC members (<xref rid="B11" ref-type="bibr">11</xref>, <xref rid="B12" ref-type="bibr">12</xref>).</p><p>Phylogenetic analyses using insertion/deletion DNA sequence polymorphisms (indels), such as the variable regions of difference (RD – see below) and whole gene and genome sequences have revealed that the evolutionary history of the MTBC represents a pattern of genome downsizing characterized by chromosomal DNA sequence deletions and the inability of these species to repair deletions through recombinogenic processes (<xref rid="B10" ref-type="bibr">10</xref>, <xref rid="B13" ref-type="bibr">13</xref>). These studies support a distinct phylogenetic position of <italic>M. canettii</italic> from all other MTBC members: <italic>M. canettii</italic> strains possess intact RD sequences that are absent from the other MTBC species together with one species-specific deletion (RD<sup>can</sup>). <italic>M. canettii</italic> strains also have 26 additional spacer sequences that are not found in other MTBC species [Figure <xref ref-type="fig" rid="F1">1</xref>] (<xref rid="B10" ref-type="bibr">10</xref>). Indeed, it has been recently proposed that <italic>M. canettii</italic> and other smooth tubercle bacilli (STB) lineages diverged from the common ancestor of all tubercle bacilli prior to the clonal radiation of non-smooth MTBC lineages and that non-smooth MTBC lineages evolved from an STB-like mycobacterial ancestor, sometimes referred to as <italic>Mycobacterium prototuberculosis</italic> (<xref rid="B14" ref-type="bibr">14</xref>).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Neighbor-joining phylogeny of selected mycobacteria species and strains based on partial 16S rRNA sequences</bold>. Bootstrap supports are based on 1,000 pseudoreplicates. Species and strains belonging to the <italic>Mycobacterium tuberculosis</italic> complex are shaded in red; members of the <italic>Mycobacterium avium</italic> complex are shaded in green.</p></caption><graphic xlink:href="fimmu-05-00536-g001"/></fig><p>Genomic comparisons across the MTBC revealed a number of “regions of difference” (RD loci), with the presence or absence of these loci capable of differentiating the constituent strains of the MTBC. Notably, deletion of chromosomal region RD9 distinguishes <italic>M. tuberculosis</italic> strains from animal-adapted lineages, including <italic>M. bovis</italic>, while RD1 [which encodes, among other genes, the ESX-1 secretion system plus key secreted effectors including early secretory antigen target 6 (ESAT-6) and culture filtrate protein 10 (CFP-10)] is deleted in <italic>M. bovis</italic> Bacille Calmette–Guérin (<italic>M. bovis</italic> BCG) vaccine strains. The use of deletions to differentiate mycobacterial strains led to the proposal of an evolutionary scenario positing <italic>M. tuberculosis</italic> as being closer to the common ancestor of the MTBC and rejection of the hypothesis that <italic>M. tuberculosis</italic> infection in human populations arose from an animal pathogen such as <italic>M. bovis</italic> in parallel with cattle domestication and husbandry (<xref rid="B15" ref-type="bibr">15</xref>, <xref rid="B16" ref-type="bibr">16</xref>).</p></sec><sec id="S3"><title>The <italic>Mycobacterium avium</italic> Complex</title><p>A second major evolutionarily distinct cluster of mycobacterial species is represented by the <italic>Mycobacterium avium</italic> complex [MAC] (Figure <xref ref-type="fig" rid="F1">1</xref>). This complex shares an estimated 40% nucleotide sequence similarity with members of the MTBC based on proteome-derived DNA sequences (<xref rid="B17" ref-type="bibr">17</xref>). The MAC comprises several closely related, slow-growing, pathogenic, and non-pathogenic species. Among the pathogenic species within the MAC are <italic>M. avium</italic> and its subspecies (MAP; <italic>M. avium</italic> subsp. <italic>avium</italic>; <italic>M. avium</italic> subsp. <italic>silvaticum</italic>; and <italic>M. avium</italic> subsp. <italic>hominissuis</italic>) and <italic>Mycobacterium intracellulare</italic>. Phylogenetic analyses based on partial gene, whole gene, and complete genome DNA sequences have revealed that subspecies of <italic>M. avium</italic> typically share over ≥95% nucleotide sequence identity, while nucleotide identity between <italic>M. avium</italic> subspecies and <italic>M. intracellulare</italic> was estimated between 80 and 94% (<xref rid="B18" ref-type="bibr">18</xref>–<xref rid="B21" ref-type="bibr">21</xref>).</p><p>Despite their genetic similarity, members of the MAC show differential host and tissue tropisms. For example, <italic>M. avium</italic> subsp. <italic>avium</italic> is the classical causative agent of tuberculosis in birds, while <italic>M. avium</italic> subsp. <italic>silvaticum</italic> has been shown to cause tuberculous lesions in wood pigeons. In this regard, <italic>M. avium</italic> subsp. <italic>avium</italic> and <italic>M. avium</italic> subsp. <italic>silvaticum</italic> together represent a distinct lineage of avian pathogens. <italic>M. avium</italic> subsp. <italic>hominissuis</italic> and <italic>M. intracellulare</italic> are opportunistic pathogens widely distributed in the environment and can cause disseminated tuberculosis and pulmonary disease in a range of mammalian hosts, including pigs, cattle, and human beings (<xref rid="B22" ref-type="bibr">22</xref>). From a human perspective, <italic>M. avium</italic> subsp. <italic>hominissuis</italic> is considered to be most clinically relevant member of the MAC member, where it has been previously shown to have caused disseminated infections among immunocompromised patients (<xref rid="B23" ref-type="bibr">23</xref>). MAP, in contrast, is an obligate intracellular pathogen of ruminants that causes JD characterized by chronic enteritis, with severe economic losses for the dairy industry in many countries (<xref rid="B3" ref-type="bibr">3</xref>). MAP can be differentiated from the other subspecies of <italic>M. avium</italic> by its very slow growth rate <italic>in vitro</italic> (between 8 and 24 weeks growth is required for visible colony formation). In addition, MAP is dependent on the siderophore mycobactin J, an iron-chelating cell wall component, for growth in primary cultures (<xref rid="B24" ref-type="bibr">24</xref>).</p></sec><sec id="S4"><title><italic>Mycobacterium bovis</italic> Infection and Bovine Tuberculosis</title><p>Bovine tuberculosis is caused by infection with <italic>M. bovis</italic>, which continues to pose a threat to livestock worldwide. Furthermore, as a zoonotic pathogen, <italic>M. bovis</italic> also has serious implications for human health (<xref rid="B25" ref-type="bibr">25</xref>). It has been estimated that BTB contributes losses of $3 billion to global agriculture annually (<xref rid="B26" ref-type="bibr">26</xref>, <xref rid="B27" ref-type="bibr">27</xref>), while comprehensive econometric analyses place BTB as the fourth most important livestock disease worldwide (<xref rid="B28" ref-type="bibr">28</xref>). The impacts of BTB infection are manifold, including significant economic and social effects due to the slaughter of infected animals, compensatory payments to producers, continual surveillance programs, and disruption to agricultural trade and productivity (<xref rid="B29" ref-type="bibr">29</xref>).</p><p>Bovine tuberculosis is predominantly a pulmonary disease, characterized largely by the formation of tuberculous lesions in the upper respiratory lymph nodes of the lung and thorax. In some cases, tuberculous lesions have also been detected in the cranial lymph nodes (<xref rid="B30" ref-type="bibr">30</xref>). The etiology and host immune response to <italic>M. bovis</italic> is similar to <italic>M. tuberculosis</italic> infections in human beings (<xref rid="B31" ref-type="bibr">31</xref>). Infection is normally caused by the inhalation of aerosolized respiratory secretions containing infectious bacilli, with the natural site of infection being the respiratory tract, presumably on the alveolar surface of the lung. Following inhalation, the bacilli are rapidly encountered by host alveolar macrophages and other phagocytic innate immune cells (such as dendritic cells), which serve as key innate immune effector cells that provide the first line of defense against the pathogen. At this stage, bacilli can be destroyed by the antimicrobial actions of the macrophage; however, bacilli that evade intracellular destruction can persist and multiply within infected macrophages. This results in the migration of infected macrophages to regional lymph nodes, where protective T<sub>H</sub>1 cell immunity is induced through the recruitment and interaction of additional innate and adaptive immune cells, culminating in the formation of granulomas – organized complexes of immune cells composed of lymphocytes, non-infected macrophages, and neutrophils that contain infected macrophages and prevent the dissemination of bacilli (<xref rid="B5" ref-type="bibr">5</xref>, <xref rid="B31" ref-type="bibr">31</xref>, <xref rid="B32" ref-type="bibr">32</xref>). However, in most cases, the pathogen is not eliminated by the host; rather, the pathogen persists in a dormant stage within the granuloma for prolonged periods of time, becoming metabolically and reproductively active following the breakdown of the granuloma and dysregulation of protective T-cell immunity. This results in the development of active tuberculosis, causing immunopathology in the host and enabling the transmission of infection (<xref rid="B31" ref-type="bibr">31</xref>).</p></sec><sec id="S5"><title>MAP Infection and Johne’s Disease</title><p>Johne’s disease, caused by infection with MAP, is a chronic inflammatory disease that affects the gastrointestinal tract of cattle and other ruminants. Specifically, JD presents as a granulomatous inflammation of the intestinal tissue and regional lymph nodes due to a massive influx of monocytes and macrophages. This inflammation effectively prevents absorption of nutrients, and, therefore, during the later stages of disease cattle manifest significant weight loss and diarrhea, resulting in progressive physiological wasting and death. Disease progression is generally classified into four stages: silent infection, subclinical, clinical, and advanced clinical disease; in particular, the subclinical phase can be extremely lengthy (between 2 and 5 years), with pathology largely restricted to the ileum, rendering diagnosis difficult (<xref rid="B24" ref-type="bibr">24</xref>).</p><p>Exposure to MAP in ruminants generally occurs within the first months of life, through either a fecal–oral route or by ingestion of contaminated colostrum or milk, although evidence suggests that some infections can occur <italic>in utero</italic> (<xref rid="B33" ref-type="bibr">33</xref>). Following ingestion, the bacilli colonize the mucosa-associated lymphoid tissues of the upper gastrointestinal tract and are subsequently endocytosed by the microfold cells (M-cells) that cover the ileal Peyer’s patches. The bacilli are subsequently phagocytosed by subepithelial and intraepithelial intestinal macrophages, where they reside and multiply (<xref rid="B34" ref-type="bibr">34</xref>). The subsequent host cellular immune response leads to the development of granulomas involving adjacent lymph nodes. After years of latent infection, bacilli are assumed to reactivate and trigger a state of active cellular proliferation, leading to the corrugated intestinal epithelia and clinical manifestations with shedding of bacilli into the environment and grassland completing the infection cycle. The disease finally presents as a malnutrition syndrome that culminates in the death of the animal (<xref rid="B3" ref-type="bibr">3</xref>, <xref rid="B35" ref-type="bibr">35</xref>, <xref rid="B36" ref-type="bibr">36</xref>).</p><p>Johne’s disease has major implications for domestic animal health worldwide causing significant economic loss in affected herds, which is largely due to decreased milk yields, reduced slaughter weight, premature culling of infected animals, and losses due to continued spread of infection (<xref rid="B37" ref-type="bibr">37</xref>). In cattle, JD results in estimated losses of $250 million to the US dairy industry annually, while dairy herd prevalence of JD is estimated to be greater than 50% in certain US states and European countries (<xref rid="B35" ref-type="bibr">35</xref>, <xref rid="B38" ref-type="bibr">38</xref>, <xref rid="B39" ref-type="bibr">39</xref>). Furthermore, it has been hypothesized that MAP infection may trigger or exacerbate Crohn’s disease, an inflammatory disease of the intestines in human beings with similar granulomatous pathology at the ileocecal valve; however, this proposed link between MAP infection and Crohn’s disease remains contentious (<xref rid="B40" ref-type="bibr">40</xref>).</p></sec><sec id="S6"><title>The Role of the Macrophage during Mycobacterial Infection</title><p>Although BTB and JD exhibit distinct complex etiologies, the causative agents of these diseases display a propensity to infect, reside, and replicate in host macrophages – the key host innate immune cell that mediates the immune response following infection. Macrophage recognition of mycobacteria occurs through the interaction of mycobacterial pathogen-associated molecular patterns (PAMPs) – such as lipopolysaccharide, and various lipoproteins and glycolipids (e.g., lipoarabinomannan) – with host pathogen recognition receptors (PRRs) proteins, such as Toll-like receptors (TLRs), which are expressed on the macrophage cell surface (<xref rid="B41" ref-type="bibr">41</xref>). Macrophage PRR activation induces signaling pathways resulting in the production of endogenous NF-κB-inducible cytokines and chemokines that promote a T<sub>H</sub>1 immune response characterized by the release of proinflammatory IFN-γ, primarily from CD4<sup>+</sup> T-cells, and the lysing of infected macrophages by cytotoxic CD8<sup>+</sup> T-cells. IFN-γ induces microbicidal activity in infected macrophages and enhances the expression of major histocompatibility complex (MHC) class I and II molecules necessary for the presentation of mycobacterial antigens on the macrophage cell surface to CD8<sup>+</sup> and CD4<sup>+</sup> T-cells, respectively. These mechanisms can lead to either the immediate killing of the pathogen and clearing of infection, or the containment of infection through the formation of granulomas (<xref rid="B42" ref-type="bibr">42</xref>–<xref rid="B45" ref-type="bibr">45</xref>).</p><p>Pathogenic mycobacteria have evolved a diverse range of immunoevasive mechanisms that facilitate survival and replication within the host macrophage. These immunoevasive mechanisms include inhibition of phagosome maturation necessary for destruction of the pathogen and antigen presentation (<xref rid="B46" ref-type="bibr">46</xref>, <xref rid="B47" ref-type="bibr">47</xref>); evasion of macrophage apoptosis and activation of macrophage necrosis, which facilitates release of bacilli from the macrophage and encourages dissemination of infection to other cells (<xref rid="B7" ref-type="bibr">7</xref>, <xref rid="B48" ref-type="bibr">48</xref>); and the subversion of innate cell signaling, which is critical to the establishment of infection and progression to active disease (<xref rid="B49" ref-type="bibr">49</xref>, <xref rid="B50" ref-type="bibr">50</xref>). It has also been recently demonstrated that virulent <italic>M. tuberculosis</italic> strains preferentially infect permissive macrophages and evade microbicidal macrophages through the masking of PAMPs with cell surface associated lipids (<xref rid="B51" ref-type="bibr">51</xref>).</p><p>Failure or subversion of an appropriate innate immune response is critical to the establishment of infection and progression to disease; central to this process is the macrophage response to infection (<xref rid="B31" ref-type="bibr">31</xref>). Consequently, analysis of the bovine macrophage response to <italic>in vitro</italic> infections with <italic>M. bovis</italic> and MAP may provide insights into the cellular mechanisms that underlie and govern the divergent immunopathology of BTB and JD (<xref rid="B36" ref-type="bibr">36</xref>, <xref rid="B52" ref-type="bibr">52</xref>).</p></sec><sec id="S7"><title>Functional Genomics Analysis of the Bovine Macrophage Response to Mycobacteria</title><p>Early investigations of the bovine macrophage response to mycobacterial infection focused on the analysis of the expression of single or small numbers of immunological parameters. For example, the quantification of gene or protein expression using reverse transcription quantitative real-time PCR (RT-qPCR) and ELISA technologies; however, focused studies such as these, are unable to provide a high-resolution overview of the global macrophage response to infection. Pathogen-induced activation of host macrophages is characterized by large-scale changes in the expression profile of genes critical for the control and eradication of the pathogen, while modulation of host gene expression critical for pathogen survival is also expected to be reflected in the transcriptome of the macrophage (<xref rid="B53" ref-type="bibr">53</xref>, <xref rid="B54" ref-type="bibr">54</xref>). Consequently, genomics technologies that assay pan-genomic changes in gene expression have been widely used to discern patterns of host-gene regulation during infection. In particular, the development of high-throughput gene expression technologies, such as microarrays and RNA-sequencing (RNA-seq), over the past decade, coupled with dramatic improvements in mammalian genome resources and increasingly sophisticated computational tools for the analysis of large-scale gene expression datasets are providing new opportunities for detection, cataloging, and analysis of the large numbers of host macrophage genes expressed in response to mycobacterial infection in cattle (<xref rid="B55" ref-type="bibr">55</xref>–<xref rid="B64" ref-type="bibr">64</xref>).</p><p>A primary goal of our research group is to use high-throughput functional genomics technologies to analyze the bovine macrophage transcriptome following infection with <italic>M. bovis</italic> and <italic>M. paratuberculosis</italic> to improve our understanding of host–pathogen interactions that characterize and underlie BTB and JD. Previously, we used a 24-h time course infection model to investigate the transcriptome response of bovine monocyte-derived macrophages (MDM) infected with <italic>M. bovis</italic> and MAP using data generated from the Affymetrix<sup>®</sup> GeneChip<sup>®</sup> Bovine Genome microarray platform (<xref rid="B61" ref-type="bibr">61</xref>, <xref rid="B62" ref-type="bibr">62</xref>). These analyses revealed a large number of differentially expressed (DE) genes following <italic>M. bovis</italic> infection relative to non-infected control MDM such that the number of DE genes also increased across the time course, with the highest number observed 24 h post-infection [hpi] (<xref rid="B62" ref-type="bibr">62</xref>). This contrasted with results from MAP-infected MDM relative to the same control macrophages, which showed the highest number of DE genes at the 2 hpi time point with a decrease in the number of DE genes at the later time points post-infection (i.e., 6 and 24 hpi) (<xref rid="B61" ref-type="bibr">61</xref>). These findings suggest that <italic>M. bovis</italic> and MAP have differential survival strategies once internalized by macrophages, which in turn, may underlie the divergent immunopathology associated with BTB and JD.</p></sec><sec id="S8"><title>The Monocyte-Derived Macrophage Infection Model and Gene Expression Datasets Used for Comparative Functional Analysis</title><p>To further investigate the similarities and differences of the bovine MDM response to virulent and attenuated mycobacterial species/strains, we have reanalyzed and directly compared the Affymetrix<sup>®</sup> GeneChip<sup>®</sup> Bovine Genome microarray data from our earlier work (i.e., the non-infected control MDM and the <italic>M. bovis</italic>- and MAP-infected bovine MDM) together with corresponding microarray data from MDM infected with <italic>M. bovis</italic> BCG (<xref rid="B65" ref-type="bibr">65</xref>). All infected and control MDM used to generate these data were derived from the same seven age-matched Holstein-Friesian females, while a multiplicity of infection (MOI) of 2:1 (i.e., 2 bacilli:1 MDM) was used for all MDM infections (<xref rid="B61" ref-type="bibr">61</xref>, <xref rid="B62" ref-type="bibr">62</xref>). Gene expression omnibus (GEO) data series accession numbers used for these re-analyses were GSE33309, GSE35185, and GSE59774 (<xref rid="B66" ref-type="bibr">66</xref>).</p><p>For this new comparative analysis, gene expression data from <italic>M. bovis</italic>-, MAP-, and <italic>M. bovis</italic> BCG-infected MDM together with data from the non-infected control MDM at time points 2, 6, and 24 hpi were used. Prior to differential gene expression analysis, all microarray data were quality checked using the arrayQualityMetrics package in Bioconductor (<xref rid="B67" ref-type="bibr">67</xref>). Raw data from two microarrays (one MAP- and one <italic>M. bovis</italic> BCG infected MDM sample) did not pass the QC thresholds set in the arrayQualityMetrics package and were removed from all further downstream analyses. Furthermore, all arrays generated from these two animals were excluded to ensure a balanced experimental design for comparative gene expression analysis (Figure <xref ref-type="fig" rid="F2">2</xref>). Next, all raw gene expression data were normalized and filtered for non-informative probe sets using the I/NI algorithm implemented in the FARMS Bioconductor software package (version 1.14.0) (<xref rid="B68" ref-type="bibr">68</xref>); this analysis yielded 11,842 informative probe sets for use in downstream analysis.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>The experimental design used for the comparative functional genomics analysis described in the current study</bold>. Previously, MDM from seven age-matched females were infected with <italic>M. bovis</italic> and MAP (MOI 2:1); control MDM received culture media only (<xref rid="B61" ref-type="bibr">61</xref>, <xref rid="B62" ref-type="bibr">62</xref>). We also infected, in parallel, MDM from the same animals with <italic>M. bovis</italic> BCG (MOI 2:1). Infections were performed across duplicate tissue culture plate wells (shaded circles); total RNA from duplicate treatment wells was harvested and pooled at 2, 6, and 24 hpi. Pan-genomic gene expression data for each RNA sample was generated using the Affymetrix<sup>®</sup> GeneChip<sup>®</sup> Bovine Genome microarray platform (<xref rid="B61" ref-type="bibr">61</xref>, <xref rid="B62" ref-type="bibr">62</xref>). For the comparative functional genomics analysis, microarray data from only five of these animals were used to ensure a balanced experimental design following quality control assessment (see main body text for details).</p></caption><graphic xlink:href="fimmu-05-00536-g002"/></fig><p>The 11,842 informative probe sets identified post-filtering were then used to generate a multi-dimensional scaling (MDS) plot summarizing the transcriptomic relationship between samples (Figure <xref ref-type="fig" rid="F3">3</xref>). Notably, samples clustered according to time on the first dimension, while the second dimension clustered samples according to animal ID. However, there was a noticeable clustering of RNA samples from <italic>M. bovis</italic> BCG- and MAP-infected samples within each group of samples corresponding to a particular animal at a given time. Investigation of the expressed genes with the greatest animal effect revealed that loci within the bovine MHC displayed the greatest differences in expression among animals with relatively small gene expression differences within each animal across time and treatment (Figure S1 in Supplementary Material). The magnitude of the expression differences among animals for these genes presumably contributes to the separation of the samples by animal on the second dimension in the MDS plot (Figure <xref ref-type="fig" rid="F3">3</xref>).</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Multi-dimensional scaling (MDS) plot of the infected MDM at each time point post-infection</bold>. Manhattan distances (calculated from 11,842 informative probe sets) were used to generate the MDS plot.</p></caption><graphic xlink:href="fimmu-05-00536-g003"/></fig><p>We propose two hypotheses to explain the observed expression differences among animals. First, expressed MHC loci, which are among the most polymorphic loci in mammals, generate mRNA transcripts that display considerable nucleotide differences between individuals from the same species (<xref rid="B69" ref-type="bibr">69</xref>–<xref rid="B71" ref-type="bibr">71</xref>). mRNA transcripts that vary appreciably from the reference transcript sequences used to produce the microarray probe sets may, therefore, have a reduced hybridization efficiency compared to mRNA transcripts that are identical to or differ only slightly from the reference transcript sequences. In turn, this may result in Type 1 errors for differential gene expression estimates between samples or sample groups (<xref rid="B72" ref-type="bibr">72</xref>–<xref rid="B74" ref-type="bibr">74</xref>). Consequently, the animal effect observed in these data may be due, in part, to the technical limitations of the microarray platform used. Second, it is possible that the observed inter-animal differential gene expression is due, in part, to real differences in mRNA abundance generated by genotypic differences at loci that regulate gene expression (<xref rid="B71" ref-type="bibr">71</xref>, <xref rid="B74" ref-type="bibr">74</xref>, <xref rid="B75" ref-type="bibr">75</xref>); indeed, genotypic differences at loci that regulate gene expression in response to mycobacterial infection may contribute to phenotypic differences in the ability of an animal to clear or succumb to infection. In addition, a combination of both these technical and biological factors could also explain the observed animal effect.</p></sec><sec id="S9"><title>Comparative Functional Genomics Analysis Reveals Similarities and Differences in the Macrophage Transcriptome Response to <italic>M. bovis</italic>, MAP, and <italic>M. bovis</italic> BCG</title><p>Figure <xref ref-type="fig" rid="F4">4</xref> shows the results of differential gene expression analysis for the infected MDM (i.e., <italic>M. bovis</italic>, MAP, and <italic>M. bovis</italic> BCG) relative to the non-infected control MDM [false-discovery rate (FDR) adjusted <italic>P</italic> value ≤0.05]. Notably, for each infected MDM/control, MDM contrast the number of DE genes varied with respect to time. <italic>M. bovis</italic>-infected MDM exhibited the greatest number of DE genes, with the number of DE genes increasing across the 24-h time course. Furthermore, for <italic>M. bovis</italic>, the number of downregulated genes exceeded the number of upregulated genes at each of the time points post-infection.</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>The number of DE genes found for paired comparisons of infected MDM (i.e., MDM infected with <italic>M. bovis</italic>/MAP/<italic>M. bovis</italic> BCG) relative to the control MDM for each time point post-infection (FDR ≤ 0.05)</bold>. Bars above the horizontal line (i.e., <italic>y</italic> = 0) indicate the number of genes displaying upregulation in the infected MDM relative to the control MDM; bars below the horizontal line indicate the number of genes displaying downregulation in the infected MDM relative to the control MDM.</p></caption><graphic xlink:href="fimmu-05-00536-g004"/></fig><p>In contrast, the number of DE genes observed for both the MAP- and <italic>M. bovis</italic> BCG-infected MDM (relative to the non-infected controls) was highest at the 6 hpi time point for both sample groups (Figure <xref ref-type="fig" rid="F4">4</xref>). These results indicate that for MAP and <italic>M. bovis</italic> BCG infection, MDM differential gene expression had largely abated at the 24 hpi time point and that the MDM transcriptome reverted to a transcriptional state similar to that of the control MDM. These observations support previous work that showed that differential gene expression changes in MAP-infected bovine MDM are transient and are largely undetected 24 hpi relative to non-infected control MDM (<xref rid="B76" ref-type="bibr">76</xref>, <xref rid="B77" ref-type="bibr">77</xref>).</p><p>A comparison of the lists of DE genes obtained for all mycobacteria/control contrasts at each time point (Figure <xref ref-type="fig" rid="F5">5</xref>) identified a core set of DE genes at the 2 and 6 hpi time points common to all three mycobacterial treatments consisting of 170 and 236 DE genes, respectively. Among the DE genes common to all three mycobacterial infections were <italic>IL1A</italic>, <italic>IL1B</italic>, <italic>TNF</italic>, <italic>NFKB1</italic>, and <italic>NFKB2</italic>; all of these genes were upregulated at one or more time points in all types of infected MDM, suggesting a robust inflammatory reaction to all of the mycobacteria used in this study. Ingenuity<sup>®</sup> Systems Pathway Analysis (IPA; Ingenuity Systems, Redwood City, CA, USA; <uri xlink:type="simple" xlink:href="http://www.ingenuity.com">www.ingenuity.com</uri>) was used to identify canonical pathways within the list of DE genes that were common to all three mycobacterial treatments relative to the control group. The top ranking canonical cellular pathways (based on the lowest adjusted <italic>P</italic>-values; FDR ≤ 0.05) enriched for the 170 and 236 common DE genes identified at 2 and 6 hpi included <italic>IL-10 signaling</italic> (2 hpi); <italic>dendritic cell maturation</italic> (2 and 6 hpi); <italic>IL-6 signaling</italic> (2 hpi); <italic>TNFR2 signaling</italic> (2 hpi), <italic>TWEAK signaling</italic> (2 hpi); <italic>communication between innate and adaptive immune cells</italic> (6 hpi), <italic>NF-κB signaling</italic> (6 hpi); and <italic>TREM1 signaling</italic> (2 and 6 hpi; FDR ≤ 0.0001) (see Tables S1 and S2 in Supplementary Material).</p><fig id="F5" position="float"><label>Figure 5</label><caption><p><bold>The number of shared and unique DE genes among the three infected MDM groups based on gene expression data relative to the control MDM (FDR ≤ 0.05)</bold>. Red shading denotes <italic>M. bovis</italic>-infected MDM; green shading denotes MAP-infected MDM; blue shading denotes <italic>M. bovis</italic> BCG (BCG)-infected MDM.</p></caption><graphic xlink:href="fimmu-05-00536-g005"/></fig><p>We also observed large numbers of DE genes that were specific to <italic>M. bovis</italic>-infected MDM, which increased over time from 1,013 significant genes at 2 hpi to 2,167 at 24 hpi (Figure <xref ref-type="fig" rid="F5">5</xref>). The percentages of DE genes unique to <italic>M. bovis</italic> infection relative to the total number of DE genes detected following <italic>M. bovis</italic> infection were 80.3% (2 hpi), 78.8% (6 hpi), and 96.6% (24 hpi). IPA analysis of the DE genes unique to <italic>M. bovis</italic>-infected macrophages across all three time points (Tables S3–S5 in Supplementary Material) revealed enrichment for genes involved in cell signaling, including <italic>IL-6 signaling</italic> and <italic>mitogen-activated protein kinase (MAPK) signaling</italic>, the latter of which regulates the expression of several transcription factors, such as those encoded by <italic>FOS</italic> and <italic>JUN</italic> that are critical for the activation of immune cells (<xref rid="B78" ref-type="bibr">78</xref>). In contrast, the number of DE genes specific to MAP and <italic>M. bovis</italic> BCG infection across the time course was markedly lower; for example, no gene was specific to either MAP or <italic>M. bovis</italic> BCG infection 24 hpi. Indeed, for each post-infection time point, more than 98.3% of the DE genes induced by MAP and <italic>M. bovis</italic> BCG relative to the controls were among the list of DE genes induced by <italic>M. bovis</italic> relative to the controls.</p><p>We next analyzed differential gene expression directly between each pair of mycobacteria-infected sample groups (Figure <xref ref-type="fig" rid="F6">6</xref>). Large and increasing numbers of DE genes were found across the time course in <italic>M. bovis</italic>-infected MDM relative both MAP- and BCG-infected MDM, confirming the divergence between the two types of MDM transcriptional responses. Notably, no DE genes were detected between MAP- and BCG-infected MDM at 2 and 6 hpi, while only two DE genes were identified at 24 hpi; this contrasts with the larger numbers of DE genes identified through the indirect comparison of MAP- and BCG-infected MDM using the controls as a common reference (Figure <xref ref-type="fig" rid="F5">5</xref>). This discrepancy is most readily explained by differences in variances of gene expression within each sample group as illustrated in Figure S2 in Supplementary Material.</p><fig id="F6" position="float"><label>Figure 6</label><caption><p><bold>The number of DE genes found for pairwise comparisons among infected MDM groups (FDR ≤ 0.05)</bold>. Bars above the horizontal line (i.e., <italic>y</italic> = 0) indicate the number of genes displaying upregulation; bars below the horizontal line indicate the number of genes displaying downregulation.</p></caption><graphic xlink:href="fimmu-05-00536-g006"/></fig><p>The overlap in DE genes across all three treatment groups (relative to the control groups) at the 2 and 6 hpi time points suggest that a “core” MDM transcriptional response is induced by all three mycobacterial species/strains during the early stages of infection. This core transcriptome is characterized by genes involved in innate cytokine signaling and production, which encode proteins that activate the adaptive immune response following mycobacterial infection. However, the large and increasing number of DE genes specific to <italic>M. bovis</italic> across the infection time course demonstrates that <italic>M. bovis</italic> is a more potent inducer of proinflammatory genes than <italic>M. bovis</italic> BCG or MAP and highlights the distinct MDM gene expression profile elicited in response to this pathogen. In addition, the enrichment of <italic>M. bovis</italic>-specific DE genes for roles in macrophage cell signaling, such as IL-6 and MAPK signaling, suggests that additional cellular pathways are triggered by this pathogen relative to <italic>M. bovis</italic> BCG or MAP.</p><p>Although proinflammatory cytokines and chemokines play a pivotal role in mediating the host immune response to control mycobacterial infection, several lines of evidence suggest that these molecules and their associated pathways can be exploited by virulent mycobacterial pathogens to promote granuloma formation, which recruit new macrophages to the site of infection enabling persistence within the host (<xref rid="B79" ref-type="bibr">79</xref>). For example, non-regulated production of proinflammatory cytokines and chemokines can result in immunopathology, including destructive inflammation and necrosis, allowing dissemination of the pathogen from infected cells (<xref rid="B9" ref-type="bibr">9</xref>, <xref rid="B80" ref-type="bibr">80</xref>, <xref rid="B81" ref-type="bibr">81</xref>). Therefore, the immunopathology of BTB may be associated with the increased induction of innate immune genes following infection with <italic>M. bovis</italic>. Furthermore, the divergent transcriptomic profile observed in MDM infected with virulent <italic>M. bovis</italic> relative to <italic>M. bovis</italic> BCG-infected MDM is presumably governed, in part, by the presence of several secreted virulence factors, such as those encoded by the RD1 locus (<xref rid="B57" ref-type="bibr">57</xref>, <xref rid="B82" ref-type="bibr">82</xref>, <xref rid="B83" ref-type="bibr">83</xref>). This locus is present in all virulent strains of <italic>M. bovis</italic> (and <italic>M. tuberculosis</italic>) but is absent in attenuated strains of <italic>M. bovis</italic> BCG (<xref rid="B84" ref-type="bibr">84</xref>).</p><p>Conversely, the reduced number of DE genes detected between the MAP-infected and the non-infected control and <italic>M. bovis</italic> BCG-infected MDM across the time course suggests that MAP infection does not result in a major perturbation of the MDM transcriptome; rather, bovine MDM sense and respond to MAP in a similar manner to attenuated <italic>M. bovis</italic> BCG despite their markedly distinct evolutionary histories and different pathogenicities. These results support the hypothesis that MAP infection of host MDM is achieved via a capacity to appear “benign,” which enables it to reside and replicate within the macrophage for prolonged periods of time, and may underlie the lengthy subclinical phase of infection characteristic of JD (<xref rid="B85" ref-type="bibr">85</xref>). Our results also suggest that the immunoevasive mechanisms used by MAP involve the suppression of the proinflammatory response, such that the transcriptome of an infected macrophage resembles that of a non-infected cell. In support of this, <italic>in vivo</italic> MAP infection models demonstrate that cattle initially develop an early proinflammatory and T<sub>H</sub>1-type response to infection, which gradually declines in animals that progress to active disease, favoring a T<sub>H</sub>2-type response that does not control infection (<xref rid="B3" ref-type="bibr">3</xref>, <xref rid="B86" ref-type="bibr">86</xref>–<xref rid="B88" ref-type="bibr">88</xref>). Notably, the immunosuppression observed <italic>in vivo</italic> may originate at a cellular level, whereby MAP-infected macrophages fail to properly respond to host-derived immune activators such as CD40L and IFN-γ (<xref rid="B89" ref-type="bibr">89</xref>, <xref rid="B90" ref-type="bibr">90</xref>).</p></sec><sec id="S10"><title>Transcriptional Evidence for Type I Interferon-Mediated Regulation of Interleukin I Production in <italic>M. bovis</italic>-Infected MDM: Toward a Mechanism of Pathology</title><p>Examination of the lists of DE genes for each mycobacterial infection/control contrast shows that type I interferon-inducible genes such as <italic>IFIT1</italic>, <italic>IFIT2</italic>, <italic>MX1</italic>, <italic>MX2</italic>, and <italic>IL27</italic> and interferon-dependent <italic>CXCL10</italic> were not DE following MAP and <italic>M. bovis</italic> BCG infection at any of the post-infection time points. However, all of these genes were DE for at least one time point following <italic>M. bovis</italic> infection (Table <xref ref-type="table" rid="T1">1</xref>). Similarly, the gene encoding type II interferon (i.e., <italic>IFNG</italic>) was also upregulated at 6 and 24 hpi following <italic>M. bovis</italic> infection, but was not DE at any post-infection time point following infection with MAP and <italic>M. bovis</italic> BCG. We further observed that, in general, the fold-change of upregulation of the type I and type II interferon-inducible genes increased over the time course of infection in the <italic>M. bovis</italic>-infected MDM, with an accompanying decrease in the fold-change of upregulation of the interleukin-1 genes (<italic>IL1A</italic> and <italic>IL1B</italic>).</p><table-wrap id="T1" position="float"><label>Table 1</label><caption><p><bold>Log<sub>2</sub> fold-change values of manually curated genes for several immune-related pathways</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="left" rowspan="1" colspan="1"/><th align="center" colspan="3" rowspan="1"><italic>M. bovis</italic> vs CN<hr/></th><th align="center" colspan="3" rowspan="1"><italic>M. bovis</italic> BCG vs CN<hr/></th><th align="center" colspan="3" rowspan="1">MAP vs CN<hr/></th><th align="center" colspan="3" rowspan="1"><italic>M. bovis</italic> vs <italic>M. bovis</italic> BCG<hr/></th><th align="center" colspan="3" rowspan="1"><italic>M. bovis</italic> vs MAP<hr/></th></tr><tr><th align="left" rowspan="1" colspan="1"/><th align="left" rowspan="1" colspan="1">2 hpi</th><th align="left" rowspan="1" colspan="1">6 hpi</th><th align="left" rowspan="1" colspan="1">24 hpi</th><th align="left" rowspan="1" colspan="1">2 hpi</th><th align="left" rowspan="1" colspan="1">6 hpi</th><th align="left" rowspan="1" colspan="1">24 hpi</th><th align="left" rowspan="1" colspan="1">2 hpi</th><th align="left" rowspan="1" colspan="1">6 hpi</th><th align="left" rowspan="1" colspan="1">24 hpi</th><th align="left" rowspan="1" colspan="1">2 hpi</th><th align="left" rowspan="1" colspan="1">6 hpi</th><th align="left" rowspan="1" colspan="1">24 hpi</th><th align="left" rowspan="1" colspan="1">2 hpi</th><th align="left" rowspan="1" colspan="1">6 hpi</th><th align="left" rowspan="1" colspan="1">24 hpi</th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1"><bold>IL-1 signaling genes</bold></td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>IL1B</italic></td><td align="left" style="background-color:#F97173;" rowspan="1" colspan="1">5.68</td><td align="left" style="background-color:#FA9A9C;" rowspan="1" colspan="1">4.04</td><td align="left" style="background-color:#FBB7B9;" rowspan="1" colspan="1">2.85</td><td align="left" style="background-color:#FA9799;" rowspan="1" colspan="1">4.14</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FAA0A3;" rowspan="1" colspan="1">3.77</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FBD7DA;" rowspan="1" colspan="1">1.54</td><td align="left" style="background-color:#FBB8BB;" rowspan="1" colspan="1">2.80</td><td align="left" style="background-color:#FBD0D3;" rowspan="1" colspan="1">1.80</td><td align="left" style="background-color:#FBCED0;" rowspan="1" colspan="1">1.91</td><td align="left" style="background-color:#FBBABD;" rowspan="1" colspan="1">2.71</td><td align="left" style="background-color:#FBC7CA;" rowspan="1" colspan="1">2.18</td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>IL1A</italic></td><td align="left" style="background-color:#F9797B;" rowspan="1" colspan="1">5.37</td><td align="left" style="background-color:#FBC1C4;" rowspan="1" colspan="1">2.42</td><td align="left" style="background-color:#FBBFC2;" rowspan="1" colspan="1">2.50</td><td align="left" style="background-color:#FA9598;" rowspan="1" colspan="1">4.21</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FA9DA0;" rowspan="1" colspan="1">3.88</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FCE0E3;" rowspan="1" colspan="1">1.16</td><td align="left" style="background-color:#FBCED0;" rowspan="1" colspan="1">1.91</td><td align="left" style="background-color:#FBD4D7;" rowspan="1" colspan="1">1.64</td><td align="left" style="background-color:#FCD8DB;" rowspan="1" colspan="1">1.49</td><td align="left" style="background-color:#FBCFD2;" rowspan="1" colspan="1">1.85</td><td align="left" style="background-color:#FBC8CA;" rowspan="1" colspan="1">2.16</td></tr><tr><td align="left" rowspan="1" colspan="1"><bold>Type 1 interferon-related genes</bold></td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>IFNB1</italic></td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FCEAED;" rowspan="1" colspan="1">0.76</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FCE9EC;" rowspan="1" colspan="1">0.79</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FCECEF;" rowspan="1" colspan="1">0.66</td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>IFNAC</italic></td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1">0.17</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1">0.18</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1">0.18</td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>CXCL10</italic></td><td align="left" style="background-color:#FBCACD;" rowspan="1" colspan="1">2.04</td><td align="left" style="background-color:#FBB6B8;" rowspan="1" colspan="1">2.89</td><td align="left" style="background-color:#FAB3B5;" rowspan="1" colspan="1">3.00</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FBC9CC;" rowspan="1" colspan="1">2.10</td><td align="left" style="background-color:#FBCED0;" rowspan="1" colspan="1">1.92</td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>IFIT2</italic></td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FBC5C8;" rowspan="1" colspan="1">2.27</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FBCDD0;" rowspan="1" colspan="1">1.94</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FBD3D6;" rowspan="1" colspan="1">1.70</td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>MX1</italic></td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FCE0E2;" rowspan="1" colspan="1">1.18</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FCE2E5;" rowspan="1" colspan="1">1.07</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FCE8EA;" rowspan="1" colspan="1">0.85</td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>MX2</italic></td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FBCCCE;" rowspan="1" colspan="1">1.99</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FCD9DC;" rowspan="1" colspan="1">1.43</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FCDADD;" rowspan="1" colspan="1">1.41</td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>IFNAR1</italic></td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FCF4F7;" rowspan="1" colspan="1">0.33</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1"><italic>IFNAR2</italic></td><td align="left" style="background-color:#FCF8FB;" rowspan="1" colspan="1">0.20</td><td align="left" style="background-color:#FCF8FB;" rowspan="1" colspan="1">0.24</td><td align="left" style="background-color:#FCF8FB;" rowspan="1" colspan="1">0.25</td><td align="left" style="background-color:#FCF8FB;" rowspan="1" colspan="1">0.20</td><td align="left" style="background-color:#FCF8FB;" rowspan="1" colspan="1">0.21</td><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FCF8FB;" rowspan="1" colspan="1">0.19</td><td align="left" style="background-color:#FCF8FB;" rowspan="1" colspan="1">0.17</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FCF8FB;" rowspan="1" colspan="1">0.20</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FCF8FB;" rowspan="1" colspan="1">0.25</td></tr><tr><td align="left" rowspan="1" colspan="1"><bold>Type 2 interferon-related genes</bold></td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>IFNG</italic></td><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FBCACD;" rowspan="1" colspan="1">2.06</td><td align="left" style="background-color:#FBCED1;" rowspan="1" colspan="1">1.88</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FCD8DB;" rowspan="1" colspan="1">1.49</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FBD1D4;" rowspan="1" colspan="1">1.78</td><td align="left" style="background-color:#FBCBCE;" rowspan="1" colspan="1">2.02</td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>IFNGR2</italic></td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1"><italic>IFNGR2</italic></td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1">0.15</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1"><bold>NF-κB-related genes</bold></td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>NFKB1</italic></td><td align="left" style="background-color:#FCD8DB;" rowspan="1" colspan="1">1.47</td><td align="left" style="background-color:#FCDFE2;" rowspan="1" colspan="1">1.19</td><td align="left" style="background-color:#FCEDF0;" rowspan="1" colspan="1">0.61</td><td align="left" style="background-color:#FCE5E7;" rowspan="1" colspan="1">0.98</td><td align="left" style="background-color:#FCE5E7;" rowspan="1" colspan="1">0.89</td><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FCE5E7;" rowspan="1" colspan="1">1.00</td><td align="left" style="background-color:#FCEEF1;" rowspan="1" colspan="1">0.58</td><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FCEEF1;" rowspan="1" colspan="1">0.50</td><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FCEEF1;" rowspan="1" colspan="1">0.43</td><td align="left" style="background-color:#FCEEF1;" rowspan="1" colspan="1">0.48</td><td align="left" style="background-color:#FCEEF1;" rowspan="1" colspan="1">0.61</td><td align="left" style="background-color:#FCEEF1;" rowspan="1" colspan="1">0.53</td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>NFKB2</italic></td><td align="left" style="background-color:#FBBDC0;" rowspan="1" colspan="1">2.59</td><td align="left" style="background-color:#FAB1B4;" rowspan="1" colspan="1">3.07</td><td align="left" style="background-color:#FBCFD2;" rowspan="1" colspan="1">1.84</td><td align="left" style="background-color:#FBD2D5;" rowspan="1" colspan="1">1.74</td><td align="left" style="background-color:##FBC4C7;" rowspan="1" colspan="1">2.30</td><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FBD3D6;" rowspan="1" colspan="1">1.69</td><td align="left" style="background-color:#FBD3D6;" rowspan="1" colspan="1">1.87</td><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FCE8EB;" rowspan="1" colspan="1">0.85</td><td align="left" style="background-color:#FCE8EB;" rowspan="1" colspan="1">0.77</td><td align="left" style="background-color:#FCE1E3;" rowspan="1" colspan="1">1.14</td><td align="left" style="background-color:#FCE7E9;" rowspan="1" colspan="1">0.89</td><td align="left" style="background-color:#FCE1E3;" rowspan="1" colspan="1">1.20</td><td align="left" style="background-color:#FCE1E3;" rowspan="1" colspan="1">1.38</td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>TNF</italic></td><td align="left" style="background-color:#F97F81;" rowspan="1" colspan="1">5.11</td><td align="left" style="background-color:#FA9B9D;" rowspan="1" colspan="1">3.98</td><td align="left" style="background-color:#FBC4C7;" rowspan="1" colspan="1">2.30</td><td align="left" style="background-color:#FA989A;" rowspan="1" colspan="1">4.11</td><td align="left" style="background-color:#FCDADD;" rowspan="1" colspan="1">1.39</td><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FAA6A8;" rowspan="1" colspan="1">3.55</td><td align="left" style="background-color:#FCD8DB;" rowspan="1" colspan="1">1.47</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FBBDC0;" rowspan="1" colspan="1">2.59</td><td align="left" style="background-color:#FBD7DA;" rowspan="1" colspan="1">1.52</td><td align="left" style="background-color:#FBD7DA;" rowspan="1" colspan="1">1.57</td><td align="left" style="background-color:#FBBFC2;" rowspan="1" colspan="1">2.51</td><td align="left" style="background-color:#FBCBCD;" rowspan="1" colspan="1">2.04</td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>IL6</italic></td><td align="left" style="background-color:#FA8E91;" rowspan="1" colspan="1">4.49</td><td align="left" style="background-color:#FBC8CA;" rowspan="1" colspan="1">2.15</td><td align="left" style="background-color:#FAB0B3;" rowspan="1" colspan="1">3.12</td><td align="left" style="background-color:#FAB0B3;" rowspan="1" colspan="1">2.86</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FAB0B3;" rowspan="1" colspan="1">2.40</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/><td align="left" style="background-color:#FBD4D7;" rowspan="1" colspan="1">1.63</td><td align="left" style="background-color:#FBD4D7;" rowspan="1" colspan="1">1.67</td><td align="left" style="background-color:#FBC7C9;" rowspan="1" colspan="1">2.20</td><td align="left" style="background-color:#FBC7C9;" rowspan="1" colspan="1">2.09</td><td align="left" style="background-color:#FBD6D9;" rowspan="1" colspan="1">1.57</td><td align="left" style="background-color:#FBBABC;" rowspan="1" colspan="1">2.73</td></tr></tbody></table><table-wrap-foot><p><italic>Cells containing values are statistically different for the contrasts listed in the column headings (FDR ≤ 0.05); empty cells are not statistically different (FDR ≥ 0.05). The intensity of the shading is related to relative increased fold-changes in gene expression</italic>.</p></table-wrap-foot></table-wrap><p>IL-1 and type I IFN-signaling pathways have been shown to play important, yet opposing, roles in determining the host response to infection with virulent members of the MTBC. Mice deficient in IL-1B display increased susceptibility to virulent <italic>M. tuberculosis</italic>, indicating that IL-1 signaling is required for the host control of infection (<xref rid="B91" ref-type="bibr">91</xref>, <xref rid="B92" ref-type="bibr">92</xref>). Conversely, mice deficient in type I IFN signaling show reduced bacterial loads following infection, suggesting that type I IFN plays a contributory role in tuberculosis disease progression (<xref rid="B93" ref-type="bibr">93</xref>, <xref rid="B94" ref-type="bibr">94</xref>). Studies have also shown that virulent <italic>M. tuberculosis</italic> and attenuated <italic>M. bovis</italic> BCG use distinct signaling pathways for regulating IL-1B production in human MDM. <italic>M. tuberculosis</italic> induced the expression of IFN-related genes, while induction of type I IFN-signaling inhibited IL-1B secretion (<xref rid="B95" ref-type="bibr">95</xref>, <xref rid="B96" ref-type="bibr">96</xref>). Notably, infection of human MDM with <italic>M. bovis</italic> BCG did not induce significant differential expression of IFN-related genes or IL-1B secretion. These results suggest that type I IFN-mediated suppression of IL-1B production is a key mechanism for the intracellular survival of <italic>M. tuberculosis</italic> (<xref rid="B95" ref-type="bibr">95</xref>, <xref rid="B96" ref-type="bibr">96</xref>). Comparable <italic>in vivo</italic> and <italic>in vitro</italic> studies of bovine ileal tissue and MDM have demonstrated increased transcript and protein levels of IL-1A and IL-1B (<italic>in vitro</italic> only) in response to MAP infections, with a concomitant increase in downstream expression of TRAF1 (<xref rid="B97" ref-type="bibr">97</xref>, <xref rid="B98" ref-type="bibr">98</xref>). This increase in TRAF1 has been proposed to enhance survival of MAP in macrophages due to the anti-apoptotic properties of TRAF1 and its capacity to interfere with normal macrophage activation, particularly via CD40/CD40L (<xref rid="B98" ref-type="bibr">98</xref>, <xref rid="B99" ref-type="bibr">99</xref>).</p><p>The increase in type I IFN gene expression and the concomitant decrease in <italic>IL1</italic> induction that we observed in <italic>M. bovis</italic>-infected MDM over the 24-h infection time course suggests that the intracellular survival strategy of virulent <italic>M. bovis</italic> may also involve type I IFN-mediated suppression of <italic>IL1</italic> production. In contrast to this, these results suggest that the same immunoevasive mechanism is not used by virulent MAP or the attenuated <italic>M. bovis</italic> BCG in bovine MDM. These results indicate that differential activation of macrophage immunoregulatory pathways is central to the differential intracellular survival mechanisms of these related, yet distinct, bovine mycobacterial pathogens. A proposed model of the differential response of bovine MDM to the mycobacteria examined in this comparative study is shown in Figure <xref ref-type="fig" rid="F7">7</xref>.</p><fig id="F7" position="float"><label>Figure 7</label><caption><p><bold>A proposed model for the differential responses of bovine MDM to <italic>M. bovis</italic>, <italic>M. bovis</italic> BCG, and <italic>M. avium</italic> subsp. <italic>paratuberculosis</italic> infection <italic>in vitro</italic></bold>. Differential integration of common and specific signals induced by the three mycobacterial types is shown.</p></caption><graphic xlink:href="fimmu-05-00536-g007"/></fig></sec><sec id="S11"><title>Differential Mycobacterial Virulence Factors and Their Impact on Macrophage Pathogen Recognition</title><p>The region of difference 1 (RD1) locus, which is present in <italic>M. bovis</italic>, is a major genetic difference between this species and MAP and <italic>M. bovis</italic> BCG, which both lack RD1 (<xref rid="B100" ref-type="bibr">100</xref>). Moreover, RD1 (which is also present in virulent <italic>M. tuberculosis</italic> strains) has attracted increasing interest over the last two decades, because it contains the ESX-1 type VII secretion system, responsible for the secretion of virulence factors, such as the dimer ESAT-6/CFP-10 (<xref rid="B101" ref-type="bibr">101</xref>). ESAT-6, which is proposed to facilitate escape from macrophage phagolysosomal degradation, binds to TLR2 receptors, and activates TLR signaling cascades within the macrophage that culminate in cytokine production (<xref rid="B102" ref-type="bibr">102</xref>–<xref rid="B104" ref-type="bibr">104</xref>). It has been recently reported that TLR2 receptors mediate enhanced interferon production through reprograming of murine macrophages following infection with viral ligands (<xref rid="B105" ref-type="bibr">105</xref>). In support of these observations, our results showed that differential expression of both type I and type II interferon genes is unique to <italic>M. bovis</italic>-infected MDM, which is not present in MDM infected with MAP and <italic>M. bovis</italic> BCG. Consequently, we hypothesize that the virulence factors encoded in the RD1 region and secreted by <italic>M. bovis</italic> – but not MAP or <italic>M. bovis</italic> BCG – trigger an additional cascade of signaling events, such as those mediated by TLRs and MAPKs (as revealed through IPA analyses of unique <italic>M. bovis</italic>-induced genes), relative to attenuated <italic>M. bovis</italic> BCG and the lengthy subclinical MAP. In turn, the combined activation of additional immune pathways and canonical PRR-dependent pathways may lead to a sustained (i.e., chronic) inflammatory response in infected macrophages, as opposed to a more transient inflammation following <italic>M. bovis</italic> BCG or MAP infections.</p></sec><sec id="S12"><title>Concluding Remarks</title><p>In the present study, we highlight markedly different transcriptional response of bovine MDM infected with <italic>M. bovis</italic> over a 24-h time course compared to the closely related but attenuated <italic>M. bovis</italic> BCG strain and to virulent MAP. We hypothesize that RD1-encoded virulence factors provide a mechanistic basis for this differential response, as RD1 was lost during the derivation of the <italic>M. bovis</italic> BCG vaccine strain from <italic>M. bovis</italic> and is absent from the MAP genome. We also identified a common MDM transcriptional response to both attenuated <italic>M. bovis</italic> BCG and MAP. We propose that the respective attenuated and lengthy subclinical phenotypes of <italic>M. bovis</italic> BCG and MAP may induce similar responses in infected macrophages, at least during the early stages of infection. Finally, we identified type I interferon-dependent genes among the DE genes specific to virulent <italic>M. bovis-</italic>infected MDM, adding further evidence supporting a key role for type I interferon in the establishment of active tuberculosis in cattle as has previously reported for human tuberculosis (<xref rid="B106" ref-type="bibr">106</xref>).</p><p>While the comparative functional genomics analysis presented here is based on data generated from microarrays, the changing landscape of transcriptomics, as represented by the advent of high-throughput RNA-seq, offers unprecedented opportunities to study the host macrophage response to mycobacterial infection at the nucleotide level, including investigation of the effect of genotype on gene expression levels. High-throughput sequencing technologies are providing novel insights into the cellular mechanisms governing mycobacteria–macrophage interactions, enabling further understanding of how modulation of these pathways can result in pathology. In addition, identification of transcriptional biomarkers of infection may lead to the development of novel diagnostics for BTB and JD, providing new molecular tools for disease control and eradication.</p></sec><sec id="S13"><title>Conflict of Interest Statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.The Guest Associate Editor Kieran G. Meade declares that, despite having collaborated with authors Kévin Rue-Albrecht, David A. Magee, Kate E. Killick, Nicolas C. Nalpas, Stephen V. Gordon and David E. MacHugh, the review process was handled objectively and no conflict of interest exists.</p></sec><sec sec-type="supplementary-material" id="S14"><title>Supplementary Material</title><p>The Supplementary Material for this article can be found online at <uri xlink:type="simple" xlink:href="http://www.frontiersin.org/Journal/10.3389/fimmu.2014.00536/abstract">http://www.frontiersin.org/Journal/10.3389/fimmu.2014.00536/abstract</uri></p><supplementary-material content-type="local-data" id="SM1"><media xlink:href="Table_1.XLS"><caption><p>Click here for additional data file.</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="SM2"><media xlink:href="Table_2.XLS"><caption><p>Click here for additional data file.</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="SM3"><media xlink:href="Table_3.XLS"><caption><p>Click here for additional data file.</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="SM4"><media xlink:href="Table_4.XLS"><caption><p>Click here for additional data file.</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="SM5"><media xlink:href="Table_5.XLS"><caption><p>Click here for additional data file.</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="SM6"><media xlink:href="Data_Sheet_1.DOCX"><caption><p>Click here for additional data file.</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="SM7"><media xlink:href="Image_1.TIF"><caption><p>Click here for additional data file.</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="SM8"><media xlink:href="Image_2.TIF"><caption><p>Click here for additional data file.</p></caption></media></supplementary-material></sec>
Priming of CD8<sup>+</sup> T Cell Responses to Liver Stage Malaria Parasite Antigens	<p>While the role of malaria parasite-specific memory CD8<sup>+</sup> T cells in the control of exo-erythrocytic stages of malaria infection is well documented and generally accepted, a debate is still ongoing regarding both the identity of the anatomic site where the activation of naive pathogen-specific T cells is taking place and contribution of different antigen-presenting cells (APCs) into this process. Whereas some studies infer a role of professional APCs present in the lymph nodes draining the site of parasite injection by the mosquito, others argue in favor of the liver as a primary organ and hepatocytes as stimulators of naïve parasite-specific T cell responses. This review aims to critically analyze the current knowledge and outline new lines of research necessary to understand the induction of protective cellular immunity against the malaria parasite.</p>	<contrib contrib-type="author"><name><surname>Corradin</surname><given-names>Giampietro</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="corresp" rid="cor1">*</xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/137013"/></contrib><contrib contrib-type="author"><name><surname>Levitskaya</surname><given-names>Jelena</given-names></name><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/190628"/></contrib>	Frontiers in Immunology	<sec sec-type="intro" id="S1"><title>Introduction</title><p>It is currently accepted that priming of CD8<sup>+</sup> T lymphocytes by antigen-presenting cells (APCs) takes place in the secondary lymphoid organs such as spleen and lymph nodes [reviewed in Ref. (<xref rid="B1" ref-type="bibr">1</xref>)]. Multi-photon-based intravital microscopy revealed that the first contact between naïve CD8<sup>+</sup> T cells and APC takes place in the periphery of draining lymph nodes (DLN) shortly after infection and mainly occurs in the subcapsular sinus or the interfollicular regions enriched with pathogen-derived antigens (<xref rid="B2" ref-type="bibr">2</xref>, <xref rid="B3" ref-type="bibr">3</xref>). Depending on the pathogen’s nature, the rapid relocation of naïve T cells to the periphery of the draining lymph node can be either antigen-specific (<xref rid="B2" ref-type="bibr">2</xref>) or antigen-independent, associated with decreased local levels of chemokines and the drastic alteration of the lymph node architecture by the pathogen (<xref rid="B3" ref-type="bibr">3</xref>, <xref rid="B4" ref-type="bibr">4</xref>). Data from mice infected with vesicular stomatitis virus demonstrated that, though CD169+ macrophages (<xref rid="B5" ref-type="bibr">5</xref>) residing in the subcapsular sinus were the major cell population bearing virus-derived antigens (<xref rid="B2" ref-type="bibr">2</xref>), dendritic cells (DCs) served as the primary APC triggering antigen-specific naïve CD8<sup>+</sup> T cells. The ability of immature dendritic cells to acquire exogenous antigens followed by their proteolytic processing and presentation on the MHC class I molecules, commonly referred to as “cross-presentation,” is believed to be the major requirement for the generation of primary antigen-specific CD8<sup>+</sup> T cell responses against pathogens (<xref rid="B6" ref-type="bibr">6</xref>–<xref rid="B9" ref-type="bibr">9</xref>).</p><p>Upon the initial encounter of naïve T cells with APC, a heterogeneous progeny of antigen-specific CD8<sup>+</sup> T cells including short-lived effector cells (SLEC) and memory precursor effector cells (MPEC) [reviewed in Ref. (<xref rid="B10" ref-type="bibr">10</xref>, <xref rid="B11" ref-type="bibr">11</xref>)] is generated. It is still not clear whether the SLEC versus MPEC differentiation is enforced by the asymmetric segregation of transcription factors and protein degradation machinery already at the first cell division (<xref rid="B12" ref-type="bibr">12</xref>–<xref rid="B14" ref-type="bibr">14</xref>) or it reflects the differential exposure to inflammatory and co-stimulatory “help” signals received from APCs by antigen-specific CD8<sup>+</sup> T cells during the expansion phase [reviewed in Ref. (<xref rid="B10" ref-type="bibr">10</xref>, <xref rid="B11" ref-type="bibr">11</xref>)]. While generation of primary CD8<sup>+</sup> T cell responses to non-inflammatory antigens requires CD4<sup>+</sup> T cell help, induction of primary CD8<sup>+</sup> T cells responses to <italic>Listeria</italic>, LCMV, and influenza virus is CD4<sup>+</sup> T cell-independent and results from direct activation of APCs by the pathogen (<xref rid="B15" ref-type="bibr">15</xref>–<xref rid="B17" ref-type="bibr">17</xref>). Moreover, CD4<sup>+</sup> T cell help can be replaced by the CD40 triggering on the DCs, which prime antigen-specific naïve CD8<sup>+</sup> T cells (<xref rid="B18" ref-type="bibr">18</xref>, <xref rid="B19" ref-type="bibr">19</xref>). Thus, the exact nature and requirements for “help” signals necessary for the initial triggering and subsequent expansion of primary antigen-specific CD8<sup>+</sup> T cell responses vary among different pathogens and sites of primary infection. In this report, our objective is to present and discuss the published data regarding CD8<sup>+</sup> T cell activation in <italic>Plasmodium</italic> infection, and suggest experiments to better understand the antigen presentation process.</p><p>Malaria infection is initiated through the bites by <italic>Plasmodium</italic>-carrying female Anopheles searching for blood to support egg development. As the mosquito probes the host environment under the skin for the presence of blood vessels, it injects salivary gland proteins both prior and during blood feeding to inhibit blood coagulation. Parasites deposited into the skin can also traverse surrounding cells and enter the circulation with subsequent infection of liver cells. Studies performed with parasites injected intradermally or intravenously show that the resulting liver parasite load is similar (<xref rid="B20" ref-type="bibr">20</xref>). In addition, transfer of parasites from the skin sites to DLN occurs (<xref rid="B21" ref-type="bibr">21</xref>).</p><p>Identification of the anatomical site and the type of APC, which orchestrate the induction of primary CD8<sup>+</sup> T cell responses against a particular antigen, represents an essential step in rational design of CD8<sup>+</sup> T cell-based vaccination strategies. Whereas the research on the effector phase of CD8<sup>+</sup> T cell response against malaria has been quite extensive (<xref rid="B22" ref-type="bibr">22</xref>–<xref rid="B25" ref-type="bibr">25</xref>), a rather limited number of studies attempted to dissect the issue of liver stage-specific CD8<sup>+</sup> T cell priming in the infected host.</p></sec><sec id="S2"><title>Role of Different Organs in Antigen Presentation</title><p>In this respect, the study by Chakravarty et al. (<xref rid="B21" ref-type="bibr">21</xref>) appears to be one of the most comprehensive and systematic up to date. The authors concluded that extrahepatic lymphoid tissues, in particular the DLN and spleen are the most important sites contributing to the generation of the effector T-cell pool in the liver. In agreement with these data, Obeid and colleagues demonstrated that strictly subcutaneous immunization with irradiated sporozoites led to induction of sterile immunity against pre-erythrocytic malaria with T cell priming occurring in skin-draining lymph node (<xref rid="B26" ref-type="bibr">26</xref>). It was proposed that parasite-specific CD8<sup>+</sup> T cell priming depends on cross-presentation of malaria antigens (<xref rid="B21" ref-type="bibr">21</xref>). This indicates that professional APC, rather than infected hepatocytes, trigger priming of naïve CD8<sup>+</sup> T cells directed to liver stage antigens.</p><p>Several lines of experimental evidence were presented in support of these conclusions. Thus, IFNγ production by adoptively transferred circumsporozoite protein (CSP)-specific naïve transgenic T cells was first detected in the skin-DLN as early as on day 2 after mouse immunization by microinjection or mosquito bites, whereas no detectable T cell activation was detected in other organs including spleen. Hence, Chakravarty and co-authors suggested that these temporal differences in the onset of parasite-specific T cell activation could reflect the hierarchical order of T cell priming initiated in the DLN that could be followed by migration of primed CD8<sup>+</sup> T cells to other organs, including the spleen and the liver. However, removal of lymph nodes draining the site of parasite injection prior to the adoptive transfer of parasite-specific CD8<sup>+</sup> T cells, though resulted in a 60% reduction of activated CD8<sup>+</sup> T cells in the liver, did not affect the frequencies of primed CD8<sup>+</sup> T cells in the spleen where the first signs of T cells activation were documented only 24 h later than in DLN and at the same time point as in the liver. These data indicate that temporal differences in the onset of T cell activation used as a parameter for identification of the CD8<sup>+</sup> T cell priming site should be carefully reconsidered in future studies and further strengthen the importance of the spleen as a site of induction of primary CD8<sup>+</sup> T cell responses in animal models of the infection. The latter is in agreement with the data by Sano et al. (<xref rid="B27" ref-type="bibr">27</xref>) demonstrating that spleens of infected mice support priming of parasite-specific naïve CD8<sup>+</sup> T cells following intravenous injection of sporozoites.</p><p>At the same time, several lines of evidence presented by Chakravarty and colleagues (<xref rid="B21" ref-type="bibr">21</xref>) do not firmly support the essential role of the spleen in the parasite-specific CD8<sup>+</sup> T cell priming.</p><p>First, DCs isolated from the spleens 60 h after injection of sporozoites were unable to trigger proliferation of parasite-specific CD8<sup>+</sup> T cells, whereas DCs isolated from the DLN efficiently induced T cell proliferation and, presumably, presented the antigen. Since no data with liver-resident DCs were generated, a direct role of intrahepatic professional APCs in priming of parasite-specific CD8<sup>+</sup> T cells still needs to be addressed. In addition, as the first signs of activation of parasite-specific T cells in DLN were detected at day 2 post immunization, it is not completely clear whether DCs from DLN had a greater capacity to prime CD8<sup>+</sup> T cells as compared to spleen and liver-resident DCs at time points earlier than 60 h.</p><p>Second, animals subjected to simultaneous lymphadenectomy and splenectomy prior to the adoptive transfer of CSP-specific CD8<sup>+</sup> T cells followed by immunization with sporozoites and subsequent challenge with viable parasites 10 days later had similar load of parasites in the liver as non-immunized mice, indicating that either DLN or/and spleen are required for CSP-specific CD8<sup>+</sup> T cell priming. At the same time, splenectomy alone did not affect inhibition of parasite development in the liver, prompting the authors to conclude, that DLNs are the priming site of protective CD8<sup>+</sup> T cell responses.</p><p>Interestingly, as shown by Chakravarty and co-authors (<xref rid="B21" ref-type="bibr">21</xref>), removal of both DLNs and the spleen prior to immunization with sporozoites, though drastically reduced the activated T cell pool in the liver, failed to abrogate it completely, suggesting that at least a proportion of parasite-specific CD8<sup>+</sup> T cells found in the liver had been primed outside the DLN and the spleen. These findings could reflect the process of parasite-specific CD8<sup>+</sup> T cell triggering in the liver and define it as the organ essential for the parasite development. On the other hand, animals treated with FTY720, a drug, which inhibits lymphocyte egress from lymph nodes (<xref rid="B28" ref-type="bibr">28</xref>, <xref rid="B29" ref-type="bibr">29</xref>), had substantially less IFN gamma producing parasite-specific CD8<sup>+</sup> T cells at day 7 post injection with irradiated sporozoites. Based on this observation, the authors concluded that systemic distribution of CD8<sup>+</sup> T cells, at least in part, contributes to the intrahepatic pool of parasite-specific CD8<sup>+</sup> T cells (<xref rid="B21" ref-type="bibr">21</xref>). It still needs to be seen, if treatment with FTY720 (<xref rid="B30" ref-type="bibr">30</xref>) inhibits the development of “early-primed” parasite-specific CD8<sup>+</sup> T cells in the liver and spleen, previously noted by the authors already 72 h after mosquito bite. In addition, effect of FTY720 on the protection of animals from subsequent challenge with live sporozoites has to be addressed in this model. Noteworthy, the time course of the parasite-specific clonal T cell activation in the lymph nodes, liver, and other organs is only slightly delayed (by 24 h) while it is known that activated T cells egress from the lymph nodes 4–5 days after antigen encounter (<xref rid="B31" ref-type="bibr">31</xref>, <xref rid="B32" ref-type="bibr">32</xref>). The latter suggests that either activation of parasite-specific T cells may take place simultaneously in various organs, or unusually rapid egress from the lymph node after priming is an intrinsic feature of T cells in this specific experimental model.</p></sec><sec id="S3"><title>Role of Infected Hepatocytes in Antigen Presentation</title><p>The role of infected hepatocytes in direct priming of naïve parasite-specific CD8<sup>+</sup> T cells is still a subject of controversy. Early study by Renia et al. demonstrated that intrasplenic injection of infected hepatocytes induced protective T cell-mediated immunity against infection with <italic>Plasmodium yoelii</italic> and <italic>P. berghei</italic> sporozoites (<xref rid="B33" ref-type="bibr">33</xref>). Leiriao et al. demonstrated that apoptotic hepatocytes infected with irradiated sporozoites are phagocytosed by DCs and merely serve as a source of <italic>Plasmodium</italic> antigens for the initiation of the protective immune responses via cross-priming (<xref rid="B34" ref-type="bibr">34</xref>). In contrast, Renia and collaborators argued against apoptotic infected hepatocytes as a source of antigens and suggested that liver DCs could be activated upon uptake of parasite antigens directly from viable infected hepatocytes (<xref rid="B35" ref-type="bibr">35</xref>) as previously seen in other experimental models (<xref rid="B36" ref-type="bibr">36</xref>, <xref rid="B37" ref-type="bibr">37</xref>). However, data from Chakravarty et al. implied that though cross-priming is required, it takes place in the DLNs and not in the liver (<xref rid="B21" ref-type="bibr">21</xref>). In agreement with these data, Jung et al. demonstrated that mice subjected to chemical depletion of CD11c<sup>+</sup> DCs fail to induce CD8<sup>+</sup> T cell responses to infection with <italic>Plasmodium yoelii</italic> (<xref rid="B38" ref-type="bibr">38</xref>). Neither of these studies considered hepatocytes as an APC subset capable of initiating the primary parasite-specific T cell responses.</p><p>A recent study by Balam et al. (<xref rid="B39" ref-type="bibr">39</xref>) focused on two questions: can infected hepatocytes directly prime naïve parasite-specific T cells and does stimulation of already primed CD8<sup>+</sup> T cells protect mice against parasite challenge? Administration of CD8<sup>+</sup> CSP-specific T cells but not an irrelevant T cell clone injected into TAP-deficient MHC class I mismatched recipient mice, simultaneously with infected hepatocytes bearing MHC haplotype relevant for parasite-specific T cells, resulted in 100% protection of mice from subsequent challenge with live sporozoites (<xref rid="B39" ref-type="bibr">39</xref>). As the observed protection was not due to a bystander effect or a continuous cytokine secretion by parasite-specific CD8<sup>+</sup> T cells, these data demonstrate that infected hepatocytes are capable of presenting the antigen to CD8<sup>+</sup> T cells, reactivating resting CSP-specific CD8<sup>+</sup> T cells and inducing protection.</p><p>Importantly, more than 60% of naïve BALB/c mice injected with irradiated sporozoite-infected hepatocytes were also protected from subsequent live parasite challenge, suggesting that infected hepatocytes could contribute to the priming of endogenous naïve T cell. However, T cell depletion experiments are required to confirm that protection is T cell-mediated. Finally, to formally exclude contamination with other APC potentially present in the hepatocyte preparations and capable of presenting CSP and priming the naïve CD8<sup>+</sup> T cells, isolation of pure hepatocyte population devoid of cells bearing markers of DCs, macrophages, and stellate cells should be done by flow cytometry using fluorescent transgenic parasites. On the other hand, arguing against the sole role of professional APC in priming of naïve immune responses to malaria parasites, mice depleted of DCs by treatment with cytochrome <italic>c</italic> were still protected from the challenge with live sporozoites in spite of significantly lower frequencies of endogenous parasite-specific T cells primed by the immunization with irradiated sporozoites (<xref rid="B39" ref-type="bibr">39</xref>). These data do not fully support the previously discussed role of dendritic cell function in induction of primary malaria liver stage-specific T cell responses (<xref rid="B21" ref-type="bibr">21</xref>, <xref rid="B38" ref-type="bibr">38</xref>).</p></sec><sec id="S4"><title>Other Considerations</title><p>The quality of hepatocytes as APCs capable of triggering T cells responses had been recently dissected by Ma et al. (<xref rid="B40" ref-type="bibr">40</xref>). It had been demonstrated that <italic>P. berghei</italic> and <italic>P. falciparum</italic> infected human hepatocytes retain largely unaltered expression of multiple molecules of the MHC class I pathway until very late stages of parasite development (<xref rid="B40" ref-type="bibr">40</xref>). Moreover, infected cells exhibited no obvious defects in the capacity to upregulate expression of different molecular components of the MHC class I machinery in response to pro-inflammatory lymphokines or trigger direct activation of allo-specific as well as peptide-specific human CD8<sup>+</sup> T cells (<xref rid="B40" ref-type="bibr">40</xref>). At the same time, it is not known whether or not the characteristic features of professional APC believed to be important for efficient T-cell priming, i.e., co-stimulatory molecules B7.1 and B7.2 (“signal 2”), as well as production of cytokines essential for the survival and maintenance of primed T cells (“signal 3”) are possessed by the primary human hepatocytes <italic>in vivo</italic> and/or induced upon infection.</p><p>Current literature dissecting the ability of primary hepatocytes to specifically prime naïve CD8<sup>+</sup> T cells is scarce. Bertolino et al. demonstrated that purified primary murine hepatocytes were able to induce activation and proliferation of antigen-specific naive CD8<sup>+</sup> T cells <italic>in vitro</italic>, even in the absence of exogenously added cytokines as well as CD80 and CD86 co-stimulatory molecules (<xref rid="B41" ref-type="bibr">41</xref>). Moreover, the magnitude of T cell proliferation induced by primary hepatocytes was comparable to that induced by DCs. Naïve T cell priming by hepatocytes did not require CD4<sup>+</sup> T cell help and induced expression of early T cell activation markers and transient CD8<sup>+</sup> T cell effector activity followed by rapid cell death of activated T cells. Thus, primary hepatocytes were able to prime naïve T cells but failed to sustain productive antigen-specific CD8<sup>+</sup> T cell responses (<xref rid="B41" ref-type="bibr">41</xref>). In agreement with these data, <italic>in vivo</italic> experiments using endogenous expression of alloantigens under hepatocyte-specific promoters demonstrated that activation of primary T cells by hepatocytes as antigen-presenting cells leads to T cell apoptosis rather than formation of antigen-specific memory T cell pool (<xref rid="B42" ref-type="bibr">42</xref>–<xref rid="B44" ref-type="bibr">44</xref>). It was further demonstrated that T cells activated by hepatocytes died “by neglect” and lack of IL-2 and low expression of pro-survival genes due to insufficient co-stimulation during the priming phase (<xref rid="B45" ref-type="bibr">45</xref>). Hence, taking into account the inability of primary hepatocytes to provide appropriate co-stimulation during T cells priming along with the immunosuppressive microenvironment created by multiple subsets of the liver-resident APC [reviewed in Ref. (<xref rid="B46" ref-type="bibr">46</xref>, <xref rid="B47" ref-type="bibr">47</xref>)], it may appear unlikely that hepatocytes infected with malaria parasites play a major role in the generation of effective parasite-specific CD8<sup>+</sup> T cell memory responses. However, it does not preclude the possibility that CD8<sup>+</sup> T cells specific to malaria antigens could be primed and activated, at least shortly, by hepatocytes supporting development of exo-erythrocytic forms. Indeed, given proper stimuli, such T cells can be rescued to full immunological competence and longer survival (<xref rid="B48" ref-type="bibr">48</xref>, <xref rid="B49" ref-type="bibr">49</xref>). In the case of malaria, proper activation stimuli could be induced by <italic>Plasmodium</italic> infection leading to activation of numerous genes in hepatocytes (<xref rid="B50" ref-type="bibr">50</xref>, <xref rid="B51" ref-type="bibr">51</xref>) including those involved in native immunity and antigen presentation. Since no transcriptional analysis has been performed in Kupffer cells traversed by sporozoites so far, it would be important to understand whether or not liver-resident macrophages change their immunomodulatory properties in the site of malaria infection.</p><p>At this point, a word of caution should be expressed to the fact that all animal studies discussed above were based on a single mouse strain, BALB/c, as well as a single CD8<sup>+</sup> T cell epitope derived from the CSP. Future studies on the induction of primary T cell responses to exo-erythrocytic forms of malaria need to be extended to other protective CD8<sup>+</sup> T cell epitopes including responses, which appear later in the liver stage by using either radiation attenuated (RAS) or genetically attenuated (GAS) sporozoites or sporozoites combined with chloroquine chemoprophylaxis (CPS) (<xref rid="B52" ref-type="bibr">52</xref>).</p></sec><sec id="S5"><title>Field Studies</title><p>It is still unclear to what extend animal models dissecting induction of primary T cell responses to malaria as well as human studies involving vaccinated volunteers reflect the acquisition of natural cellular immune responses in malaria endemic areas. An acquisition of a sterile immune protection following immunization with RAS, GAS, or using CPS regime in animals and humans sharply contrasts with the situation in the field, where, in spite of frequent (up to 30 per month in certain areas) biting by infected mosquitoes (<xref rid="B53" ref-type="bibr">53</xref>, <xref rid="B54" ref-type="bibr">54</xref>), no sterile protection is usually obtained in both adults and children in response to natural infection under drug treatment or intermittent preventive treatment (IPT). Several hypotheses could be proposed to explain this discrepancy: (1) sporozoite “charge” (and, as a result, supply of parasite antigens) is too small in the field as compared to that given under experimental conditions; (2) down-regulation of the parasite-specific CD8<sup>+</sup> T cell responses by the content of mosquito salivary glands delivered together with sporozoites to the site of T cell priming; and (3) excessive and/or preceding induction of immune responses to salivary gland proteins. As for the latter, given the fact that only a fraction (0–25% depending on the seasons and location) of mosquitoes are infected (<xref rid="B54" ref-type="bibr">54</xref>–<xref rid="B56" ref-type="bibr">56</xref>), a memory T-cell pool specific for salivary gland antigens is most likely established prior to parasite infection. As a result, secondary T cell responses directed to mosquito antigens could be preferentially activated at the expense of parasite-specific T-cell activation via, for example, competition for IL-2, homeostatic niche, or by active secretion of inhibitory molecules (<xref rid="B57" ref-type="bibr">57</xref>–<xref rid="B64" ref-type="bibr">64</xref>). If this is the case, the efficacy of sporozoite-based pre-erythrocytic vaccines may turn out to be low in endemic areas due to even subtle contamination with the salivary gland proteins. Both the second and third hypotheses may explain the low frequency of parasite-specific helper and CD8<sup>+</sup> T cell found in humans from malaria endemic regions, as well as the general failure [with the exception for a single donor so far (<xref rid="B65" ref-type="bibr">65</xref>)] to obtain stable human T-cell clones specific to malaria liver stage antigens.</p></sec><sec id="S6"><title>Final Remarks</title><p>In conclusion, existing experimental data obtained from animal models suggest that: (1) both DCs and hepatocytes can prime naïve malaria parasite-specific CD8<sup>+</sup> T cells, at least those directed to epitopes derived from CSP and (2) either DCs or hepatocytes are sufficient to induce protective CSP-specific T cell responses if the parasite load is not excessive. Identification of the essential site for priming of malaria liver stage-directed CD8<sup>+</sup> T cell responses of broader antigen specificity as well as mimicking the conditions of the natural exposure to the uninfected mosquito vector will pave the way for the optimal design of T cell-based vaccines. We hope that experimental approaches suggested above in the context of the reviewed original data (<xref rid="B21" ref-type="bibr">21</xref>, <xref rid="B27" ref-type="bibr">27</xref>, <xref rid="B34" ref-type="bibr">34</xref>, <xref rid="B39" ref-type="bibr">39</xref>) will prompt further studies on the induction and maintenance of protective T cell responses against exo-erythrocytic stages of malaria infection.</p></sec><sec id="S7"><title>Conflict of Interest Statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec>
Segregation of a QTL cluster for home-cage activity using a new mapping method based on regression analysis of congenic mouse strains	<p>Recent genetic studies have shown that genetic loci with significant effects in whole-genome quantitative trait loci (QTL) analyses were lost or weakened in congenic strains. Characterisation of the genetic basis of this attenuated QTL effect is important to our understanding of the genetic mechanisms of complex traits. We previously found that a consomic strain, B6-Chr6C<sup>MSM</sup>, which carries chromosome 6 of a wild-derived strain MSM/Ms on the genetic background of C57BL/6J, exhibited lower home-cage activity than C57BL/6J. In the present study, we conducted a composite interval QTL analysis using the F2 mice derived from a cross between C57BL/6J and B6-Chr6C<sup>MSM</sup>. We found one QTL peak that spans 17.6 Mbp of chromosome 6. A subconsomic strain that covers the entire QTL region also showed lower home-cage activity at the same level as the consomic strain. We developed 15 congenic strains, each of which carries a shorter MSM/Ms-derived chromosomal segment from the subconsomic strain. Given that the results of home-cage activity tests on the congenic strains cannot be explained by a simple single-gene model, we applied regression analysis to segregate the multiple genetic loci. The results revealed three loci (loci 1–3) that have the effect of reducing home-cage activity and one locus (locus 4) that increases activity. We also found that the combination of loci 3 and 4 cancels out the effects of the congenic strains, which indicates the existence of a genetic mechanism related to the loss of QTLs.</p>	<contrib contrib-type="author"><name><surname>Kato</surname><given-names>S</given-names></name><xref ref-type="aff" rid="aff1">1</xref><xref ref-type="aff" rid="aff2">2</xref></contrib><contrib contrib-type="author"><name><surname>Ishii</surname><given-names>A</given-names></name><xref ref-type="aff" rid="aff3">3</xref><xref ref-type="aff" rid="aff4">4</xref></contrib><contrib contrib-type="author"><name><surname>Nishi</surname><given-names>A</given-names></name><xref ref-type="aff" rid="aff3">3</xref><xref ref-type="aff" rid="aff4">4</xref></contrib><contrib contrib-type="author"><name><surname>Kuriki</surname><given-names>S</given-names></name><xref ref-type="aff" rid="aff1">1</xref><xref ref-type="aff" rid="aff2">2</xref></contrib><contrib contrib-type="author"><name><surname>Koide</surname><given-names>T</given-names></name><xref ref-type="aff" rid="aff2">2</xref><xref ref-type="aff" rid="aff3">3</xref><xref ref-type="aff" rid="aff4">4</xref><xref ref-type="corresp" rid="caf1">*</xref></contrib><aff id="aff1"><label>1</label><institution>The Institute of Statistical Mathematics</institution>, Tachikawa, Tokyo, <country>Japan</country></aff><aff id="aff2"><label>2</label><institution>Transdisciplinary Research Integration Center, Research Organization of Information and Systems</institution>, Minato-ku, Tokyo, <country>Japan</country></aff><aff id="aff3"><label>3</label><institution>Mouse Genomics Resource Laboratory, National Institute of Genetics</institution>, Mishima, Shizuoka, <country>Japan</country></aff><aff id="aff4"><label>4</label><institution>Department of Genetics, The Graduate University for Advanced Studies (SOKENDAI)</institution>, Mishima, Shizuoka, <country>Japan</country></aff>	Heredity	<sec sec-type="intro"><title>Introduction</title><p>Quantitative trait loci (QTL) analysis of complex traits is one of the most important approaches available for understanding the genetic basis of common phenotypes. The identification of genes for QTLs mapped by genetic studies is making a substantial contribution to better understanding the mechanisms that underlie complex traits. However, many recent attempts to identify genes for QTLs have failed (<xref ref-type="bibr" rid="bib7">Flint <italic>et al.</italic>, 2005</xref>). Numerous studies showed that genetic loci that had significant effects in whole-genome QTL analyses turned out to be lost or weakened once congenic strains were established to show the effect of the QTL (<xref ref-type="bibr" rid="bib27">Saad <italic>et al.</italic>, 2008</xref>; <xref ref-type="bibr" rid="bib26">Rapp and Joe, 2012</xref>; <xref ref-type="bibr" rid="bib29">Stewart <italic>et al.</italic>, 2012</xref>). Even after more than 2000 QTLs had been mapped, only 20 were identified as causative genes; this means that <1% of these applications of the QTL approach were successful (<xref ref-type="bibr" rid="bib7">Flint <italic>et al.</italic>, 2005</xref>). These problems have also been referred to as ‘missing heritability' although possible causes, such as a large number of genes with small effects, epistatic interaction among genes, gene–environment interactions and parent-of-origin effects have been proposed, the actual mechanism(s) responsible for the high level of failure remains unclear (<xref ref-type="bibr" rid="bib18">Manolio <italic>et al.</italic>, 2009</xref>; <xref ref-type="bibr" rid="bib6">Eichler <italic>et al.</italic>, 2010</xref>; <xref ref-type="bibr" rid="bib17">Makowsky <italic>et al.</italic>, 2011</xref>). It is therefore essential to address the genetic basis behind the loss of the QTL effect during the fine mapping of candidate genes.</p><p>Home-cage activity is a general activity that is influenced by the rhythm of an animal's spontaneous activity. We previously showed that the pattern of home-cage activity of a wild-derived mouse strain MSM/Ms (MSM) differs from that of C57BL/6J (B6), a commonly used laboratory strain (<xref ref-type="bibr" rid="bib22">Nishi <italic>et al.</italic>, 2010</xref>). To identify QTLs associated with the total amount of home-cage activity (total activity), we used consomic strains established from crosses between B6 and MSM. Given that each consomic strain has a particular chromosome derived from MSM, whereas the rest are from B6, the detection of a phenotypic difference from B6 enables efficient mapping of QTLs at the chromosomal level. As a result, we found that five consomic strains, for chromosomes 2T (telomere side), 3, 4, 13 and 14, showed significantly higher total activity than B6. In contrast, another five consomic strains, for chromosomes 6C (centromere side), 7T, 9, 11 and 15, were less active than B6. These results indicated that multigenic factors located on different chromosomes regulate the total activity. In order to address the genetic mechanism related to the difference in total activity, we chose chromosome 6, which was associated with one of the lowest scores of total activity, for further genetic analysis in the present study. By applying a two-step approach, namely QTL analysis followed by congenic analysis, we conducted high-resolution genetic analysis of loci related to the total activity. In the studies of a series of congenic strains, we found that the genetic effects behind the different level of total activity were lost in some of the congenic strains, which made it difficult to explain the genetic mechanism with a simple single-gene model. In order to characterise genetic factors related to the total activity in more detail, we used a regression model to reveal the existence of multigenic factors and to map them to different chromosomal regions. In this paper, we report the complex characteristics of the genetic basis for this quantitative trait.</p></sec><sec sec-type="materials\|methods"><title>Materials and methods</title><sec><title>Animals</title><p>Mice of strains MSM/Ms (MSM), C57BL/6JJcl (B6), a consomic strain B6-Chr6C<sup>MSM</sup> (B6-Chr6C), a subconsomic strain B6.MSM-(<italic>D6Nig92.7-D6Mit374</italic>)/Ms (C7) and congenic strains were used in this study. MSM was established as an inbred strain from Japanese wild mice and maintained at the National Institute of Genetics, Mishima, Japan (<xref ref-type="bibr" rid="bib19">Moriwaki <italic>et al.</italic>, 2009</xref>). B6 mice were purchased from CLEA Japan, Inc. (Tokyo, Japan), and bred at the animal facility of the National Institute of Genetics. The consomic strain that carried a large segment of chromosome 6 was established by replacing chromosome 6 of B6 with that of MSM (<xref ref-type="bibr" rid="bib30">Takada <italic>et al.</italic>, 2008</xref>). As a consomic strain carrying the entirety of chromosome 6 was not successfully established, the proximal side of chromosome 6, a 59.6-cM chromosomal segment from the centromere to D6Mit12 derived from MSM, was used for a consomic strain, B6-Chr6C (<xref ref-type="fig" rid="fig1">Figure 1a</xref>). The subconsomic strain and all the congenic strains have the same genetic background as B6, except for a replaced chromosomal region of chromosome 6, derived from B6-Chr6C.</p><p>All the animals were housed with their same-sex littermates until the time of home-cage activity testing in a plastic cage (measuring 19.5 × 29.5 × 15 cm) containing wood chips as the bedding material. The mice were maintained at a constant temperature of 23±2 °C and a 12-h light/dark cycle with lights on at 0600 h, with food and water available <italic>ad libitum</italic>. The home-cage activity of all male mice was tested at 9–12 weeks of age unless otherwise stated. The mice were maintained in accordance with National Institute of Genetics guidelines, and all procedures were carried out with the approval of our institutional animal care and use committee.</p></sec><sec><title>Behavioural testing</title><p>Male mice of B6, consomic, subconsomic and congenic strains were used for home-cage activity tests. Before these tests were conducted, the mice were kept individually for 1 day in their home cage to habituate them to isolation. The home-cage test involved the recording of individual spontaneous home-cage activity from 0600 h on the day after transfer for 3 days. An infrared sensor, Activity Sensor (Ohara Co. Ltd., Tokyo, Japan), was used to evaluate the spontaneous activity of each mouse in their home cage. This sensor was located above the lid (made of stainless steel wire) of each cage. The motion of the mouse inside the home cage was recorded as counts detected by the sensor. Using this apparatus, many kinds of activity, such as horizontal locomotion, climbing the cage lid, hanging on the lid and jumping could be detected efficiently as combined activity counts. However, grooming or foraging behaviour that occurred in one place was detected less efficiently. The test apparatus used can record activity in the home cage for 24 mice simultaneously. In each test session, we analysed up to 24 mice from several different strains at once. Four sets of analyses, namely QTL analysis, subconsomic analysis and two congenic analyses, constitute this study. These four sets of studies were conducted in different periods, with careful calibration of the test apparatus between all analyses. Mice of the control strain B6 were also separated into several groups and subjected to the same analyses as mice of the other strains. The activity in each 1-min bin was measured by accumulated counts if the animal was active in any area of the cage. The activity counts for 72 h were summed and the scores of total activity corresponded to the average counts for a 1-day period over the 3 days.</p></sec><sec><title>QTL analysis and genotyping of microsatellite polymorphisms</title><p>For QTL mapping analysis, F1 mice made by crossing B6 and B6-Chr6C were intercrossed to make F2 progeny. We performed the home-cage activity tests using 174 males of F2 progeny at 8–16 weeks of age. Genomic DNA was isolated from the tails using an automatic nucleic acid isolation system, the NA-2000 (Kurabo, Osaka, Japan). The genotype of each mouse at each of the microsatellite markers was determined. Information on the physical position and sequence information of each microsatellite marker and polymorphism between B6 and MSM for each microsatellite marker was obtained from the National Center for Biotechnology Information (<ext-link xmlns:xlink="http://www.w3.org/1999/xlink" ext-link-type="uri" xlink:href="http://www.ncbi.nlm.nih.gov/">http://www.ncbi.nlm.nih.gov/</ext-link>) and Mouse Microsatellite Database of Japan (<ext-link xmlns:xlink="http://www.w3.org/1999/xlink" ext-link-type="uri" xlink:href="http://www.shigen.nig.ac.jp/mouse/polymorphism/top/top.jsp">http://www.shigen.nig.ac.jp/mouse/polymorphism/top/top.jsp</ext-link>). In addition to the available microsatellite markers, we designed new microsatellite markers as shown in <xref ref-type="supplementary-material" rid="sup1">Supplementary Table S1</xref>. Genomic DNA prepared from tails of F2 mice was amplified by PCR with primer sets of the above microsatellite markers. The amplified DNA fragments were separated on agarose gels by electrophoresis and stained with ethidium bromide.</p><p>To analyse the locations of the QTLs associated with total activity, composite interval mapping based on a multiple QTL model (<xref ref-type="bibr" rid="bib38">Zeng, 1993</xref>, <xref ref-type="bibr" rid="bib39">1994</xref>) was conducted using Windows QTL Cartographer Version 2.5 software (<xref ref-type="bibr" rid="bib3">Basten <italic>et al.</italic>, 1997</xref>). For composite interval mapping, we used model 6 as a standard and a forward regression method with five control markers and 0.5-cM increments with a window size of 10 c<sc>M</sc>. The likelihood-ratio statistic is −2ln(<italic>L</italic><sub>0</sub>/<italic>L</italic><sub>1</sub>), where <italic>L</italic><sub>0</sub> and <italic>L</italic><sub>1</sub> are the maximum likelihoods under the null hypothesis <italic>H</italic><sub>0</sub> (there is no QTL at the test site) and alternative hypothesis <italic>H</italic><sub>1</sub> (there is a QTL at the test site), respectively. Next, the logarithm of odds scores were calculated as 4.605 likelihood ratio=1 logarithm of odds. To determine the 5% level of significance, 3000 replications of a permutation test were carried out.</p></sec><sec><title>Establishment of subconsomic and congenic strains</title><p>To analyse the effect of QTLs, we established a subconsomic strain from B6-Chr6C by introducing shorter chromosomal segments of MSM into B6. From the results of QTL analyses, we selected mice that carried an MSM-derived segment including the QTL region between D6Mit132 and D6Mit55, and made a subconsomic strain, B6.MSM-(<italic>D6Nig92.7-D6Mit374</italic>)/Ms (C7). Further recombinants were made by backcrossing C7 to B6. We made mice that have breakpoints by meiotic recombination at several different points between D6Nig92.7 and D6Mit374. The selected mice were crossed with B6 to obtain pairs of progeny that carried the recombined chromosomal segment. By intercrossing the mice, we obtained homozygotes of the MSM genotype for the loci of interest, and established a series of congenic strains. The genotypes of each microsatellite marker for the subconsomic and the congenic strains are shown in Figure 3.</p></sec><sec><title>Statistical analysis</title><p>Statistical analyses of behavioural test data were performed using the StatView software package (SAS Institute Inc., Cary, NC, USA). Analyses of variance were carried out for between- and within-group factors. One-way between-group analysis of variance (strain) was performed to compare the total activity of the B6 and congenic strains. In the case of significant F values, a <italic>t</italic>-test with Bonferroni correction (<italic>α</italic>=0.05/16 (number of test strains)) was conducted to compare B6 with each congenic strain.</p></sec><sec><title>Regression analysis using data of congenic strains</title><p>In order to identify chromosomal regions that influence the total activity and measure the levels of their influence, we used regression models. In these models, it is assumed that the response <italic>Y</italic><sub><italic>i</italic></sub> (<italic>i</italic>=1,…,17) is the mean of the total activities of mice in congenic strain <italic>i</italic>, where <italic>i</italic> corresponds to one of the 17 different congenic strains established in the study. The regressors or dependent variables <italic>x</italic><sub><italic>i</italic>1</sub>,…, <italic>x</italic><sub><italic>i</italic>30</sub>, <italic>x</italic><sub><italic>i</italic>1</sub> : <italic>x</italic><sub><italic>i</italic>2</sub>, <italic>x</italic><sub><italic>i</italic>2</sub> : <italic>x</italic><sub><italic>i</italic>3</sub>,…, <italic>x</italic><sub><italic>i</italic>29</sub> : <italic>x</italic><sub><italic>i</italic>30</sub> consist of 30 genotypes of the microsatellite markers of the congenic strain <italic>i</italic> and 29 interactions between the adjacent markers. Here, the value of <italic>x</italic><sub><italic>ij</italic></sub> (<italic>j</italic>=1,…,30) is defined as 0 if the genotype of the <italic>j</italic>th marker of the mouse with the congenic strain <italic>i</italic> is the same as that of the mouse B6 and 1 if the genotype of the <italic>j</italic>th marker is the same as that of the mouse B6-Chr6C. The interaction between adjacent markers <italic>x</italic><sub><italic>ij</italic></sub> : <italic>x</italic><sub><italic>ij</italic>+1</sub> is defined as the product of the values of the two adjacent markers, namely, <italic>x</italic><sub><italic>ij</italic></sub> : <italic>x</italic><sub><italic>ij</italic>+1</sub>=<italic>x</italic><sub><italic>ij</italic></sub> <italic>x</italic><sub><italic>ij</italic>+1</sub>.</p><p>One technical problem arising from the regression analysis of these data was the so-called <italic>n</italic><<<italic>p</italic> problem, which arises when the number of microsatellite markers exceeds the number of congenic strains. It is known that for such data there are no unique solutions to the estimating equation of the least squares linear regression model. To solve this problem, our regression analysis was conducted using the following two steps.</p><p>First, we performed model selection via the least absolute shrinkage and selection operator (lasso) (<xref ref-type="bibr" rid="bib34">Tibshirani, 1996</xref>). In our setting, the lasso was applied as follows:</p><p><disp-formula id="equ1"><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="hdy201442e1.jpg"/></disp-formula></p><p>where <italic>i</italic>=1,…,17, <italic>β</italic><sub>0</sub>,…,<italic>β</italic><sub>59</sub> are the regression coefficients, <italic>ɛ</italic><sub><italic>i</italic></sub>'s are independent and identically distributed as Gaussian distributions with mean 0 and variance <italic>σ</italic><sup>2</sup> and <italic>c</italic> is the tuning parameter. Before the lasso was fitted, the dimension of the regressors was reduced by identifying those variables that took the same values as the adjacent variables for all the strains as one variable. Next, the regression coefficients of the lasso were estimated by minimising the penalised least squares:</p><p><disp-formula id="equ2"><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="hdy201442e2.jpg"/></disp-formula></p><p>for a given value of the tuning parameter, where <italic>λ</italic> is uniquely determined by the tuning parameter <italic>c</italic>. The tuning parameter <italic>λ</italic> was estimated using five-fold cross-validation. We generated 100 cross-validation samples and obtained the ultimate estimate of <italic>λ</italic> as the minimum mode of 100 estimates of <italic>λ</italic>.</p><p>Second, the least squares linear regression model was applied to the data whose number of dimensions of the regressors had become less than the number of congenic strains. In order to carry out further variable selection, we adopted all subsets based on the Akaike Information Criterion (<xref ref-type="bibr" rid="bib13">Konishi and Kitagawa, 2008</xref>). The subset that minimised the Akaike Information Criterion value was selected as the model ultimately used for fitting. Using the theory of regression analysis (see, for example, <xref ref-type="bibr" rid="bib16">McCullagh and Nelder, 1989</xref>), estimates of the regression coefficients and the corresponding <italic>P</italic>-values were obtained. The residual analysis was conducted to assess the goodness-of-fit of the model.</p><p>Regression analysis was performed by using the statistical software R. To estimate the regression coefficients and tuning parameter of the lasso, the R package ‘glmnet' was used. The generic function ‘lm' was used for estimation of the least squares linear regression model.</p></sec></sec><sec sec-type="results"><title>Results</title><sec><title>Comparison of home-cage activity between the consomic strain and B6</title><p>In a previous study, we conducted a systematic analysis of home-cage activity using a series of consomic strains and found that five consomic strains including B6-Chr6C, a consomic strain for chromosome 6, showed significantly lower activity than B6 (<xref ref-type="bibr" rid="bib22">Nishi <italic>et al.</italic>, 2010</xref>). In the present study, we conducted home-cage activity tests and confirmed that B6-Chr6C showed lower total activity than B6 (<italic>P</italic><0.0001) (<xref ref-type="fig" rid="fig1">Figure 1b</xref>).</p></sec><sec><title>QTL mapping of a locus for total activity on Chr 6</title><p>To map genetic loci related to total activity, we conducted QTL analysis with males of the F2 population made from an intercross of F1 mice between B6-Chr6C and B6. <xref ref-type="fig" rid="fig2">Figure 2a</xref> shows the results of the analysis by composite interval mapping for the total activity in which one QTL was mapped between D6Mit132 and D6Mit55 (17.6 Mbp). The logarithm of odds score of the QTL was 5.0.</p><p>In order to show the effect of the QTL on the total activity directly, we made a subconsomic strain, C7, which has a chromosome fragment that spans the region between D6Nig92.7 and D6Mit374 (41.3 Mbp), including the QTL region between D6Mit132 and D6Mit55, derived from MSM (<xref ref-type="fig" rid="fig2">Figure 2b</xref>). The observation that C7 exhibited significantly lower total activity than B6 (<italic>P</italic>=0.0104) suggested the existence of a gene for the reduced total activity in the region between D6Mit132 and D6Mit55 (<xref ref-type="fig" rid="fig2">Figure 2c</xref>). Total activities of B6 and C7 were analysed with larger numbers of animals in the next experiment.</p></sec><sec><title>Analysis of total activity using congenic strains</title><p>In order to map QTL related to total activity as precisely as possible, we made a series of congenic strains from C7 (<xref ref-type="fig" rid="fig3">Figure 3</xref>). Total activities of the congenic strains are shown in <xref ref-type="fig" rid="fig3">Figure 3</xref>. One-way analysis of variance showed a significant main effect of strain on the total home-cage activity (F(16,492)=15.97, <italic>P</italic><0.0001). The results showed that four strains, C7-4, C7-15, C7-24 and C7-31, as well as parental strains B6-Chr6C and C7, exhibited significantly lower activities than B6 (<italic>P</italic><0.05). In contrast, one strain, C7-1, showed significantly higher activity than B6, but another strain with a chromosome fragment that covered the entire region of C7-1, such as C7-5, did not. In the results, effects of both decreasing and increasing the total activity, as well as the offsetting of these effects, were observed among these congenic strains. These findings indicate that multiple loci exist in the chromosomal segment of the C7 region on chromosome 6. Given that the relationship of total activity and the regions specifically associated with each of the congenic strains varied among strains, it was difficult to map the QTL precisely using the currently available methods.</p></sec><sec><title>Fine genetic mapping for total activity using regression analysis</title><p>To identify chromosomal regions that influence the total activity more precisely, we used a regression model in which the decreasing and increasing effects in the chromosome regions partitioned into different strains are summed over the entire strain. As a method of pretreatment, the variables in the regressors that took the same values as the adjacent variables for all the strains were grouped together as one variable. For example, the values of D6Nig92.7M, D6Nig93.7M and their interaction are the same for each strain; hence, we grouped these three variables together. This process is necessary because it is known in statistical theory that the estimates of those variables are not unique unless the variables are identified. This enabled us to reduce the dimensions of the regressors from 59 to 18 (<xref ref-type="supplementary-material" rid="sup1">Supplementary Table S2</xref>). We then applied the lasso for model selection and the least squares linear regression model for further variable selection and fitting.</p><p>First we discuss the results of the lasso. <xref rid="tbl1" ref-type="table">Table 1</xref> displays non-zero estimates of the regression coefficients corresponding to the estimated tuning parameter <italic>λ</italic>=43.17. The lasso reduced the number of variables from 18 to 9. The average of the 100 mean cross-validated errors is 2 454 988. The average of 100 estimates of standard error based on the cross-validation samples is 749 904. The total amount of variance is 332 457.</p><p>Second, the least squares linear regression model was fitted to data whose number of dimensions of the regressors was now less than the number of strains. <xref rid="tbl1" ref-type="table">Table 1</xref> shows the estimates of the regression coefficients and the corresponding <italic>P</italic>-values of the selected regression model. The number of selected variables is 6 (out of 9). The maximum log-likelihood is −111.75 and the value of Akaike Information Criterion is 226.3. The total amount of variance is 265 183, which is ∼20% less than that of the estimated lasso. The results of the residual analysis are given in <xref ref-type="fig" rid="fig4">Figure 4a</xref>. These results suggest that the model generally fits well to the data. The maximum absolute value of the residuals is given for strain C7-Q as 1160.32, which is <10% of the total activity of the strain of 13 984.12. <xref ref-type="fig" rid="fig4">Figure 4b</xref> implies that there is no clear relationship between the values of the predictors and residuals. It seems from these observations that the fitted model provides a satisfactory fit to the data.</p><p>From the results of our regression analysis, we mapped the QTL regions related to total activity. <xref rid="tbl1" ref-type="table">Table 1</xref> implies that there are four chromosomal regions where the <italic>P</italic>-values are significant (<italic>P</italic><0.05). The regression coefficients in those regions suggest that strain B6-Chr6C contains loci that decrease the total activity in the following chromosomal regions: adjacent to the marker D6Nig100.2M and/or between D6Nig100.2M and D6Nig102.1M (locus 1), between D6Mit36 and D6Mit105 (locus 2), and between D6Mit104 and D6Mit23 (locus 3). In addition, it is likely to be that there is a locus that increases the total activity in the region adjacent to the marker D6Nig108.6M (locus 4). The levels of influence of these chromosomal regions on total activity can be estimated from the values of the regression coefficients.</p></sec><sec><title>Loci for increasing home-cage activity and the suppressor</title><p>Among these four loci, locus 4, which has a positive effect, shows the opposite effect to the total cumulative effect on activity of B6-Chr6C, which covers a larger substituted region and is associated with lower home-cage activity. In order to confirm the positive effect of locus 4, we established further congenic strains and analysed their total activity (<xref ref-type="fig" rid="fig5">Figure 5</xref>). Two strains, C7-1-163 and C7-1-298, were established from C7-1, and a third strain, C7-5-325, was established from C7-5. One-way analysis of variance showed the significant main effect of strain on the total home-cage activity (F(4,109)=7.67, <italic>P</italic><0.0001). C7-1-163 showed higher total activity than B6, at a level similar to that of C7-1, but C7-1-298 showed no significant increase, which indicated that the region between D6Nig32 and D6Nig29 has an effect of increasing total activity. However, when the region extends towards the proximal side up to 4.1 Mb in C7-5-325, the congenic strain does not exhibit increased total activity. These results suggest that the region between D6Mit36 and D6Nig30 has the effect of suppressing the increasing effect for total activity caused by a factor located in the region between D6Nig32 and D6Nig239. These results are consistent with the results of the regression analysis in that locus 4, which increases the total activity, is mapped around D6Nig108.6M, and locus 2, which decreases it, is mapped in the region between D6Mit36 and D6Mit105 (<xref ref-type="fig" rid="fig3">Figure 3a</xref>).</p></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>Home-cage activity will be influenced by many environmental factors, such as food intake, temperature, bedding materials and social conditions (<xref ref-type="bibr" rid="bib14">Lightfoot <italic>et al.</italic>, 2004</xref>). It is well known that interactions between the genetic apparatus and environment might have a substantial impact on the behavioural and physiological phenotype (<xref ref-type="bibr" rid="bib23">Overall <italic>et al.</italic>, 2013</xref>; <xref ref-type="bibr" rid="bib15">Meek <italic>et al.</italic>, 2014</xref>). We did not see significant difference in body weights between C7-31 and C7-1, which are strains that exhibit the extremes of the opposite deviations in total activity from B6. Thus, we do not have any data that suggest association of differences in physiological metabolism with the difference in home-cage activities between the strains. However, this point needs to be studied further to understand the mechanisms that underlie the differences in the home-cage activities in these congenic strains. In addition, it will be interesting to examine how changes in environmental factors affect the home-cage activities in these strains. Given that our current method for measuring home-cage activity depends on the infrared sensor located on the top of each mouse cage, we were unable to characterise the behaviour exhibited by the mouse in the home-cage in detail. Further studies to better characterise the quality of this behaviour will provide more in-depth understanding of the biological relevance of each genetic locus.</p><p>We previously conducted genetic analyses of behavioural traits using a panel of consomic strains established from MSM and B6 (<xref ref-type="bibr" rid="bib31">Takahashi <italic>et al.</italic>, 2008a</xref>, <xref ref-type="bibr" rid="bib33">2010</xref>; <xref ref-type="bibr" rid="bib22">Nishi <italic>et al.</italic>, 2010</xref>; <xref ref-type="bibr" rid="bib12">Ishii <italic>et al.</italic>, 2011</xref>). The results showed that there are many QTLs related to complex traits distributed over multiple chromosomes. The advantage of using consomic strains is that it makes it possible to focus on a particular chromosome that shows a significant effect on the phenotype under the same genetic background. In this case, further genetic mapping can be conducted under the assumption that a single gene associated with the trait of the consomic strain might exist on each mapped chromosome. Indeed, the result of the QTL analysis for total activity using the F2 population established using B6 and B6-Chr6C showed one significant QTL mapped to the telomere side of chromosome 6, although the QTL peak was relatively broad. In addition, the subconsomic strain C7, which included the QTL region spanning up to 43.3 Mb derived from MSM, showed lower activity, at a level similar to that of B6-Chr6C. However, the results of the regression analysis of the data from a series of congenic strains clearly mapped multiple loci that have effects of increasing activity, decreasing activity and suppression of the increase of activity separately within a small chromosomal region.</p><p>Several other studies using subconsomic or congenic strains showed the existence of multiple loci related to the phenotype associated with the substituted chromosome (<xref ref-type="bibr" rid="bib37">Youngren, 2003</xref>; <xref ref-type="bibr" rid="bib28">Shao <italic>et al.</italic>, 2008</xref>; <xref ref-type="bibr" rid="bib32">Takahashi <italic>et al.</italic>, 2008b</xref>; <xref ref-type="bibr" rid="bib25">Prevorsek <italic>et al.</italic>, 2010</xref>; <xref ref-type="bibr" rid="bib24">Parker <italic>et al.</italic>, 2013</xref>). Complex regulation of the susceptibility to testicular germ cell tumours was reported using congenic strains derived from a consomic strain for chromosome 19, 129.MOLF-Chr19 (<xref ref-type="bibr" rid="bib37">Youngren, 2003</xref>). In this previous study, it was indicated that susceptibility to such tumours is influenced by five genetic regions with additive and epistatic effects. These regions were found along the length of chromosome 19. Several other reports have shown that multiple genetic factors for various quantitative phenotypes are clustered in a small region (<xref ref-type="bibr" rid="bib5">Christians and Keightley, 2004</xref>; <xref ref-type="bibr" rid="bib35">Yalcin <italic>et al.</italic>, 2004</xref>; <xref ref-type="bibr" rid="bib1">Ashikari <italic>et al.</italic>, 2005</xref>; <xref ref-type="bibr" rid="bib2">Ashley-Koch <italic>et al.</italic>, 2006</xref>; <xref ref-type="bibr" rid="bib4">Christians <italic>et al.</italic>, 2006</xref>; <xref ref-type="bibr" rid="bib9">Ghazalpour <italic>et al.</italic>, 2006</xref>; <xref ref-type="bibr" rid="bib36">Yazbek <italic>et al.</italic>, 2011</xref>; <xref ref-type="bibr" rid="bib26">Rapp and Joe, 2012</xref>; <xref ref-type="bibr" rid="bib29">Stewart <italic>et al.</italic>, 2012</xref>). For example, a single QTL that affects body size appeared to be a cluster of at least four closely linked QTLs (<xref ref-type="bibr" rid="bib5">Christians and Keightley, 2004</xref>; <xref ref-type="bibr" rid="bib4">Christians <italic>et al.</italic>, 2006</xref>). These multiple genetic factors might influence body size independently. In another case, two closely linked QTLs were suggested to influence the atherosclerotic phenotype by interacting with each other (<xref ref-type="bibr" rid="bib9">Ghazalpour <italic>et al.</italic>, 2006</xref>). In order to understand the biological functions of these clustered QTLs, it is important to characterise their molecular basis in more detail.</p><p>In the present study, we have mapped four loci: three (loci 1–3) that negatively regulate and one (locus 4) that positively regulates the total home-cage activity. Among these loci, locus 4, which is critical for increasing the total activity, has been mapped to a 1.4-Mb genomic region. It is noteworthy that both the low-activity strain C7-31 and the high-activity strain C7-1 carry locus 4. This result also indicates the complex nature of these multiple genetic loci for regulating home-cage activity. Locus 4 contains three annotated genes: <italic>Bhlhe40</italic>, <italic>Arl8b</italic> and <italic>Edem1</italic>. BHLHE40, also known as DEC1, is a member of the basic helix-loop-helix family and modulates the circadian phase expression of the <italic>Clock</italic> gene (<xref ref-type="bibr" rid="bib21">Nakashima <italic>et al.</italic>, 2008</xref>). ARL8B is an Arf-like GTPase that has a role in directing cargo traffic to lysosomes (<xref ref-type="bibr" rid="bib8">Garg <italic>et al.</italic>, 2011</xref>). EDEM1 is a type II endoplasmic reticulum transmembrane protein that is involved in the endoplasmic reticulum-associated protein degradation pathway to regulate the degradation of misfolded glycoproteins (<xref ref-type="bibr" rid="bib11">Hosokawa <italic>et al.</italic>, 2001</xref>). None of the functions of these genes is strongly suggestive of an association with home-cage activity; however, there is still a possibility of this in all cases. We conducted sequence analyses of these genes and found no non-synonymous polymorphisms between B6 and MSM. It is therefore possible that a different level of expression of one of these genes in the responsible tissue might affect the total activity, although further experiments are needed to clarify this.</p><p>Given that QTL analysis using a panel of congenic strains is one of the most powerful approaches to address multigenic factors behind complex traits, the method using a regression model should be very useful for numerous studies. The regression model used in the present study was the lasso, which is well known as a penalised regression model, that is, a regression model that imposes a certain relationship on unknown parameters. In addition to the lasso, there are several other penalised regression models, such as the ridge regression (<xref ref-type="bibr" rid="bib10">Hoerl and Kennard, 1970</xref>) and elastic net (<xref ref-type="bibr" rid="bib40">Zou and Hastie, 2005</xref>) models. Given that the purpose of the statistical analysis is to identify influential chromosomal regions, it seems that the lasso, which is commonly used for variable selection, is the most appropriate among the familiar penalised regression models.</p><p>The regression model applied in this study supposed that the effects of decreasing and increasing the home-cage activity in the chromosome regions partitioned into different strains are summed over the entire strain. Given that the residuals calculated in the congenic strains were <10% of the total activity of the strain, the current model fits well in this mapping study. In contrast, epistatic interaction between two loci that are not adjacent has a minor effect in the present case. Although we might need to consider the epistatic interaction between two distant loci in another mapping study, this would involve a heavy computational burden for the calculations. Therefore, this method for fine mapping with the regression model will not be applicable to mapping studies that require consideration of the epistatic interaction model.</p><p>In this study, we conducted high-resolution genetic mapping of home-cage activity using a series of congenic strains. This is a highly reliable approach because we can easily make multiple congenic strains from the consomic strains, and the phenotype data of each congenic strain can be collected from multiple mice, as noted previously (<xref ref-type="bibr" rid="bib20">Nadeau <italic>et al.</italic>, 2000</xref>). To identify genes associated with this phenotype, we are currently establishing more recombinants from the congenic strains carrying the shorter chromosomal segments. These further recombinant strains will aid in the identification of genes that increase or decrease home-cage activity in mice.</p></sec><sec><title>Data archiving</title><p>Data are deposited in the Dryad repository.</p></sec>
Genetics of decayed sexual traits in a parasitoid wasp with endosymbiont-induced asexuality	<p>Trait decay may occur when selective pressures shift, owing to changes in environment or life style, rendering formerly adaptive traits non-functional or even maladaptive. It remains largely unknown if such decay would stem from multiple mutations with small effects or rather involve few loci with major phenotypic effects. Here, we investigate the decay of female sexual traits, and the genetic causes thereof, in a transition from haplodiploid sexual reproduction to endosymbiont-induced asexual reproduction in the parasitoid wasp <italic>Asobara japonica</italic>. We take advantage of the fact that asexual females cured of their endosymbionts produce sons instead of daughters, and that these sons can be crossed with sexual females. By combining behavioral experiments with crosses designed to introgress alleles from the asexual into the sexual genome, we found that sexual attractiveness, mating, egg fertilization and plastic adjustment of offspring sex ratio (in response to variation in local mate competition) are decayed in asexual <italic>A. japonica</italic> females. Furthermore, introgression experiments revealed that the propensity for cured asexual females to produce only sons (because of decayed sexual attractiveness, mating behavior and/or egg fertilization) is likely caused by recessive genetic effects at a single locus. Recessive effects were also found to cause decay of plastic sex-ratio adjustment under variable levels of local mate competition. Our results suggest that few recessive mutations drive decay of female sexual traits, at least in asexual species deriving from haplodiploid sexual ancestors.</p>	<contrib contrib-type="author"><name><surname>Ma</surname><given-names>W-J</given-names></name><xref ref-type="aff" rid="aff1">1</xref><xref ref-type="corresp" rid="caf1">*</xref></contrib><contrib contrib-type="author"><name><surname>Pannebakker</surname><given-names>B A</given-names></name><xref ref-type="aff" rid="aff2">2</xref></contrib><contrib contrib-type="author"><name><surname>Beukeboom</surname><given-names>L W</given-names></name><xref ref-type="aff" rid="aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Schwander</surname><given-names>T</given-names></name><xref ref-type="aff" rid="aff1">1</xref><xref ref-type="aff" rid="aff3">3</xref><xref ref-type="author-notes" rid="note1"><sup>4</sup></xref></contrib><contrib contrib-type="author"><name><surname>van de Zande</surname><given-names>L</given-names></name><xref ref-type="aff" rid="aff1">1</xref><xref ref-type="author-notes" rid="note1"><sup>4</sup></xref></contrib><aff id="aff1"><label>1</label><institution>Evolutionary Genetics, Center for Ecological and Evolutionary Studies, University of Groningen</institution>, Groningen, <country>The Netherlands</country></aff><aff id="aff2"><label>2</label><institution>Laboratory of Genetics, Wageningen University, Droevendaalsesteeg 1</institution>, Wageningen, <country>The Netherlands</country></aff><aff id="aff3"><label>3</label><institution>Department of Ecology and Evolution, University of Lausanne</institution>, Lausanne, <country>Switzerland</country></aff>	Heredity	<sec sec-type="intro"><title>Introduction</title><p>Owing to environmental or life style changes and associated shifts in selective pressures, formerly adaptive traits may become non-functional or even maladaptive, and as a consequence they might decay (<xref ref-type="bibr" rid="bib8">Fong <italic>et al.</italic>, 1995</xref>; <xref ref-type="bibr" rid="bib58">Wiens, 2001</xref>; <xref ref-type="bibr" rid="bib7">Ellers <italic>et al.</italic>, 2012</xref>). Trait decay has been observed for morphological, behavioral and physiological features. Examples include reduced wings in flightless birds (<xref ref-type="bibr" rid="bib28">McNab, 1994</xref>), loss of eye function, decay of pigmentation in cave-dwelling animals (<xref ref-type="bibr" rid="bib17">Jeffery, 2009</xref>; <xref ref-type="bibr" rid="bib38">Protas <italic>et al.</italic>, 2011</xref>) and loss of lipid synthesis pathways in parasitoid wasps (<xref ref-type="bibr" rid="bib52">Visser <italic>et al.</italic>, 2010</xref>). Such trait decay, especially when occurring in parallel in independent lineages, highlights the importance of natural selection for the maintenance of adaptations (<xref ref-type="bibr" rid="bib8">Fong <italic>et al.</italic>, 1995</xref>; <xref ref-type="bibr" rid="bib58">Wiens, 2001</xref>; <xref ref-type="bibr" rid="bib23">Lahti <italic>et al.</italic>, 2009</xref>). Despite the fact that trait decay is a common evolutionary phenomenon, its underlying genetic mechanisms are poorly understood. It is largely unknown if trait decay typically stems from multiple mutations with small effects, or rather has a simple genetic architecture involving few loci with major phenotypic effects (<xref ref-type="bibr" rid="bib17">Jeffery, 2009</xref>; <xref ref-type="bibr" rid="bib23">Lahti <italic>et al.</italic>, 2009</xref>).</p><p>In particular, many traits are expected to decay following a transition from sexual to asexual reproduction, making this transition of special interest for studies of trait decay and the genetic causes thereof (for example, <xref ref-type="bibr" rid="bib2">Carson <italic>et al.</italic>, 1982</xref>; <xref ref-type="bibr" rid="bib33">Pannebakker <italic>et al.</italic>, 2004a</xref>; <xref ref-type="bibr" rid="bib18">Jeong and Stouthamer, 2005</xref>; <xref ref-type="bibr" rid="bib21">Kraaijeveld <italic>et al.</italic>, 2009</xref>; <xref ref-type="bibr" rid="bib42">Russell and Stouthamer, 2011</xref>; <xref ref-type="bibr" rid="bib43">Schwander <italic>et al.</italic>, 2013</xref>). For example, any trait specific to the male sex is useless under asexual reproduction in all-female species. The same holds for sexual traits expressed in females, such as those involved in attracting mates, mating behavior and egg fertilization.</p><p>Here, we investigate the decay of female sexual traits and its genetic basis in the asexual wasp <italic>Asobara japonica. A. japonica</italic> is a parasitoid wasp that uses <italic>Drosophila</italic> larvae as host. It consists of both sexual and all-female asexual strains (<xref ref-type="bibr" rid="bib30">Murata <italic>et al.</italic>, 2009</xref>). Genetic analyses can be difficult in asexual organisms because of the inability to perform crosses. An exception to this constraint applies to those species in which asexuality is induced by infection with bacterial endosymbionts, such as <italic>A. japonica</italic> (<xref ref-type="bibr" rid="bib22">Kremer <italic>et al.</italic>, 2009</xref>; <xref ref-type="bibr" rid="bib41">Reumer <italic>et al.</italic>, 2012</xref>). Endosymbiont-induced asexuality is mainly found among wasps and among other groups in which sexual species are characterized by haplodiploid sex determination (<xref ref-type="bibr" rid="bib54">Werren, 1997</xref>; <xref ref-type="bibr" rid="bib55">Werren <italic>et al.</italic>, 2008</xref>; <xref ref-type="bibr" rid="bib29">Mateo Leach <italic>et al.</italic>, 2009</xref>; <xref ref-type="bibr" rid="bib9">Giorgini <italic>et al.</italic>, 2010</xref>; <xref ref-type="bibr" rid="bib19">Kageyama <italic>et al.</italic>, 2012</xref>; <xref ref-type="bibr" rid="bib26">Ma <italic>et al.</italic>, 2014</xref>), although it has also been suggested to occur in species with other sex determination systems (for example, <xref ref-type="bibr" rid="bib37">Pike and Kingcombe, 2009</xref>). Under haplodiploidy, females develop from fertilized, diploid eggs, whereas males develop from unfertilized, haploid eggs (<xref ref-type="bibr" rid="bib57">Whiting, 1933</xref>). However, unfertilized haploid eggs laid by endosymbiont-infected females undergo diploidization in the absence of fertilization with sperm (<xref ref-type="bibr" rid="bib49">Suomalainen <italic>et al.</italic>, 1987</xref>; <xref ref-type="bibr" rid="bib46">Stouthamer <italic>et al.</italic>, 1990</xref>; <xref ref-type="bibr" rid="bib54">Werren, 1997</xref>; <xref ref-type="bibr" rid="bib56">Werren and O'Neil, 1997</xref>; <xref ref-type="bibr" rid="bib11">Gottlieb and Zchori-Fein 2001</xref>; <xref ref-type="bibr" rid="bib34">Pannebakker <italic>et al.</italic>, 2004b</xref>). In species with endosymbiont-induced asexuality, the genetics of traits involved in sexual reproduction can be studied because infected asexual females can often be cured of their endosymbionts via treatment with antibiotics. Such cured asexual females produce males, and these males can be crossed with females from related sexual strains (<xref ref-type="bibr" rid="bib18">Jeong and Stouthamer, 2005</xref>; <xref ref-type="bibr" rid="bib35">Pannebakker <italic>et al.</italic>, 2005</xref>; <xref ref-type="bibr" rid="bib42">Russell and Stouthamer, 2011</xref>). In addition, asexual females occasionally produce males under natural conditions and in the absence of antibiotic treatments. This is possibly caused by incomplete endosymbiont transmission or by the occasional inability of endosymbionts to manipulate host reproduction (<xref ref-type="bibr" rid="bib14">Heath <italic>et al.</italic>, 1999</xref>; <xref ref-type="bibr" rid="bib41">Reumer <italic>et al.</italic>, 2012</xref>). Although these males usually do not have any mating opportunities or success in the asexual populations, they can be mated with females of sexual strains, similar to the males induced by antibiotic treatment.</p><p>We investigated four female sexual traits for signs of decay in asexual <italic>A. japonica</italic> and studied the genetic architecture of decayed traits via introgression of alleles from the asexual strain into the sexual one. As the expressed level of decay should depend on the degree of introgression of the asexual genome, we can make inferences about the genetic architecture underlying sexual trait decay, using simple genetic models to estimate the number of loci involved. Specifically, we investigated attractiveness to males, mating behavior and egg fertilization in two different contexts specific to sexual reproduction in haplodiploid parasitoids. We tested which fraction of females with different levels of sexual–asexual admixture would still fertilize eggs, and to what extent these females would adjust the proportion of fertilized eggs (that is, the proportion of females among their offspring) to different conditions. Females of many parasitoid species adjust the sex ratio of their offspring adaptively when mating occurs among the offspring of a few females in isolated patches (local mate competition (LMC); <xref ref-type="bibr" rid="bib13">Hamilton, 1967</xref>). If only a single female is present in a patch, her sons will compete with each other to mate with their sisters. Hence, it is in the female's interest to produce the minimum number of sons required to fertilize all her daughters and allocate more energy into the production of daughters (<xref ref-type="bibr" rid="bib13">Hamilton, 1967</xref>). When several females are present in a patch, sons from different females will compete, favoring a larger investment in sons. Sexual females are therefore predicted to produce a smaller proportion of sons when ovipositing alone than when other females are present in the same patch (<xref ref-type="bibr" rid="bib13">Hamilton, 1967</xref>; <xref ref-type="bibr" rid="bib53">Werren, 1980</xref>), a pattern we also found in a sexual <italic>A. japonica</italic> strain in a pilot experiment. This plasticity in sex-ratio adjustment makes it an interesting sexual trait to consider in the context of decaying sexual functions under asexual reproduction in parasitoid wasps, as there is no selection for the maintenance of plasticity under asexuality. Our results point to a surprisingly simple genetic architecture underlying the decay of female sexual traits, which most likely facilitated the spread of reduced sexual traits in the asexual population.</p></sec><sec sec-type="materials\|methods"><title>Materials and methods</title><sec><title>Wasp culturing and collection of individuals</title><p>A sexual and a <italic>Wolbachia</italic>-induced, asexual strain of <italic>A. japonica</italic> were used in this study. Both strains were collected from Japan and have been cultured in the laboratory since 2009; the sexual strain originated from the island of Amami-oshima and the asexual strain from Kagoshima on the mainland of Japan. These two strains are closely related (<xref ref-type="bibr" rid="bib30">Murata <italic>et al.</italic>, 2009</xref>; <xref ref-type="bibr" rid="bib41">Reumer <italic>et al.</italic>, 2012</xref>), which minimizes the probability of genetic incompatibility in crosses. <italic>A. japonica</italic> was cultured on second-instar <italic>Drosophila melanogaster</italic> larvae as hosts at 25 °C, with a 16L:8D light–dark cycle and 60% relative humidity (for details see <xref ref-type="bibr" rid="bib25">Ma <italic>et al.</italic>, 2013</xref>).</p><p>We used five different classes of virgin individuals: sexual females, sexual males, asexual females cured of their <italic>Wolbachia</italic> infection via antibiotic treatment, untreated asexual females and males produced by asexual females. The males produced by asexual females were either induced via antibiotic treatment or directly collected from our mass culture with the highest incidence of accidental males (we collected 40–70 males among ∼6000 individuals). After using these males in our experiments, we verified, via flow cytometry, that they were haploid (as expected under normal haplodiploid sex determination in Hymenoptera; for details, see <xref ref-type="bibr" rid="bib25">Ma <italic>et al.</italic>, 2013</xref>). Virgin sexual males and females were also collected directly from standard laboratory cultures. Sexual males were collected first because they emerge one or two days earlier than females. After the emergence of males, sexual females were collected by individually isolating wasp pupae in plastic vials (diameter 2.4 cm, height 7.5 cm) containing a layer of agar to control humidity (<xref ref-type="bibr" rid="bib25">Ma <italic>et al.</italic>, 2013</xref>).</p><p><italic>Wolbachia</italic>-infected asexual females were cured of their bacteria with antibiotics applied to the <italic>Drosophila</italic> host larvae. In total, 10 mg of rifampicin was added to 1 g yeast powder, which was mixed with water to feed the second-instar <italic>Drosophila</italic> larvae. Rifampicin treatment has been shown to have little impact on the development of <italic>Asobara</italic> wasps (<xref ref-type="bibr" rid="bib5">Dedeine <italic>et al.</italic>, 2001</xref>), and no effects on life-history traits, such as brood size and pupal mortality, were found in our experiments (data not shown). Female wasps that emerged from rifampicin-fed <italic>Drosophila</italic> hosts were individually collected in plastic vials containing a layer of agar. To confirm complete removal of <italic>Wolbachia</italic> by antibiotic treatment, 23 emerged females were allowed to lay eggs in antibiotic-free host larvae, and the production of only male offspring was verified.</p></sec><sec><title>Observation of courtship behavior and spermatheca dissections</title><p>To investigate sexual attractiveness and mating behavior of females, we set up 109 no-choice mating trials. For each trial, asexual females (<italic>n</italic>=20 cured, 40 untreated) or sexual females (<italic>n</italic>=49) were individually paired with a sexual male for 20 min. We evaluated female attractiveness to sexual males by scoring if the male attempted to court the females (that is, displayed wing vibration and actively approached the female). For female mating behavior, we scored if females responded to the copulation attempts of sexual males with escape behavior or acceptance. A successful copulation was scored when it lasted at least 7 s. As these experiments revealed that cured asexual females were relatively unattractive to sexual males (see results), we evaluated if decreased attractiveness could stem from lineage divergence that is unrelated to the asexual mode of reproduction (that is, as could be observed between two sexual species). To this end, we repeated similar tests with males produced by asexual females, as there is no lineage divergence between these males and asexual females.</p><p>To investigate if cured asexual females fertilize their eggs and produce daughters when mated with sexual males, two experiments were performed. First, we paired each of 20 cured asexual females with a sexual male for 24 h. Second, to increase the chance of obtaining any mated asexual females, we paired each of 13 additional cured asexual females with a group of at least 50 sexual males in one mating vial for 24 h. Females were then offered 50–100 second-instar host larvae for egg laying for ∼24 h (both experiments), to test if they produce any daughters, indicative of successful egg fertilization. Next, the spermathecae of all asexual females were dissected to check for the presence of sperm, as an indication of successful mating and sperm transfer.</p><p>For spermatheca dissections, an individual female wasp was placed in a drop of <italic>Drosophila</italic> Ringer's solution (<xref ref-type="bibr" rid="bib40">Rajaram <italic>et al.</italic>, 2005</xref>) on a microscope slide. The wasp abdomen was first separated from the rest of the body using a very fine needle. The spermatheca was then carefully separated from the rest of the abdomen. The spermatheca was isolated under a Zeiss Stemi SV 6 stereo microscope (with 25 × 2.4 magnification, Carl Zeiss AG, Oberkochen, Germany), and a cover slide was gently put on top of it. Motile sperm were scored as present or absent under a Zeiss Axio Lab.A1 binocular microscope (with 10 × 40 magnification, Carl Zeiss GmbH, Göttingen, Germany).</p></sec><sec><title>Introgression experiment</title><p>To determine the genetic architecture of decayed female sexual traits, we introgressed alleles from the asexual into the sexual strain. Upon emergence of virgin sexual females, they were individually paired with a male from the asexual strain for 24 h. Each female was then offered ∼100 second-instar <italic>D. melanogaster</italic> larvae for oviposition during 36 h. After 12–14 days, wasp pupae were isolated from parasitized hosts to prevent matings between individuals upon emergence. Females emerging from these pupae (the F1 hybrid generation) were collected and individually paired with a male from the asexual strain to produce the offspring for the next generation (for further details, see <xref ref-type="bibr" rid="bib25">Ma <italic>et al.</italic>, 2013</xref>). This experimental procedure was continued for successive generations of introgressions. Because almost no daughters were produced after the 4th generation (‘G4') of introgression (because females with high proportions of their genome stemming from the asexual strain do not produce daughters, see results), the introgression experiment was stopped after the 4th generation of introgression. We repeated the same introgression experiment three times independently over the course of 3 months, because only a few asexually produced males were available on specific days for crosses. Because there was no difference for brood size or offspring sex ratio or any other parameter between different experiments (analyses not shown), all data were pooled for further analysis. Given our crossing design, the proportion of asexual alleles in each generation increased from 50% in F1 hybrid females to 75% in females of G2, 87.5% in G3 and a final 93.8% in females of G4. For each generation of introgression, the emerging wasps were anaesthetized with CO<sub>2</sub>, counted and sexed (<xref ref-type="bibr" rid="bib25">Ma <italic>et al.</italic>, 2013</xref>).</p><p>The combination of decayed sexual attractiveness, mating behavior and egg-fertilization results in cured asexual females producing only sons. Therefore, we monitored the proportion of admixed females producing only sons across successive generations of introgression. For each generation of introgression, a subset of hybrid females (those not used to continue the introgression crosses) were paired with a sexual male for 24 h followed by oviposition for 36 h. The sex ratio of offspring produced during these 36 h was then determined for each female to infer the proportion of females producing only sons. Using the same experimental conditions, we also determined the sex ratio produced by sexual females and cured asexual females.</p></sec><sec><title>Local mate competition experiment</title><p>To investigate if <italic>A. japonica</italic> females produce different offspring sex ratios (proportion of male offspring) in response to different situations of LMC, we investigated offspring sex ratios produced by females that oviposited alone or in groups of three females per patch. LMC theory predicts that the proportion of sons increases with increasing numbers of females in a patch. A pilot experiment (data not shown) showed that sexual females do indeed change the sex ratio among their offspring depending on the number of females (one or three) per patch. Thus we tested if sexual females, cured asexual females and females from different introgression generations would produce different offspring sex ratios when alone in a vial relative to when in a group of three. To avoid any interactions between females in different vials (<xref ref-type="bibr" rid="bib45">Shuker <italic>et al.</italic>, 2007</xref>), the vials were spaced 5 cm apart. As not all hybrid females mate, which would constrain their sex allocation, females were observed while paired with a sexual male and only those females that copulated were used to test for sex-ratio adjustments. Prior to this experiment, each female was offered ∼50 host larvae to get oviposition experience in order to minimize super-parasitism (oviposition in already-parasitized hosts; <xref ref-type="bibr" rid="bib50">van Alphen and Nell, 1982</xref>; <xref ref-type="bibr" rid="bib25">Ma <italic>et al.</italic>, 2013</xref>). To control for effects of host larva density on wasp offspring sex ratio, we adjusted the number of the host larvae to the number of females, that is, 50 larvae were offered to a single and 150 to a group of three females for ∼15 h. Fifty larvae per female are an excess as a single female can handle a maximum of 30–40 larvae in 15 h (W-J Ma, personal observation). Offspring were counted and sexed upon their emergence from the host pupae.</p></sec><sec><title>Statistical analyses</title><p>Fisher's exact tests were used to compare the proportion of courting attempts by males, rejection rates of males and the frequency of successful copulations of sexual and cured asexual females. Offspring sex ratios of females from different generations of introgression were compared with logistic regressions specified in generalized linear models (glm). The proportions of females producing only sons between different introgression generations were tested using glm models with a logit link function, and a quasi-binomial error structure to correct for over-dispersion (<xref ref-type="bibr" rid="bib3">Crawley, 2007</xref>). The proportion of asexual genome in hybrid females was used as a quantitative explanatory variable and the progeny type (either only sons or at least one daughter) as the response. To pinpoint significant differences between introgression generations, a sequential multiple comparison was used. To conduct the multiple comparison, generation had to be used as a categorical variable. For comparison of offspring sex ratios of females producing at least one daughter among different generations of introgression, a glm model was used with a logit link function and a quasi-binomial error structure to correct for over-dispersion (<xref ref-type="bibr" rid="bib3">Crawley, 2007</xref>). In this model, the proportion of males was used as the response variable (weighted by brood size via the logit link function), and generation as the explanatory variable. A similar glm model was used for comparing the proportion of male offspring in the LMC experiments, that is, between single and triple ovipositing females. All statistical analyses were performed with R 2.13.0 (<xref ref-type="bibr" rid="bib39">R Development Core Team, 2011</xref>), multiple comparisons of traits among generations were done using the Tukey test as implemented in the R package multcomp for general linear hypotheses (<xref ref-type="bibr" rid="bib15">Hothorn <italic>et al.</italic>, 2008</xref>).</p></sec></sec><sec sec-type="results"><title>Results</title><sec><title>Sexual traits in asexual females</title><p>First, we evaluated whether sexual traits were functional in <italic>Wolbachia</italic>-cured asexual <italic>A. japonica</italic> females. Specifically, we tested whether cured asexual females were able to attract sexual males, mate, and store sperm in their spermathecae, and, if so, whether they fertilize eggs and display plastic sex-allocation behavior under different levels of LMC.</p><p>No-choice mating trials (20 min) revealed that cured asexual females were less attractive to sexual males than were sexual females. Sexual males courted a significantly smaller percentage of asexual (55% 11 out of 20) than sexual females (88% 43 out of 49; Fisher's exact test, <italic>P</italic>=0.008, <xref rid="tbl1" ref-type="table">Table 1</xref>). In addition, among the females that were courted by sexual males, cured asexual females rejected all mounting and copulation attempts of males (100%, <italic>n</italic>=11), whereas rejections were significantly less frequent for sexual females (23% 10 out of 43, Fisher's exact test, <italic>P</italic><0.0001, <xref rid="tbl1" ref-type="table">Table 1</xref>). As a consequence, not a single 1 out of 20 cured asexual females successfully copulated with a sexual male during the 20 min observation period, as compared with over 67% (33 out of 49) sexual females (Fisher's exact test, <italic>P</italic>=0.0003, <xref rid="tbl1" ref-type="table">Table 1</xref>). The reduced attractiveness and copulation propensity of asexual females as compared with sexual females are unlikely to be caused by the treatment with antibiotics, as untreated asexual females were also unattractive to sexual males; only 3 out of 40 untreated asexual females (8%) were courted, compared with 43 out of 49 sexual females (88% Fisher's exact test, <italic>P</italic>=0.0001, <xref rid="tbl1" ref-type="table">Table 1</xref>). The comparison of cured and untreated asexual females further revealed that sexual males are more inclined to court the former (Fisher's exact test, <italic>P</italic>=0.0001). The reduced attractiveness of asexual females to sexual males is unlikely to stem from divergence of functional sexual signals between sexual and asexual strains (as could be expected for comparisons between two diverged sexual strains). This is revealed by the fact that males produced by asexual females also courted sexual females more often (79% 50 out of 63) than cured asexual females (52% 11 out of 21; Fisher's exact test, <italic>P</italic>=0.024).</p><p>To investigate if cured asexual females were able to fertilize their eggs and produce daughters, we conducted two experiments to increase the chances of finding mated asexual females, given the above-described 20 min no-choice trials did not result in any copulations. In the first experiment, 20 cured asexual females were each paired with a sexual male for 24 h, but all 11 females that produced offspring had only sons (40–60 offspring per female, over 400 offspring in total). The remaining nine females produced no offspring at all (for unknown reasons, subsequent dissections revealed a normal egg load and apparently developed ovaries in all females). In the second experiment, we paired each of 13 cured asexual females with groups of at least 50 sexual males for 24 h. Again, nine females produced only sons (40–60 offspring per female, over 300 offspring in total), even though dissecting them after the experiment revealed that three of these contained sperm in their spermathecae. The remaining four females, of which one contained sperm in her spermatheca, did not produce any offspring (despite having active ovaries). Thus, even in the rare cases where asexual females do mate and store sperm, sperm is apparently not used to fertilize eggs.</p></sec><sec><title>Patterns of sexual trait change across different levels of introgression</title><p>The combination of low attractiveness to males, low copulation propensity and the absence of egg fertilization, leads to cured asexual females producing only sons under conditions where the majority of sexual females (89.3% <xref ref-type="fig" rid="fig1">Figure 1</xref>) produce daughters in addition to sons. Hence, we monitored the proportion of females producing only sons, as a measure of sexual trait decay, across increasing levels of introgression of the asexual genome into the sexual one. As expected, given the phenotype of the sexual and asexual strains for this trait, the proportion of females producing only sons increased with increasing levels of introgression (<xref ref-type="fig" rid="fig1">Figure 1</xref>, glm, F<sub>1,405</sub>=72.3, <italic>P</italic><0.0001). Importantly, the first significant increase was in the second generation (glm, <italic>t</italic>=−4.5, <italic>P</italic><0.0001); no significant increase was observed in the first generation (glm, <italic>t</italic>=1.0, <italic>P</italic>=0.340; <xref ref-type="fig" rid="fig1">Figure 1</xref>). This indicates that recessive genetic effects cause females to produce only sons. More precisely, the quantitative increase of the proportion of females producing only sons in the second generation is consistent with a simple genetic architecture of this trait (<xref rid="tbl2" ref-type="table">Table 2</xref>). Indeed, for hybrid females with 75% of their genome stemming from asexual strains, 46% (104 out of 225) produced only sons, which is in agreement with the expected 50% under a single-locus model, but differs significantly from the expected frequencies under models with two or more loci (<xref rid="tbl2" ref-type="table">Table 2</xref>).</p><p>Given that in addition to the females producing only sons, there was a (decreasing) fraction of females producing daughters in addition to sons, we also monitored how the offspring sex ratios of these females changed across increasing levels of introgression. However, in contrast to the proportion of females producing only sons, we had no specific prediction for how this trait would change with an increased representation of the asexual genome, given that this trait would derive from the sexual rather than the asexual genome. Indeed, as revealed by the experiments described above, cured asexual females do not fertilize any of their eggs, even if they have copulated. In our introgression experiment, we needed hybrid females for initiating each successive generation, hence we indirectly selected for females that copulated and fertilized at least some of their eggs. Following this indirect selection, we found that the sex ratio among offspring of females with at least some daughters fluctuated across increasing levels of introgression (<xref ref-type="fig" rid="fig2">Figure 2</xref>). Although females from all different introgression levels produced significantly more female-biased sex ratios than ‘pure' sexual females, the most female-biased offspring sex ratios were found for the F1 hybrid females (<xref ref-type="fig" rid="fig2">Figure 2</xref>, glm, Tukey contrasts, all <italic>P</italic><0.001). Even though these females also produced significantly fewer offspring (76.6±2.6) than ‘pure' sexual females (88±3.1; Welch <italic>t</italic>-test, <italic>t</italic><sub>91.6</sub>=2.9, <italic>P</italic>=0.004), the low sex ratios are unlikely to stem from high mortality of males as a consequence of hybrid breakdown. Hybrid breakdown should cause increased mortality of sons produced by F1 hybrid females independently of rearing conditions, yet we found no indication of hybrid breakdown among the females producing only sons (described above), and there was also no evidence for it in the LMC experiment (see the next paragraph). Hence, the reason for the initial drop in sex ratio and offspring number of F1 hybrid females remains unknown.</p><p>As asexual females produce only daughters, there is no selection for plastic sex allocation in response to different levels of LMC in asexual strains. We first verified that sexual <italic>A. japonica</italic> females indeed displayed plastic sex allocation. This was the case, as revealed by sexual females producing a significantly larger proportion of sons when in groups of three than when alone (<xref ref-type="fig" rid="fig3">Figure 3</xref>, glm, F<sub>1,57</sub>=13.9, <italic>P</italic><0.001). A similar response was also displayed by sexual–asexual F1 hybrid females (<xref ref-type="fig" rid="fig3">Figure 3</xref>, glm, F<sub>1,22</sub>=5.5, <italic>P</italic>=0.029). However, in the second generation (females with 75% of their genome stemming from asexual strains), females did not adjust their offspring sex ratio to different levels of LMC (all of these females produced at least one daughter; <xref ref-type="fig" rid="fig3">Figure 3</xref>, glm, F<sub>1,10</sub>=0.2, <italic>P</italic>=0.66). We did not perform the LMC experiment for G3 and G4 introgression generations because plastic sex allocation was already completely lost in generation G2. In combination, these patterns suggest that there is decay of sex-allocation plasticity in asexual females, and that this decay is due to recessive genetic effects, similar to the decay of attractiveness and copulation propensity. However, an estimation of the number of loci affecting this trait cannot be provided, given that the presence vs absence of sex-allocation plasticity is evaluated qualitatively at the group level.</p></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>In this study, we used the parasitoid wasp <italic>A. japonica</italic> with <italic>Wolbachia</italic>-induced asexuality to investigate if female sexual traits decay under asexuality, and if so, whether trait decay is caused by many loci with small effects or by only a few major-effect loci. A combination of behavioral experiments with crosses designed to introgress alleles from the asexual into sexual genome revealed decay of all investigated sexual traits and suggested a surprisingly simple genetic architecture underlying trait decay.</p><p>We found evidence for moderate decay of sexual attractiveness and extensive decay of mating and egg-fertilization behavior in asexual <italic>A. japonica</italic>. Asexual females were always less likely to be courted by males than sexual females, but the percentage of asexual females that were courted depended on whether these females were infected with <italic>Wolbachia</italic> (8%) or cured of their infection by antibiotic treatment (55%). This suggests that <italic>Wolbachia</italic> negatively affects the sexual attractiveness of infected females via an unknown mechanism. <italic>Wolbachia</italic> may, for example, modify sexual signals expressed in asexual females. Bacteria are known to alter pheromone production in some insects, such as commensal bacteria in <italic>D. melanogaster</italic> (<xref ref-type="bibr" rid="bib44">Sharon <italic>et al.</italic>, 2010</xref>), and gut bacteria in the desert locust <italic>Schistocerca geraria</italic> (<xref ref-type="bibr" rid="bib6">Dillon and Charnley, 2002</xref>). In <italic>D. melanogaster,</italic> curing individuals from endosymbiont infection decreases levels of mate discrimination between populations by about 50%, an effect likely mediated by the effect of endosymbionts on female pheromone production (<xref ref-type="bibr" rid="bib20">Koukou <italic>et al.</italic>, 2006</xref>; see also <xref ref-type="bibr" rid="bib44">Sharon <italic>et al.</italic>, 2010</xref>). <italic>Wolbachia</italic> could also affect female attractiveness by altering cuticular hydrocarbon profiles, which function as mating cues in many insect species (for example, <xref ref-type="bibr" rid="bib16">Ivy <italic>et al.</italic>, 2005</xref>; <xref ref-type="bibr" rid="bib59">Yew <italic>et al.</italic>, 2009</xref>).</p><p>Mate attraction is often associated with significant costs in sexual species (<xref ref-type="bibr" rid="bib4">Daly, 1978</xref>), because of resource investment in the production of mate attraction signals and/or because it can increase the risk of predation (for example, <xref ref-type="bibr" rid="bib60">Zuk <italic>et al.</italic>, 2006</xref>). As a consequence, mate attraction is expected to be under strong negative selection in asexual species where it is superfluous, a prediction largely supported by empirical data (for example, <xref ref-type="bibr" rid="bib24">Lehmann <italic>et al.</italic>, 2011</xref>; reviewed by <xref ref-type="bibr" rid="bib51">van der Kooi and Schwander, 2014</xref>). However, once mate attraction is low, for example as a consequence of endosymbiont infection as we found for asexual <italic>A. japonica</italic>, the strength of selection for decreased expression of signals involved in mating interactions might be reduced. This may explain why asexual <italic>A. japonica</italic> females cured of their endosymbionts were still somewhat attractive to sexual males, although less than sexual females.</p><p>In contrast to the moderate decay of sexual attractiveness, mating behavior is largely disrupted in asexual females. This disruption is not caused by <italic>Wolbachia</italic> infection. Both infected and cured asexual females rejected all mating attempts by males under conditions where the majority of sexual females (88%) would accept mates and copulate. The mechanisms underlying disrupted mating interactions in asexual females remain to be investigated.</p><p>We also aimed at investigating if asexual females fertilize their eggs upon mating with a sexual male. In order to obtain any mated asexual females, we had to pair cured asexual females individually with groups of over 50 sexual males. The four asexual females that contained sperm in their spermathecae either produced no offspring at all (one female) or did not fertilize their eggs and produced only sons (three females). Although the sample is small, this pattern suggests that the ability to produce fertilized embryos decayed in asexual <italic>A. japonica</italic> females. Decay of egg fertilization can have different causes, for example egg modifications such as impermeability to sperm, and/or a lack of structures for sperm maintenance in the spermatheca. In the asexual parasitoid wasp <italic>Muscidifurax uniraptor,</italic> an essential spermatheca-associated muscle is completely absent (<xref ref-type="bibr" rid="bib11">Gottlieb and Zchori-Fein, 2001</xref>). Similarly, non-reproductive workers of many ant species have degenerated spermathecae with a flattened reservoir epithelium with few organelles (<xref ref-type="bibr" rid="bib10">Gobin <italic>et al.</italic>, 2008</xref>).</p><p>Different mechanisms have been suggested to drive the decay of female sexual traits. In <italic>A. japonica</italic> and other asexual species deriving from haplodiploid sexual ancestors, low copulation propensity and the absence of egg fertilization lead to cured asexual females producing only sons even when presented with mating opportunities, a pattern referred to as ‘functional virginity' (<xref ref-type="bibr" rid="bib18">Jeong and Stouthamer, 2005</xref>; <xref ref-type="bibr" rid="bib48">Stouthamer <italic>et al.</italic>, 2010</xref>; <xref ref-type="bibr" rid="bib42">Russell and Stouthamer, 2011</xref>). It has been suggested that such ‘functional virginity' would be favored selectively in sexual populations where only a fraction of females are asexual as a consequence of endosymbiont infection (<xref ref-type="bibr" rid="bib48">Stouthamer <italic>et al.</italic> 2010</xref>). Because ‘functional virginity' would enhance male production in populations with female-biased sex ratios, uninfected females might benefit from such a trait, assuming their sons could reproduce sexually with the majority of females in that population. ‘Functional virginity' may therefore become fixed in the population before the fixation of endosymbiont-induced asexuality. This could explain why female mating and egg-fertilization behaviors decay rapidly in species with endosymbiont-induced asexuality (<xref ref-type="bibr" rid="bib48">Stouthamer <italic>et al.</italic>, 2010</xref>), a pattern reported for many haplodiploid species (<xref ref-type="bibr" rid="bib36">Pijls <italic>et al.</italic>, 1996</xref>; <xref ref-type="bibr" rid="bib1">Arakaki <italic>et al.</italic>, 2000</xref>; <xref ref-type="bibr" rid="bib33">Pannebakker <italic>et al.</italic>, 2004a</xref>, <xref ref-type="bibr" rid="bib35">2005</xref>; <xref ref-type="bibr" rid="bib18">Jeong and Stouthamer, 2005</xref>; <xref ref-type="bibr" rid="bib42">Russell and Stouthamer, 2011</xref>; this study).</p><p>The spread of ‘functional virginity' mutations along with endosymbiont-induced asexuality is, however, not the only possible cause leading to the decay of mating behavior and egg fertilization in asexual females. A more general driving force underlying trait decay under asexuality might be strong negative selection on sexual traits that are expressed in females (<xref ref-type="bibr" rid="bib36">Pijls <italic>et al.</italic>, 1996</xref>; <xref ref-type="bibr" rid="bib51">van der Kooi and Schwander, 2014</xref>). Indeed, reduced female sexual attractiveness, lower copulation propensity and the absence of egg fertilization are ‘the norm' in asexual lineages (<xref ref-type="bibr" rid="bib51">van der Kooi and Schwander, 2014</xref>). The majority of known asexual lineages derive from sexual ancestors with genetic sex determination systems other than haplodiploidy. In these cases, virgin sexual females produce no offspring at all (rather than producing sons as under haplodiploidy), such that the decay of their mating behavior and lack of egg fertilization cannot be explained by the spread of ‘functional virginity' mutations.</p><p>To gain insights into the genetics of cured asexual females producing only sons, we monitored the proportion of such females across increasing levels of introgression of alleles from the asexual into the sexual strain. The proportion of females producing only sons increased gradually with an increasing representation of the asexual genome, indicating that alleles underlying asexuality-specific traits can indeed be introgressed into a sexual genome. The first significant increase was observed in the second generation, indicating that recessive genetic effects underlie an increased propensity of females to produce only sons (<xref ref-type="fig" rid="fig1">Figure 1</xref>). In addition, the proportion of second-generation (G2) hybrid females producing only sons matched the predicted proportion for a single-locus model (50%, <xref rid="tbl2" ref-type="table">Table 2</xref>). Cured asexual females produce only sons because of the combination of decreased sexual attractiveness, low copulation propensity and the absence of egg fertilization. Because we did not monitor each of these traits separately across the introgression generations, we cannot infer whether the single-locus effect that we found stems from changes at only one or some combination of these traits. Independently of whether the locus acts on a single or multiple traits, our study points to a rather simple genetic architecture underlying trait decay, possibly a single locus. A simple genetic architecture for other decayed sexual traits was also uncovered in three previous studies of independently derived wasp lineages with endosymbiont-induced asexuality. <xref ref-type="bibr" rid="bib33">Pannebakker <italic>et al.</italic> (2004a)</xref> found a single major locus underlying reduced male fertility using quantitative trait locus mapping. Two other studies investigated the decay of egg fertilization and found evidence for a single mutation with recessive effects in one case and dominant effects in the other (<xref ref-type="bibr" rid="bib18">Jeong and Stouthamer, 2005</xref>; <xref ref-type="bibr" rid="bib42">Russell and Stouthamer, 2011</xref>).</p><p>Two of the three studies investigating the genetic basis of female sexual trait decay in endosymbiont-induced asexuals found trait decay to be caused by recessive alleles (<xref ref-type="bibr" rid="bib18">Jeong and Stouthamer, 2005</xref>; this study). This raises the question of how recessive alleles could spread and fix within a short evolutionary time; asexuality in <italic>A. japonica</italic> evolved very recently, as indicated by little or no molecular genetic differentiation between sexual and asexual populations for mitochondrial DNA (<xref ref-type="bibr" rid="bib30">Murata <italic>et al.</italic>, 2009</xref>; <xref ref-type="bibr" rid="bib41">Reumer <italic>et al.</italic>, 2012</xref>). One possible explanation may lie in the cytological mechanism of <italic>Wolbachia</italic>-induced parthenogenesis. Diploidy of the embryo is frequently restored through gamete duplication, which results in complete homozygosity (for example, <xref ref-type="bibr" rid="bib47">Stouthamer and Kazmer, 1994</xref>; <xref ref-type="bibr" rid="bib12">Gottlieb <italic>et al.</italic>, 2002</xref>; <xref ref-type="bibr" rid="bib34">Pannebakker <italic>et al.</italic>, 2004b</xref>). Such homozygosity would facilitate spread of adaptive recessive mutations, similar to dominance favoring the spread of adaptive mutations in diploid species (<xref ref-type="bibr" rid="bib32">Otto and Goldstein, 1992</xref>; <xref ref-type="bibr" rid="bib31">Orr and Otto, 1994</xref>). It remains to be investigated whether recessive mutations may also frequently cause trait decay in species with forms of asexuality that maintain heterozygosity or whether trait decay in these cases is mostly caused by dominant mutations.</p><p>In addition to the females producing only sons, there was a (decreasing) fraction of females producing daughters in addition to sons across successive introgression generations. Copulation and egg-fertilization behaviors required for production of daughters most likely stem from the sexual rather than the asexual genome, given that cured asexual females almost never copulate, and do not fertilize their eggs in the rare event of copulation. As we needed daughters for initiating each successive introgression generation, we may have indirectly selected for a portion of the sexual genome favoring copulation and egg fertilization. Importantly, however, such putative indirect selection does not affect our conclusions for simple genetic effects underlying the increased tendency of females to produce only sons across successive introgression generations. Indeed, given the recessive effects underlying this tendency are only expressed in the second introgression generation, indirect selection would not have occurred prior to this generation, and we only used the first two generations to make inferences on genetic architecture. Following indirect selection for copulation and fertilization, we found that sex ratios among offspring of females with at least one daughter fluctuated across increasing levels of introgression, but were always more female-biased than sex ratios produced by ‘pure' sexual females (<xref ref-type="fig" rid="fig2">Figure 2</xref>).</p><p>We found that sexual <italic>A. japonica</italic> females change the sex ratio of their offspring according to the level of LMC their sons are exposed to (<xref ref-type="fig" rid="fig3">Figure 3</xref>), and we therefore monitored sex allocation in single vs multiple female set-ups across successive introgression generations. Plastic sex allocation was displayed by first generation sexual–asexual F1 hybrid females, but not by females with a higher level of introgression of the asexual genome. This indicates that plastic sex-allocation behavior has decayed in asexual females. Similar to the traits discussed above, the decay of plasticity is most likely also caused by recessive genetic effects, given that plasticity was still expressed in the first, but not in the second or any of the later introgression generations. The mechanisms underlying this decay remain a matter of speculation, but might be due to a loss of control over egg fertilization or to a loss of perception of levels of competition.</p><p>The decay of plastic sex-allocation response under asexuality has not been investigated previously, hence it is impossible to evaluate if it represents a general trend in asexuals derived from haplodiploid ancestors or if such decay is specific to asexual <italic>A. japonica</italic>. Nevertheless, selection experiments in the (sexual) spider mite <italic>Tetranychus urticae</italic> have revealed that plastic sex-allocation behavior can decay rapidly (within 54 generations) under situations of relaxed selection (<xref ref-type="bibr" rid="bib27">Macke <italic>et al.</italic>, 2011</xref>). Such rapid decay, whatever the causes underlying it, would likely result in decay of plastic sex allocation in most if not all asexual species.</p><p>In conclusion, we have investigated the potential decay of four female sexual traits in asexual <italic>A. japonica</italic>: sexual attractiveness, mating behavior, egg fertilization and plastic sex allocation under different levels of LMC, and we have found evidence for decay of all four traits. We have shown that the propensity for females to produce only sons is likely caused by a recessive allele at a single locus. Recessive genetic effects also caused the decay of sex-allocation plasticity in asexuals. Whether trait decay frequently stems from recessive genetic effects, or whether recessive effects may be specific to the decay of sexual traits in asexuals characterized by gamete duplication remains to be investigated. Genetic mapping studies in progress, facilitated by next-generation sequencing techniques, will provide insights into this question, as well as into the molecular causes of sexual trait decay.</p></sec><sec><title>Data archiving</title><p>Data available from the Dryad Digital Repository; doi:<ext-link xmlns:xlink="http://www.w3.org/1999/xlink" ext-link-type="doi" xlink:href="10.5061/dryad.rf0bh">10.5061/dryad.rf0bh</ext-link>.</p></sec>
Comparative evolutionary diversity and phylogenetic structure across multiple forest dynamics plots: a mega-phylogeny approach	<p>Forest dynamics plots, which now span longitudes, latitudes, and habitat types across the globe, offer unparalleled insights into the ecological and evolutionary processes that determine how species are assembled into communities. Understanding phylogenetic relationships among species in a community has become an important component of assessing assembly processes. However, the application of evolutionary information to questions in community ecology has been limited in large part by the lack of accurate estimates of phylogenetic relationships among individual species found within communities, and is particularly limiting in comparisons between communities. Therefore, streamlining and maximizing the information content of these community phylogenies is a priority. To test the viability and advantage of a multi-community phylogeny, we constructed a multi-plot mega-phylogeny of 1347 species of trees across 15 forest dynamics plots in the ForestGEO network using DNA barcode sequence data (<italic>rbc</italic>L, <italic>mat</italic>K, and <italic>psb</italic>A-<italic>trn</italic>H) and compared community phylogenies for each individual plot with respect to support for topology and branch lengths, which affect evolutionary inference of community processes. The levels of taxonomic differentiation across the phylogeny were examined by quantifying the frequency of resolved nodes throughout. In addition, three phylogenetic distance (PD) metrics that are commonly used to infer assembly processes were estimated for each plot [PD, Mean Phylogenetic Distance (MPD), and Mean Nearest Taxon Distance (MNTD)]. Lastly, we examine the partitioning of phylogenetic diversity among community plots through quantification of inter-community MPD and MNTD. Overall, evolutionary relationships were highly resolved across the DNA barcode-based mega-phylogeny, and phylogenetic resolution for each community plot was improved when estimated within the context of the mega-phylogeny. Likewise, when compared with phylogenies for individual plots, estimates of phylogenetic diversity in the mega-phylogeny were more consistent, thereby removing a potential source of bias at the plot-level, and demonstrating the value of assessing phylogenetic relationships simultaneously within a mega-phylogeny. An unexpected result of the comparisons among plots based on the mega-phylogeny was that the communities in the ForestGEO plots in general appear to be assemblages of more closely related species than expected by chance, and that differentiation among communities is very low, suggesting deep floristic connections among communities and new avenues for future analyses in community ecology.</p>	<contrib contrib-type="author"><name><surname>Erickson</surname><given-names>David L.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="author-notes" rid="fn001"><sup>*</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/127818"/></contrib><contrib contrib-type="author"><name><surname>Jones</surname><given-names>Frank A.</given-names></name><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><xref ref-type="aff" rid="aff3"><sup>3</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/126634"/></contrib><contrib contrib-type="author"><name><surname>Swenson</surname><given-names>Nathan G.</given-names></name><xref ref-type="aff" rid="aff4"><sup>4</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/175611"/></contrib><contrib contrib-type="author"><name><surname>Pei</surname><given-names>Nancai</given-names></name><xref ref-type="aff" rid="aff5"><sup>5</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/164730"/></contrib><contrib contrib-type="author"><name><surname>Bourg</surname><given-names>Norman A.</given-names></name><xref ref-type="aff" rid="aff6"><sup>6</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/165622"/></contrib><contrib contrib-type="author"><name><surname>Chen</surname><given-names>Wenna</given-names></name><xref ref-type="aff" rid="aff7"><sup>7</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/177343"/></contrib><contrib contrib-type="author"><name><surname>Davies</surname><given-names>Stuart J.</given-names></name><xref ref-type="aff" rid="aff8"><sup>8</sup></xref></contrib><contrib contrib-type="author"><name><surname>Ge</surname><given-names>Xue-jun</given-names></name><xref ref-type="aff" rid="aff7"><sup>7</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/188376"/></contrib><contrib contrib-type="author"><name><surname>Hao</surname><given-names>Zhanqing</given-names></name><xref ref-type="aff" rid="aff9"><sup>9</sup></xref></contrib><contrib contrib-type="author"><name><surname>Howe</surname><given-names>Robert W.</given-names></name><xref ref-type="aff" rid="aff10"><sup>10</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/186254"/></contrib><contrib contrib-type="author"><name><surname>Huang</surname><given-names>Chun-Lin</given-names></name><xref ref-type="aff" rid="aff11"><sup>11</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/186210"/></contrib><contrib contrib-type="author"><name><surname>Larson</surname><given-names>Andrew J.</given-names></name><xref ref-type="aff" rid="aff12"><sup>12</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/180560"/></contrib><contrib contrib-type="author"><name><surname>Lum</surname><given-names>Shawn K. Y.</given-names></name><xref ref-type="aff" rid="aff13"><sup>13</sup></xref></contrib><contrib contrib-type="author"><name><surname>Lutz</surname><given-names>James A.</given-names></name><xref ref-type="aff" rid="aff14"><sup>14</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/165207"/></contrib><contrib contrib-type="author"><name><surname>Ma</surname><given-names>Keping</given-names></name><xref ref-type="aff" rid="aff15"><sup>15</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/187076"/></contrib><contrib contrib-type="author"><name><surname>Meegaskumbura</surname><given-names>Madhava</given-names></name><xref ref-type="aff" rid="aff16"><sup>16</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/189897"/></contrib><contrib contrib-type="author"><name><surname>Mi</surname><given-names>Xiangcheng</given-names></name><xref ref-type="aff" rid="aff15"><sup>15</sup></xref></contrib><contrib contrib-type="author"><name><surname>Parker</surname><given-names>John D.</given-names></name><xref ref-type="aff" rid="aff17"><sup>17</sup></xref></contrib><contrib contrib-type="author"><name><surname>Fang-Sun</surname><given-names>I.</given-names></name><xref ref-type="aff" rid="aff18"><sup>18</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/186258"/></contrib><contrib contrib-type="author"><name><surname>Wright</surname><given-names>S. Joseph</given-names></name><xref ref-type="aff" rid="aff3"><sup>3</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/188421"/></contrib><contrib contrib-type="author"><name><surname>Wolf</surname><given-names>Amy T.</given-names></name><xref ref-type="aff" rid="aff10"><sup>10</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/186255"/></contrib><contrib contrib-type="author"><name><surname>Ye</surname><given-names>W.</given-names></name><xref ref-type="aff" rid="aff7"><sup>7</sup></xref></contrib><contrib contrib-type="author"><name><surname>Xing</surname><given-names>Dingliang</given-names></name><xref ref-type="aff" rid="aff9"><sup>9</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/164755"/></contrib><contrib contrib-type="author"><name><surname>Zimmerman</surname><given-names>Jess K.</given-names></name><xref ref-type="aff" rid="aff19"><sup>19</sup></xref></contrib><contrib contrib-type="author"><name><surname>Kress</surname><given-names>W. John</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref></contrib>	Frontiers in Genetics	<sec sec-type="introduction" id="s1"><title>Introduction</title><p>Phylogenetic hypotheses have played an increasingly important role in ecology over the last decade and their use in understanding community processes has been well reviewed (Webb et al., <xref rid="B55" ref-type="bibr">2002</xref>; Cavender-Bares et al., <xref rid="B6" ref-type="bibr">2009</xref>; Swenson, <xref rid="B51a" ref-type="bibr">2013</xref>). Knowledge of phylogenetic relationships among species has been used to quantify various aspects of ecology, including competition (Webb, <xref rid="B53" ref-type="bibr">2000</xref>; Kembel and Hubbell, <xref rid="B25" ref-type="bibr">2006</xref>; Webb et al., <xref rid="B54" ref-type="bibr">2008</xref>; Cavender-Bares et al., <xref rid="B6" ref-type="bibr">2009</xref>; Lebrija-Trejos et al., <xref rid="B30" ref-type="bibr">2013</xref>), environmental filtering (Cavender-Bares et al., <xref rid="B5" ref-type="bibr">2004</xref>; Uriarte et al., <xref rid="B52" ref-type="bibr">2010</xref>; Liu et al., <xref rid="B32" ref-type="bibr">2013</xref>; Pearse et al., <xref rid="B41" ref-type="bibr">2013</xref>), pathogen and herbivore selection (Gilbert and Webb, <xref rid="B14" ref-type="bibr">2007</xref>; Whitfeld et al., <xref rid="B58" ref-type="bibr">2012</xref>), succession (Whitfeld et al., <xref rid="B58" ref-type="bibr">2012</xref>) and the spatial differentiation of phylogenetic diversity (Weiblen et al., <xref rid="B57" ref-type="bibr">2006</xref>; Graham and Fine, <xref rid="B15" ref-type="bibr">2008</xref>; Fine and Kembel, <xref rid="B12" ref-type="bibr">2011</xref>). In the context of conservation biology, phylogenetic information has also been used to quantify diversity within and among communities (Faith, <xref rid="B11" ref-type="bibr">1992</xref>; Hardy and Senterre, <xref rid="B19" ref-type="bibr">2007</xref>). The best measure of diversity that is most relevant for conservation assessment remains an important question. For example, does species diversity or phylogenetic diversity best capture the full spectrum of organismal diversity and traits in a community or habitat to be conserved (e.g., Swenson, <xref rid="B51a" ref-type="bibr">2013</xref>)? Nonetheless, the ability of phylogenetic data to precisely quantify evolutionary history within and among communities provides a framework for addressing how best to quantify, manage and conserve biodiversity and communities.</p><p>The application of evolutionary information to questions in community ecology has been limited in large part by the lack of accurate estimates of phylogenetic relationships among individual species found within communities. This dearth of information has been particularly true for the most species- and ecologically-diverse communities in the tropics where existing phylogenetic data are most limiting (Webb and Donoghue, <xref rid="B56" ref-type="bibr">2005</xref>; Kress et al., <xref rid="B28" ref-type="bibr">2009</xref>). Traditionally, phylogenetic systematists have focused on taxonomic groups and lineages, not communities, on the assumption that phylogenetic treatments are most robust when all members of a clade are included in the analysis. In communities where diverse sets of species are present, the very large evolutionary divergences among co-occurring taxa and more sparse taxonomic sampling have been thought to hinder accurate reconstructions of phylogenetic relationships (Poe and Swofford, <xref rid="B43" ref-type="bibr">1999</xref>).</p><p>Newly emerging tools for constructing community phylogenies have largely ameliorated these concerns. Supertree methods, which prune and graft taxa from existing phylogenetic trees, can be used to construct phylogenetic relationships among species in a community (Bininda-Emonds and Sanderson, <xref rid="B3" ref-type="bibr">2001</xref>; Webb and Donoghue, <xref rid="B56" ref-type="bibr">2005</xref>). However, these methods have two drawbacks. Firstly, a phylogeny assembled from separate phylogenetic trees carries topological information, but contain no information on the evolutionary distances connecting species (i.e., branch lengths). Because the use of phylogenies in community ecology is specifically dependent upon evolutionary distances, branch lengths must be inferred. Assigning branch lengths to a topology with no intrinsic branch length information requires assumptions (e.g., bladj; Webb et al., <xref rid="B54" ref-type="bibr">2008</xref>) where the branch lengths between any two dated nodes are evenly divided among the nodes separating the dates, which is unrealistic. Secondly, unless the reference trees from which the super-phylogeny is constructed contain all members of the community, which is extremely unlikely particularly for diverse tropical communities, the relationships of many species will be inferred only at higher taxonomic levels where relationships are completely resolved (Kress et al., <xref rid="B28" ref-type="bibr">2009</xref>) and information about the tips of the phylogeny will be lost. Despite these limitations supertree-based community phylogenies have in many ways revolutionized community ecology. The availability of supertree tools, such as phylomatic (Webb and Donoghue, <xref rid="B56" ref-type="bibr">2005</xref>), has resulted in an explosion of interest in the merging of community ecology and phylogenetic systematics (Swenson, <xref rid="B51a" ref-type="bibr">2013</xref>).</p><p>A relatively new source of phylogenetic character information available to complement supertree methods in community ecology is DNA barcode sequence data. Multi-locus DNA barcodes for plants are composed of genes or parts of genes that have traditionally been used in molecular systematics (Soltis et al., <xref rid="B50" ref-type="bibr">2011</xref>). The community phylogenies that have been estimated from DNA barcode sequence data are robust and congruent with overall phylogenetic expectations for vascular plants (Kress et al., <xref rid="B28" ref-type="bibr">2009</xref>; Pei et al., <xref rid="B42" ref-type="bibr">2011</xref>; Whitfeld et al., <xref rid="B58" ref-type="bibr">2012</xref>; Yessoufou et al., <xref rid="B60" ref-type="bibr">2013</xref>). The advantage of these DNA barcode phylogenies is their ability to (1) better resolve relationships at the species-level in clades where supertree methods are less robust and (2) provide direct estimates of evolutionary distances (e.g., branch lengths) that connect clades within the phylogeny (Kress et al., <xref rid="B28" ref-type="bibr">2009</xref>).</p><p>Recently supertree methods have been combined with DNA barcode sequence data to enhance resolution in community phylogenies (e.g., Kress et al., <xref rid="B29" ref-type="bibr">2010</xref>). In these cases the phylogenetic relationships generated through supertree algorithms are a combination of broadly accepted patterns of taxonomic relationships at the deepest phylogenetic nodes provided by a guide or constraint tree while phylogenetic resolution among genera and species at the tips of the branches is provided by the rapidly evolving DNA barcode markers. Equally important is that branch lengths may be estimated with the DNA barcode sequence data throughout the tree, including the parts of the tree that are constrained. This merging of the two methods has been particularly fruitful in a number of community studies (e.g., Kress et al., <xref rid="B29" ref-type="bibr">2010</xref>; Uriarte et al., <xref rid="B52" ref-type="bibr">2010</xref>; Lebrija-Trejos et al., <xref rid="B30" ref-type="bibr">2013</xref>).</p><p>The next step in community analyses is to build multiple local phylogenies simultaneously that can be quantitatively compared. Currently most community phylogenies are constructed for one community at a time using different genes and different algorithms for estimating the phylogeny, as well as employing different dating methods, all of which will likely limit the ability to compare results among the communities. A few studies have employed molecular phylogenies to multiple communities (Swenson, <xref rid="B51" ref-type="bibr">2012</xref>), but most comparisons among communities have relied upon either species taxonomic lists (Ricklefs et al., <xref rid="B46" ref-type="bibr">2012</xref>) or taxonomic supertree methods (e.g., phylomatic). If we are to use phylogenetics to compare the structure, diversity, and ecological determinants of diversity among communities, then we must develop robust methods to build and employ multi-community phylogenies. Furthermore, an area in which the application of phylogenetic hypotheses to understanding ecological processes remains relatively less well explored is the geographic distribution of phylogenetic diversity and structure (Hardy and Jost, <xref rid="B18" ref-type="bibr">2008</xref>). The power of sequence-based phylogenies to resolve evolutionary relationships and calculate evolutionary distances within communities can now be applied to determining genetic differentiation and phylogenetic diversity among sites and communities by combining DNA barcode sequence data from multiple communities into a mega-phylogeny across these communities. The value of using these measures of phylogenetic diversity to assess the conservation status of communities representing various habitat types and regions across the globe should not be underestimated (e.g., Faith, <xref rid="B11" ref-type="bibr">1992</xref>).</p><p>In this study the ForestGEO (<ext-link ext-link-type="uri" xlink:href="http://www.forestgeo.si.edu">http://www.forestgeo.si.edu</ext-link>) global network of forest dynamics plots was used as the focus for developing a single large phylogeny for comparing measures of phylogenetic structure within and among plots. These plots have been developed over the last three decades to monitor forest change in different forest types around the world. Recently an effort has been initiated to generate DNA barcodes for tree species in each plot as a new tool for forensic ecology and community phylogenetics (e.g., Kress et al., <xref rid="B28" ref-type="bibr">2009</xref>, <xref rid="B29" ref-type="bibr">2010</xref>; Jones et al., <xref rid="B22" ref-type="bibr">2011</xref>; Pei et al., <xref rid="B42" ref-type="bibr">2011</xref>; Swenson, <xref rid="B51" ref-type="bibr">2012</xref>). Here a method is developed for reconstructing species relationships based on the DNA barcode sequence data in fifteen different ForestGEO plots simultaneously by constructing a single mega-phylogeny. The benefits of a simultaneous phylogenetic reconstruction are addressed by estimating branch lengths and evolutionary divergence within and among the individual plots. Finally, analyses of the geographic distribution of community structure, measures of phylogenetic diversity across these plots (e.g., Phylogenetic Diversity, Mean Phylogenetic Diversity, and Mean Nearest Taxon Density), and inferences into the mechanisms that produce these observed patterns are provided.</p></sec><sec sec-type="materials\|methods" id="s2"><title>Materials and methods</title><sec><title>Community sampling and genotyping</title><p>The samples for our analyses were obtained from 15 forest dynamics plots, which are part of the ForestGEO network organized by the Smithsonian Institution (<ext-link ext-link-type="uri" xlink:href="http://www.forestgeo.si.edu">http://www.forestgeo.si.edu</ext-link>; Figure <xref ref-type="fig" rid="F1">1</xref>). Some of these sites have been the focus of investigations into the application of DNA barcodes in understanding the processes of community ecology (e.g., Kress et al., <xref rid="B28" ref-type="bibr">2009</xref>, <xref rid="B29" ref-type="bibr">2010</xref>; Uriarte et al., <xref rid="B52" ref-type="bibr">2010</xref>; Pei et al., <xref rid="B42" ref-type="bibr">2011</xref>; Swenson, <xref rid="B51" ref-type="bibr">2012</xref>). We used samples from four plots in tropical Asia, two from sub-tropical Asia, one from temperate Asia, two from the neotropics, five from temperate North America, and one from temperate Europe (Table <xref ref-type="table" rid="T1">1</xref>). A total of 1347 species were included in the final dataset, encompassing 553 genera in 125 families and 43 orders.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>The distribution of the 15 ForestGEO plots incorporated into the mega-phylogeny are shown</bold>. The plots encompass temperate, sub-tropical and tropical habitats and are distributed globally.</p></caption><graphic xlink:href="fgene-05-00358-g0001"/></fig><table-wrap id="T1" position="float"><label>Table 1</label><caption><p><bold>Descriptions of the ForestGEO plots examined in this study are given</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="left" rowspan="1" colspan="1"><bold>Plot</bold></th><th align="center" rowspan="1" colspan="1"><bold>Species</bold></th><th align="center" rowspan="1" colspan="1"><bold>Genera</bold></th><th align="center" rowspan="1" colspan="1"><bold>Families</bold></th><th align="left" rowspan="1" colspan="1"><bold>Geography</bold></th><th align="left" rowspan="1" colspan="1"><bold>Habitat</bold></th><th align="center" rowspan="1" colspan="1"><bold>Coordinates</bold></th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">BCI</td><td align="center" rowspan="1" colspan="1">337</td><td align="center" rowspan="1" colspan="1">205</td><td align="center" rowspan="1" colspan="1">55</td><td align="left" rowspan="1" colspan="1">New-world</td><td align="left" rowspan="1" colspan="1">Tropics</td><td align="center" rowspan="1" colspan="1">8.63, −77.81</td></tr><tr><td align="left" rowspan="1" colspan="1">Bukit-Timah</td><td align="center" rowspan="1" colspan="1">326</td><td align="center" rowspan="1" colspan="1">177</td><td align="center" rowspan="1" colspan="1">61</td><td align="left" rowspan="1" colspan="1">Asian</td><td align="left" rowspan="1" colspan="1">Tropics</td><td align="center" rowspan="1" colspan="1">3.37, 98.92</td></tr><tr><td align="left" rowspan="1" colspan="1">Dinghushan</td><td align="center" rowspan="1" colspan="1">192</td><td align="center" rowspan="1" colspan="1">114</td><td align="center" rowspan="1" colspan="1">20</td><td align="left" rowspan="1" colspan="1">Asian</td><td align="left" rowspan="1" colspan="1">Sub-tropics</td><td align="center" rowspan="1" colspan="1">23.30, 114.54</td></tr><tr><td align="left" rowspan="1" colspan="1">Gutianshan</td><td align="center" rowspan="1" colspan="1">146</td><td align="center" rowspan="1" colspan="1">97</td><td align="center" rowspan="1" colspan="1">44</td><td align="left" rowspan="1" colspan="1">Asian</td><td align="left" rowspan="1" colspan="1">Sub-tropics</td><td align="center" rowspan="1" colspan="1">28.04, 121.08</td></tr><tr><td align="left" rowspan="1" colspan="1">Luquillo</td><td align="center" rowspan="1" colspan="1">141</td><td align="center" rowspan="1" colspan="1">107</td><td align="center" rowspan="1" colspan="1">39</td><td align="left" rowspan="1" colspan="1">New-world</td><td align="left" rowspan="1" colspan="1">Tropics</td><td align="center" rowspan="1" colspan="1">17.61, −67.68</td></tr><tr><td align="left" rowspan="1" colspan="1">Lienhuachih</td><td align="center" rowspan="1" colspan="1">129</td><td align="center" rowspan="1" colspan="1">79</td><td align="center" rowspan="1" colspan="1">49</td><td align="left" rowspan="1" colspan="1">Asian</td><td align="left" rowspan="1" colspan="1">Tropics</td><td align="center" rowspan="1" colspan="1">25.44, 120.27</td></tr><tr><td align="left" rowspan="1" colspan="1">Fushan</td><td align="center" rowspan="1" colspan="1">98</td><td align="center" rowspan="1" colspan="1">62</td><td align="center" rowspan="1" colspan="1">30</td><td align="left" rowspan="1" colspan="1">Asian</td><td align="left" rowspan="1" colspan="1">Tropics</td><td align="center" rowspan="1" colspan="1">24.21, 123.59</td></tr><tr><td align="left" rowspan="1" colspan="1">SCBI</td><td align="center" rowspan="1" colspan="1">62</td><td align="center" rowspan="1" colspan="1">37</td><td align="center" rowspan="1" colspan="1">52</td><td align="left" rowspan="1" colspan="1">New-world</td><td align="left" rowspan="1" colspan="1">Temperate</td><td align="center" rowspan="1" colspan="1">38.89, −78.14</td></tr><tr><td align="left" rowspan="1" colspan="1">Changbaishan</td><td align="center" rowspan="1" colspan="1">54</td><td align="center" rowspan="1" colspan="1">35</td><td align="center" rowspan="1" colspan="1">17</td><td align="left" rowspan="1" colspan="1">Asian</td><td align="left" rowspan="1" colspan="1">Temperate</td><td align="center" rowspan="1" colspan="1">42.38, 128. 08</td></tr><tr><td align="left" rowspan="1" colspan="1">Nanjenshan</td><td align="center" rowspan="1" colspan="1">42</td><td align="center" rowspan="1" colspan="1">36</td><td align="center" rowspan="1" colspan="1">17</td><td align="left" rowspan="1" colspan="1">Asian</td><td align="left" rowspan="1" colspan="1">Tropics</td><td align="center" rowspan="1" colspan="1">22.070, 122.73</td></tr><tr><td align="left" rowspan="1" colspan="1">Waibikon lake</td><td align="center" rowspan="1" colspan="1">30</td><td align="center" rowspan="1" colspan="1">23</td><td align="center" rowspan="1" colspan="1">18</td><td align="left" rowspan="1" colspan="1">New-world</td><td align="left" rowspan="1" colspan="1">Temperate</td><td align="center" rowspan="1" colspan="1">45.551, −88.78</td></tr><tr><td align="left" rowspan="1" colspan="1">SERC</td><td align="center" rowspan="1" colspan="1">28</td><td align="center" rowspan="1" colspan="1">20</td><td align="center" rowspan="1" colspan="1">15</td><td align="left" rowspan="1" colspan="1">New-world</td><td align="left" rowspan="1" colspan="1">Temperate</td><td align="center" rowspan="1" colspan="1">38.89, −76.56</td></tr><tr><td align="left" rowspan="1" colspan="1">Wytham</td><td align="center" rowspan="1" colspan="1">18</td><td align="center" rowspan="1" colspan="1">12</td><td align="center" rowspan="1" colspan="1">5</td><td align="left" rowspan="1" colspan="1">Europe</td><td align="left" rowspan="1" colspan="1">Temperate</td><td align="center" rowspan="1" colspan="1">51.77, −1.338</td></tr><tr><td align="left" rowspan="1" colspan="1">Wind river</td><td align="center" rowspan="1" colspan="1">7</td><td align="center" rowspan="1" colspan="1">4</td><td align="center" rowspan="1" colspan="1">3</td><td align="left" rowspan="1" colspan="1">New-world</td><td align="left" rowspan="1" colspan="1">Temperate</td><td align="center" rowspan="1" colspan="1">45.82, −121.95</td></tr><tr><td align="left" rowspan="1" colspan="1">Yosemite</td><td align="center" rowspan="1" colspan="1">7</td><td align="center" rowspan="1" colspan="1">5</td><td align="center" rowspan="1" colspan="1">4</td><td align="left" rowspan="1" colspan="1">New-world</td><td align="left" rowspan="1" colspan="1">Temperate</td><td align="center" rowspan="1" colspan="1">37.77, −119.82</td></tr><tr><td align="left" rowspan="1" colspan="1">Mega-phylogeny</td><td align="center" rowspan="1" colspan="1">1347</td><td align="center" rowspan="1" colspan="1">553</td><td align="center" rowspan="1" colspan="1">125</td><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/><td rowspan="1" colspan="1"/></tr></tbody></table><table-wrap-foot><p><italic>For each plot, the number of species, genera, and families is shown, as are general classification of the Geography, habitat type, and GPS coordinates. The number of species in the Mega-phylogeny is given, and is smaller than the sum among all communities due to shared species in some communities</italic>.</p></table-wrap-foot></table-wrap><p>Three samples per species were directly sequenced at three separate loci corresponding to the commonly used DNA barcode markers: (1) 552 bp of the ribulose-bisphosphate/carboxylase Large-subunit gene (<italic>rbc</italic>L;), (2) approximately 760 bp of the maturase-K gene (<italic>mat</italic>K), and (3) the <italic>psb</italic>A-<italic>trn</italic>H intergenic spacer (median 450 bp). All three markers are derived from the chloroplast genome. Methods for DNA extraction, PCR, and sequencing follow Kress et al. (<xref rid="B28" ref-type="bibr">2009</xref>) and Pei et al. (<xref rid="B42" ref-type="bibr">2011</xref>). Sequences for some of taxa were retrieved from GenBank (trees in Yosemite, Wind-River, and Wytham plots); for an individual species we used only our original sequence data or GenBank data and never combined original DNA barcode sequence data with GenBank data for the same species. All DNA barcode data generated for the study have been submitted to GenBank (see Supplemental Table <xref ref-type="supplementary-material" rid="SM1">S1</xref> for accession numbers for our original sequences and those retrieved from GenBank).</p></sec><sec><title>Sequence alignment</title><p>DNA barcode sequence data for trees collected from the 15 forest dynamics plots at each of the three separate markers were aligned across all species then concatenated together in an alignment supermatrix for estimation of phylogenetic relationships. The <italic>rbc</italic>L gene data were aligned through back-translation, using transAlign (Bininda-Emonds, <xref rid="B2" ref-type="bibr">2005</xref>). The <italic>mat</italic>K gene was also initially aligned using transAlign, and then adjusted manually to remove gaps corresponding to frame-shift mutations. Following manual adjustment of the alignment to remove gaps, the matrix was aligned a second time using MAFFT (Katoh and Standley, <xref rid="B23" ref-type="bibr">2013</xref>), implementing the FFT-NS-2 option for larger datasets. The <italic>psb</italic>A-<italic>trn</italic>H marker was aligned using SATe (Liu et al., <xref rid="B31" ref-type="bibr">2012</xref>), implementing the PRANK aligner (Löytynoja and Goldman, <xref rid="B33" ref-type="bibr">2005</xref>) for sub-groupings and the MUSCLE aligner (Edgar, <xref rid="B10" ref-type="bibr">2004</xref>) for merging sub-alignments. SATe is a “divide and conquer” style algorithm where an initial set of sequences is subdivided into smaller sets which are aligned and then joined back into a single alignment using a consensus alignment algorithm. SATe is iterative and goes through many cycles of generating sub-alignments and merging to consensus alignment using the likelihood score of a phylogenetic tree to determine an optimal alignment state. To improve the estimate of alignment in SATe, a guide tree derived from the Phylomatic portal (Webb and Donoghue, <xref rid="B56" ref-type="bibr">2005</xref>) was used as a starting tree in the alignment. The guide tree used in SATe was not a constraint tree, and thus the tree inferred from a final alignment in SATe may differ from the phylomatic input tree. SATe allowed us to generate a single alignment block for the hyper-variable <italic>psb</italic>A-<italic>trn</italic>H marker for all species, in contrast to sets of nested alignments as used previously (Kress et al., <xref rid="B28" ref-type="bibr">2009</xref>).</p></sec><sec><title>Phylogenetic reconstruction</title><p>The aligned 3-gene matrix was fully analyzed in the phylogenetic tree-building algorithm GARLI (Zwickl, <xref rid="B61" ref-type="bibr">2006</xref>) via the CIPRES portal (Miller et al., <xref rid="B36" ref-type="bibr">2010</xref>) to produce the 1347 taxon phylogeny that we call the “mega-phylogeny.” The configuration file used with GARLI is given in Supplemental Table <xref ref-type="supplementary-material" rid="SM2">S2</xref>. In addition to the aligned 3-gene matrix we utilized a phylogenetic constraint tree (described below). The aligned data-file was also partitioned by locus for use in GARLI, so that each of the three genes had separate model parameters estimated using the program MODELTEST 3.7 (Posada and Crandall, <xref rid="B44" ref-type="bibr">1998</xref>). The use of SATe greatly assisted model estimation at this stage because only a single model was required for the <italic>psb</italic>A-<italic>trn</italic>H marker, whereas with nested alignments either a single model would need to be chosen for all discrete alignment blocks (which would be artificial since the same model would not readily be chosen for all alignment partitions), or a very large number of models would be estimated separately for the same genetic locus. For a best tree search, 100 search replicates were initiated, each starting from random tree, to search for a best, most likely phylogeny. Further, we implemented a separate set of 100 bootstrap runs under the CAT-GAMMA model in GARLI, while still using the ordinal level constraint tree, to quantify support for the topology used in subsequent analyses.</p><p>Because of the relatively rapidly evolving sequence data provided by the DNA barcode markers and the inclusion of a large number of species spanning broad evolutionary distances, we employed a constraint tree to fix the deep phylogenetic relationships (Kress et al., <xref rid="B29" ref-type="bibr">2010</xref>). The search for the best tree was performed with a constraint tree derived from Phylomatic using the R20120829 phylogenetic tree for plants, derived from the Angiosperm Phylogeny Group III reconstruction (APGIII, <xref rid="B1" ref-type="bibr">2009</xref>). The constraint was modified in Mesquite (Maddison and Maddison, <xref rid="B34" ref-type="bibr">2014</xref>) in which each taxonomic order was reduced to a polytomy. This effect enforced phylogenetic relationships at the level of order and above. The molecular data were then responsible for reconstructing family, generic, and species relationships within orders. The quality of the phylogenetic reconstructions was evaluated by quantifying the fraction of resolved nodes, and the level of monophyly at the taxonomic family- and genus-levels. Although the constraint tree fixed relationships among orders according to APGIII, the branch lengths for all groups of taxa, including those fixed by the constraint-tree, were calculated from the aligned DNA barcode sequence alignment. As such, the combination of the constraint and sequences enabled phylogeny reconstruction by limiting the searched tree space and estimation of branch lengths across the depth of the tree.</p><p>In addition to constructing a single phylogeny for 15 ForestGEO community plots, phylogenetic relationships were estimated in each of the 15 plots separately. Taxa corresponding to each plot were pruned out from the aligned 3-marker matrix produced for the full 1347 taxon set and a phylogeny was constructed using the alignment for the taxa present in each plot as described above. Any benefits of high-taxon density to sequence alignment in the larger dataset were accordingly propagated to the estimates of alignment for each individual plot. For each of the 15 community plots, a best tree search with 100 independent search replicates was conducted in GARLI via the CIPRES portal using the same configuration parameters as the mega-phylogeny. The best scoring ML tree was used in subsequent comparisons between individually constructed community phylogeny and those estimated within the context of the mega-phylogeny.</p><p>To evaluate how well taxa were resolved in the mega-phylogeny and in individually constructed plot phylogenies, the fraction of non-zero length branches (that is, the fraction of resolved branches) were calculated for the entire mega-phylogeny, for individual plots that were pruned out of the mega-phylogeny, and for each individually constructed plot phylogeny. To compare how changes in taxonomic composition were associated with degree of phylogenetic resolution, spearman rank correlation was computed between the resolution of each phylogeny with species richness, Mean Phylogenetic Distance (MPD) and Mean Nearest Taxon Distance (MNTD), the latter described below. Similarly, we used spearman correlation to examine how rates of resolution changed as a function of latitude, as we moved from the tropics to temperate environments.</p></sec><sec><title>Mean path length (MPL) calibration of phylogeny</title><p>Mean Path Length (MPL) calibration (Britton et al., <xref rid="B4" ref-type="bibr">2002</xref>) was used to transform all molecular phylogenies into ultrametric chronogram. MPL estimates branch lengths using the mean of all branches descending from it, and thus is closer to molecular clock calibration. The algorithm was implemented using APE (Paradis et al., <xref rid="B40" ref-type="bibr">2004</xref>) implemented through the Picante package (Kembel et al., <xref rid="B24" ref-type="bibr">2010</xref>) of the R programming language (R Core Team, <xref rid="B45" ref-type="bibr">2012</xref>) with the “chonoMPL” command, setting the root age to 1, as opposed to attempting to assign any dates. This method was selected because (1) it most directly reflects inferred evolutionary distances (i.e., branch lengths) with the minimum of alteration of branch length relative to other methods of generating an ultrametric tree (Britton et al., <xref rid="B4" ref-type="bibr">2002</xref>), and (2) attempts to use Bayesian methods for branch length calibration (e.g., BEAST; Drummond and Rambaut, <xref rid="B9" ref-type="bibr">2007</xref>) were unable to reach a state where the optimization converged for the larger phylogenies. Thus, each of the 15 separately generated community phylogeny, and the mega-phylogeny were transformed with MPL and these transformed phylogenies were used in analysis of phylogenetic distance (PD) and diversity (Sections Phylogenetic Diversity Metrics and Comparative Community Phylogenetic Diversity and Structure).</p></sec><sec><title>Phylogenetic diversity metrics</title><p>Three common metrics of phylogenetic diversity were utilized to quantify differences among the 15 ForestGEO plot-based community phylogenies. All of these metrics were estimated within the Picante package (Kembel et al., <xref rid="B24" ref-type="bibr">2010</xref>) of the R programming language. For each plot community, the phylogenetic diversity was calculated and then the values observed were compared for individually constructed phylogenies and for those estimated within the mega-phylogeny. The PD metric (Faith, <xref rid="B11" ref-type="bibr">1992</xref>), which sums the branch lengths for any defined set of taxa in a phylogeny, is correlated with species richness, but greatly refines estimates of diversity by incorporating a quantitative measure of evolutionary divergence (Faith, <xref rid="B11" ref-type="bibr">1992</xref>; Forest et al., <xref rid="B13" ref-type="bibr">2007</xref>; Morlon et al., <xref rid="B37" ref-type="bibr">2011</xref>). For individually constructed community phylogenies, PD was simply the sum of all branch lengths in the phylogeny. For community phylogenies within the mega-phylogeny, PD was the sum of all branch lengths within the mega-phylogeny connecting the species belonging to that community.</p><p>The second metric utilized was MPD (Webb et al., <xref rid="B55" ref-type="bibr">2002</xref>), which obtains an average for the pair-wise PD across all pairs of taxa in a community. As such, MPD is not directly correlated with species number by default, and is strongly influenced by branch lengths at the deepest nodes of the phylogeny (Swenson, <xref rid="B51a" ref-type="bibr">2013</xref>). This metric gives an estimate of the overall divergence of taxonomic clades present in a community and is sensitive to replacement of taxa that differ in broad taxonomic placement.</p><p>The third metric employed was MNTD (Webb et al., <xref rid="B55" ref-type="bibr">2002</xref>), which provides an average of the distances between each species and its nearest phylogenetic neighbor in the community. MNTD quantifies the degree that a community may be a set of closely related species vs. a heterogeneous set of taxa from disparate taxonomic clades. MNTD is necessarily sensitive to replacement of closely related taxa and is much less sensitive to changes at the basal (or oldest) nodes of the phylogeny. For each of these terms, the phylogenetic diversity is inferred through the summed branch length distances connecting species in the phylogeny, thus distance is equivalent to diversity.</p><p>The absolute values of PD, MPD, and MNTD are not relevant here; rather the differences in these metrics estimated from independently derived phylogenies vs. those estimated from the mega-phylogeny are most important. To compare how estimates of phylogenetic diversity vary, the proportional difference for the values in each community were measured and values of difference were plotted for all 15-plot communities. For each metric, 15 values were calculated representing the difference between individually constructed plot phylogeny and values inferred from the mega-phylogeny. The percentage difference was calculated as: [(M<sub>i</sub> − M<sub>j</sub>)/M<sub>j</sub>]<sup>*</sup>100 where M = the metric under evaluation (PD, MPD, or MNTD), i = the value estimated from individually constructed community phylogeny and j = the value estimated from the mega-phylogeny. A value of zero corresponds to no difference in estimates of PD between that inferred in the mega-phylogeny and that from individually constructed phylogenies. We further examined if there was a significant correlation between latitude and phylogenetic diversity using the spearman correlation coefficient with decimal values of latitude for each community plot. Whereas species richness is known to exhibit a strong latitudinal gradient, we used this correlation to evaluate if phylogenetic diversity metrics exhibit similar patterns.</p></sec><sec><title>Comparative community phylogenetic diversity and structure</title><p>To compare the phylogenetic diversity and structure among ForestGEO plots, two methods were used, both estimated within the Picante package of the R programming language, and using the MPL transformed mega-phylogeny. The first metric was the Inter-community Mean Pairwise Distance, which is a measure of phylogenetic beta diversity (Webb et al., <xref rid="B55" ref-type="bibr">2002</xref>) and is calculated as the mean for all pair-wise comparisons of PD between the taxa of two different communities (the “mpd.comdist” routine within Picante). The second metric is the MNTD among nearest-neighbor pairs of species in different communities (the “comdistnt” routine within Picante) and is sensitive to higher-level taxonomic substitutions (i.e., changes in representation of taxonomic family or order) among communities. For mpd.comdist and comdistnt, both the mean and variance of the inter-community PDs were plotted.</p><p>To further test if each of the 15 ForestGEO plots was a random sample of the larger community of species represented by the mega-phylogeny, a randomization test implemented in Picante was used to estimate the standard effects size of each of the three PD metrics. This test was run for the three phylogenetic diversity metrics PD, MPD, and MNTD using the MPL transformed mega-phylogeny. For each of the three metrics, the algorithm in Picante was run using 999 randomizations of the community within the mega-phylogeny applying the “taxa.labels.” The “taxa.labels” model maintains the species richness of each community as well as the number of forest plots a particular species may be assigned to (i.e., a species observed in one forest can only be found in one forest in the randomized data), but alters the evolutionary relationships (i.e., branch lengths connecting species) in that community by randomizing the names of the species at the tip of the phylogenetic tree (Webb et al., <xref rid="B55" ref-type="bibr">2002</xref>). The model generates a distribution from the 999 independent randomizations, against which the observed value of phylogenetic diversity (PD, MPD, or MNTD) may then be compared and a <italic>p</italic>-value assigned to it. Communities with a <italic>p</italic>-value of <0.05 were judged to be significantly different from random within the context of the 15 plot mega-phylogeny. <italic>Z</italic>-values, observed and expected values of diversity, and <italic>p</italic>-value are given as supplemental data (Supplemental Tables <xref ref-type="supplementary-material" rid="SM3">S3</xref>–S5, respectively, for PD, MTD, and MNTD). Departures from random have been interpreted as a signal for local-level processes within communities, such that species with observed PDs significantly less than the randomized mean are more closely related than expected (i.e., phylogenetically clustered) and hence the result of environmental filtering on phylogenetically structured traits (Webb, <xref rid="B53" ref-type="bibr">2000</xref>). Alternatively, species with evolutionary distances significantly greater than the observed mean are more distantly related than expected (i.e., phylogenetically overdispersed), which is consistent with the role of competition in structuring species composition (Webb et al., <xref rid="B55" ref-type="bibr">2002</xref>). The entire ForestGEO mega-phylogeny was treated in essence as a global “meta-community” and as such these metrics provide evidence for similar ecological processes among communities that are linked to the environment or taxonomic structure.</p></sec></sec><sec sec-type="results" id="s3"><title>Results</title><sec><title>Phylogenetic reconstruction</title><p>Phylogenetic resolution, which is the fraction of non-zero length branches in a phylogeny, varied among the 15 single-plot phylogenies and the 15-plot mega-phylogeny. The 15-plot mega-phylogeny with molecular branch lengths selected from the most likely of 100 independent maximum-likelihood tree searches is shown in Figure <xref ref-type="fig" rid="F2">2</xref>. The distribution of the Orders throughout the 15-plot mega-phylogeny are presented in Figure <xref ref-type="fig" rid="F3a">3A</xref>; with the diversity of orders within each plot shown in Figure <xref ref-type="fig" rid="F3b">3B</xref>. The fraction of resolved species for the mega-phylogeny was over 78% using the phylogeny with the best likelihood score derived from 100 independent search replicates. A consensus tree from rapid bootstrapping of the mega-phylogeny found 70.2% of all nodes were supported using majority rule 50% criterion, which closely mirrored the 78% resolution in the highest scoring ML tree. The rates of resolution for the independently derived community phylogenies (Table <xref ref-type="table" rid="T2">2</xref>) ranged from 81% (Dinghushan) to 100% (Wytham and Yosemite). A significant relationship was found between phylogenetic resolution and species richness (<italic>r</italic> = −0.799, <italic>p</italic> > 0.001), as smaller community phylogenies (and those at higher latitudes) were more likely to be fully resolved. Importantly, however, phylogenetic resolution for a plot was consistently higher when estimated within the context of the mega-phylogeny (Table <xref ref-type="table" rid="T2">2</xref>). On average a 3.5% increase in resolution was found, ranging from an 8% increase for Bukit-Timah and Changbaishan to no increase for Wind-River and Yosemite (Table <xref ref-type="table" rid="T2">2</xref>).</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Representation of the ForestGEO 15-plot mega-phylogeny, reconstructed with Maximum-Likelihood, shown with un-transformed branch lengths</bold>.</p></caption><graphic xlink:href="fgene-05-00358-g0002"/></fig><fig id="F3a" position="float"><label>Figure 3A</label><caption><p><bold>Phylogenetic relationships of taxa in the 15 ForestGEO plots as a mega-phylogeny and as separate plots resolved at the level of taxonomic family</bold>. A cladogram of the ForestGEO 15-plot mega-phylogeny, with 1347 taxa derived from molecular data is presented. Seven separate major phylogenetic groups of vascular plants are indicated to demonstrate the evolutionary diversity of species included in the mega-phylogeny. The composition of the mega-phylogeny is broadly congruent with land plant relationships showing high diversity in the Asterid, Rosid, and Basal Eudicot clades, and very low diversity among Monilophytes and Gymnosperm clades.</p></caption><graphic xlink:href="fgene-05-00358-g0003"/></fig><fig id="F3b" position="float"><label>Figure 3B</label><caption><p><bold>Individual cladograms for each of the 15 separate ForestGEO plots arranged by species richness</bold>. The families that are present in each individual plot are mapped on the mega-phylogeny in red to show the evolutionary and taxonomic diversity present in each plot.</p></caption><graphic xlink:href="fgene-05-00358-g0004"/></fig><table-wrap id="T2" position="float"><label>Table 2</label><caption><p><bold>Fraction of resolved nodes within the ForestGEO15 mega-phylogeny and each of the individual plots when estimated separately</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="left" rowspan="1" colspan="1"><bold>Plot</bold></th><th align="center" rowspan="1" colspan="1"><bold># Taxa</bold></th><th align="center" rowspan="1" colspan="1"><bold>Individually constructed</bold></th><th align="center" rowspan="1" colspan="1"><bold>Mega-phylogeny</bold></th><th align="center" rowspan="1" colspan="1"><bold>Difference</bold></th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">ForestGEO15</td><td align="center" rowspan="1" colspan="1">1347</td><td align="center" rowspan="1" colspan="1">n/a</td><td align="center" rowspan="1" colspan="1">0.78</td><td align="center" rowspan="1" colspan="1">n/a</td></tr><tr><td align="left" rowspan="1" colspan="1">BCI</td><td align="center" rowspan="1" colspan="1">337</td><td align="center" rowspan="1" colspan="1">0.89</td><td align="center" rowspan="1" colspan="1">0.93</td><td align="center" rowspan="1" colspan="1">0.04</td></tr><tr><td align="left" rowspan="1" colspan="1">Bukit-Timah</td><td align="center" rowspan="1" colspan="1">326</td><td align="center" rowspan="1" colspan="1">0.86</td><td align="center" rowspan="1" colspan="1">0.94</td><td align="center" rowspan="1" colspan="1">0.08</td></tr><tr><td align="left" rowspan="1" colspan="1">Dinghushan</td><td align="center" rowspan="1" colspan="1">192</td><td align="center" rowspan="1" colspan="1">0.81</td><td align="center" rowspan="1" colspan="1">0.85</td><td align="center" rowspan="1" colspan="1">0.04</td></tr><tr><td align="left" rowspan="1" colspan="1">Gutianshan</td><td align="center" rowspan="1" colspan="1">146</td><td align="center" rowspan="1" colspan="1">0.87</td><td align="center" rowspan="1" colspan="1">0.93</td><td align="center" rowspan="1" colspan="1">0.06</td></tr><tr><td align="left" rowspan="1" colspan="1">Luquillo</td><td align="center" rowspan="1" colspan="1">141</td><td align="center" rowspan="1" colspan="1">0.95</td><td align="center" rowspan="1" colspan="1">0.97</td><td align="center" rowspan="1" colspan="1">0.02</td></tr><tr><td align="left" rowspan="1" colspan="1">Lienhuachih</td><td align="center" rowspan="1" colspan="1">129</td><td align="center" rowspan="1" colspan="1">0.88</td><td align="center" rowspan="1" colspan="1">0.92</td><td align="center" rowspan="1" colspan="1">0.04</td></tr><tr><td align="left" rowspan="1" colspan="1">Fushan</td><td align="center" rowspan="1" colspan="1">98</td><td align="center" rowspan="1" colspan="1">0.89</td><td align="center" rowspan="1" colspan="1">0.91</td><td align="center" rowspan="1" colspan="1">0.02</td></tr><tr><td align="left" rowspan="1" colspan="1">SCBI</td><td align="center" rowspan="1" colspan="1">62</td><td align="center" rowspan="1" colspan="1">0.89</td><td align="center" rowspan="1" colspan="1">0.94</td><td align="center" rowspan="1" colspan="1">0.05</td></tr><tr><td align="left" rowspan="1" colspan="1">Changbaishan</td><td align="center" rowspan="1" colspan="1">54</td><td align="center" rowspan="1" colspan="1">0.85</td><td align="center" rowspan="1" colspan="1">0.93</td><td align="center" rowspan="1" colspan="1">0.08</td></tr><tr><td align="left" rowspan="1" colspan="1">Nanjenshan</td><td align="center" rowspan="1" colspan="1">42</td><td align="center" rowspan="1" colspan="1">0.95</td><td align="center" rowspan="1" colspan="1">0.96</td><td align="center" rowspan="1" colspan="1">0.01</td></tr><tr><td align="left" rowspan="1" colspan="1">SERC</td><td align="center" rowspan="1" colspan="1">30</td><td align="center" rowspan="1" colspan="1">0.92</td><td align="center" rowspan="1" colspan="1">0.97</td><td align="center" rowspan="1" colspan="1">0.05</td></tr><tr><td align="left" rowspan="1" colspan="1">Wabikon lake</td><td align="center" rowspan="1" colspan="1">28</td><td align="center" rowspan="1" colspan="1">0.95</td><td align="center" rowspan="1" colspan="1">0.98</td><td align="center" rowspan="1" colspan="1">0.03</td></tr><tr><td align="left" rowspan="1" colspan="1">Wytham</td><td align="center" rowspan="1" colspan="1">18</td><td align="center" rowspan="1" colspan="1">1</td><td align="center" rowspan="1" colspan="1">1</td><td align="center" rowspan="1" colspan="1">0</td></tr><tr><td align="left" rowspan="1" colspan="1">Wind river</td><td align="center" rowspan="1" colspan="1">7</td><td align="center" rowspan="1" colspan="1">1</td><td align="center" rowspan="1" colspan="1">1</td><td align="center" rowspan="1" colspan="1">0</td></tr><tr><td align="left" rowspan="1" colspan="1">Yosemite</td><td align="center" rowspan="1" colspan="1">7</td><td align="center" rowspan="1" colspan="1">1</td><td align="center" rowspan="1" colspan="1">1</td><td align="center" rowspan="1" colspan="1">0</td></tr></tbody></table><table-wrap-foot><p><italic>The fraction of non-zero length nodes in the phylogeny was used to determine the percent resolution for the best-supported ML phylogeny</italic>.</p></table-wrap-foot></table-wrap><p>A significant relationship was found between MNTD for a plot and its phylogenetic resolution (<italic>r</italic> = 0.874; <italic>p</italic> > 0.001), with higher MNTD equating to improved resolution. A similar effect was seen with MPD (<italic>r</italic> = 0.658; <italic>p</italic> = 0.008). The relationship of MNTD with phylogenetic resolution paralleled the observation of species richness and phylogenetic resolution, and was similar to correlation with latitude (<italic>r</italic> = 0.397, <italic>p</italic> = 0.142), such that as communities were composed of fewer species, it was easier to distinguish among them topologically.</p></sec><sec><title>Community phylogenetic diversity and structure</title><p>The three diversity metrics (PD, MPD, and MNTD) calculated for each plot varied for those derived from the mega-phylogeny vs. the individually constructed plot phylogenies (Figure <xref ref-type="fig" rid="F4">4</xref>). A weak relationship was observed between species richness and the proportional difference for PD (<italic>r</italic> = 0.393, <italic>p</italic> = 0.083), but exhibited a significant positive relationship for MPD (<italic>r</italic> = 0.741, <italic>p</italic> = 0.002) and MNTD (<italic>r</italic> = 0.525, <italic>p</italic> = 0.028) as larger plots exhibited less differentiation in the estimated metrics (Figure <xref ref-type="fig" rid="F4">4</xref>). Averaged over all communities, the percent difference in estimated PD was, PD = 14.38%, MPD = 2.297%, and MNTD = 38.76%. The percent difference for MNTD was striking, and is most evident in the smallest plots with a range of 60% divergence for Changbaishan, to 15% divergence for BCI (Figure <xref ref-type="fig" rid="F4">4</xref>), which reflects the difficulty that phylogenetic reconstruction methods may have in inferring evolutionary distances when the mean of those distances is very large. The improvements in estimates of PD within the mega-phylogeny are most dramatic for the smallest plots where the higher taxon density of the mega-phylogeny greatly improves estimates of branch lengths among all species found in those communities. The inter-plot Mean Phylogenetic Distance (inter-MPD) was broadly similar for 13 of the 15 plots (Figure <xref ref-type="fig" rid="F5">5</xref>), with only the most species poor plots (e.g., Wind-River and Yosemite) differing significantly from the other 13 plots. This reflects the wide taxonomic composition of many of the plots, where high variation within plots obscures differentiation among the plots, as seen through taxonomic representation of different orders within each plot (Figure <xref ref-type="fig" rid="F3b">3B</xref>). Similarly, the inter-plot Mean Nearest Taxon Distance (inter-MNTD) exhibited no differentiation among any of the ForestGEO plots, regardless of geographic location or species richness (Figure <xref ref-type="fig" rid="F4">4</xref>).</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>The percentage difference in observed value of PD, MNTD, and MPD are plotted for each community</bold>. Each point is the percent difference in the value of a metric calculated from individually constructed community phylogeny vs. that observed for the same community in the mega-phylogeny. Values are plotted as a function of Species Richness of the ForestGEO community.</p></caption><graphic xlink:href="fgene-05-00358-g0005"/></fig><fig id="F5" position="float"><label>Figure 5</label><caption><p><bold>Two methods to infer differentiation among communities are shown, with the inter-community MNTD (top) and inter-community MPD (bottom)</bold>. Boxplots for each community show the mean (dark bar within box), interquartile range (box), and 95% confidence interval (whisker bars), computed from all pairwise contrasts between plots.</p></caption><graphic xlink:href="fgene-05-00358-g0006"/></fig><p>In contrast to the inter-community diversity metrics, randomization tests, which evaluate if communities are a random subsample of the larger phylogeny, found that the communities were not a random set of species (Table <xref ref-type="table" rid="T3">3</xref>). In the three PD metrics used, all three exhibited significant differences from random in the most speciose plots, with a consistent trend toward their being significantly clustered (Table <xref ref-type="table" rid="T3">3</xref>, and Supplemental Tables <xref ref-type="supplementary-material" rid="SM3">S3</xref>–S5 for PD, MPD, and MNTD, respectively). For PD, the five temperate sites exhibited no departure from random, whereas each of the plots with more than 62 species (excepting Luquillo) was significantly clustered. For MNTD the result was even more skewed with 12 of the 15 plots exhibiting significant clustering. For MPD significant clustering was found for the four most species rich tropical plots (BCI, Bukit-Timah, Dinghushan, and Gutianshan), whereas the most species-poor community plots were inferred to be overdispersed (Wabikon Lake, Wind River, Wytham, and Yosemite). Overall the eight tropical or sub-tropical plots, when considered over all three PD metrics, were significantly clustered in 15 out of 24 cases. In the remaining nine cases they were not different from random, and none were inferred to be over-dispersed. Alternatively for the seven species-poor temperate plots, four were overdispersed (only with MPD), eight were significantly clustered (seven for MNTD and one for PD with Changbaishan), and the remaining 12 showed no departure from random (Table <xref ref-type="table" rid="T3">3</xref>). Two plots, Luquillo and Nanjenshan, were consistent in exhibiting no significant departures from random for any of the phylogenetic diversity metrics whereas all other of the plot phylogenies exhibited some significant departure from random for at least one of the metrics.</p><table-wrap id="T3" position="float"><label>Table 3</label><caption><p><bold>Values for three species richness (SR) and three Phylogenetic Diversity metrics Phylogenetic Distance (PD), Mean Phylogenetic Distance (MPD), and Mean Nearest Taxon Distance (MNTD) are given for each plot</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="left" rowspan="1" colspan="1"><bold>Plot</bold></th><th align="center" rowspan="1" colspan="1"><bold>SR</bold></th><th align="center" rowspan="1" colspan="1"><bold>PD</bold></th><th align="center" rowspan="1" colspan="1"><bold>MPD</bold></th><th align="center" rowspan="1" colspan="1"><bold>MNTD</bold></th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">BCI</td><td align="center" rowspan="1" colspan="1">337</td><td align="center" rowspan="1" colspan="1"><bold>28.88</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td><td align="center" rowspan="1" colspan="1"><bold>0.61</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td><td align="center" rowspan="1" colspan="1">0.09</td></tr><tr><td align="left" rowspan="1" colspan="1">Bukit-Timah</td><td align="center" rowspan="1" colspan="1">326</td><td align="center" rowspan="1" colspan="1"><bold>25.80</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td><td align="center" rowspan="1" colspan="1"><bold>0.60</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td><td align="center" rowspan="1" colspan="1"><bold>0.08</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td></tr><tr><td align="left" rowspan="1" colspan="1">Dinghushan</td><td align="center" rowspan="1" colspan="1">192</td><td align="center" rowspan="1" colspan="1"><bold>18.55</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td><td align="center" rowspan="1" colspan="1"><bold>0.72</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td><td align="center" rowspan="1" colspan="1"><bold>0.09</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td></tr><tr><td align="left" rowspan="1" colspan="1">Gutianshan</td><td align="center" rowspan="1" colspan="1">146</td><td align="center" rowspan="1" colspan="1"><bold>16.1</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td><td align="center" rowspan="1" colspan="1"><bold>0.60</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td><td align="center" rowspan="1" colspan="1"><bold>0.12</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td></tr><tr><td align="left" rowspan="1" colspan="1">Luquillo</td><td align="center" rowspan="1" colspan="1">141</td><td align="center" rowspan="1" colspan="1"><bold>19.57</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td><td align="center" rowspan="1" colspan="1">0.67</td><td align="center" rowspan="1" colspan="1">0.14</td></tr><tr><td align="left" rowspan="1" colspan="1">Lienhuachih</td><td align="center" rowspan="1" colspan="1">129</td><td align="center" rowspan="1" colspan="1"><bold>14.17</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td><td align="center" rowspan="1" colspan="1">0.62</td><td align="center" rowspan="1" colspan="1"><bold>0.11</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td></tr><tr><td align="left" rowspan="1" colspan="1">Fushan</td><td align="center" rowspan="1" colspan="1">98</td><td align="center" rowspan="1" colspan="1"><bold>12.67</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td><td align="center" rowspan="1" colspan="1">0.86</td><td align="center" rowspan="1" colspan="1"><bold>0.12</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td></tr><tr><td align="left" rowspan="1" colspan="1">SCBI</td><td align="center" rowspan="1" colspan="1">62</td><td align="center" rowspan="1" colspan="1"><bold>8.67</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td><td align="center" rowspan="1" colspan="1">0.61</td><td align="center" rowspan="1" colspan="1"><bold>0.13</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td></tr><tr><td align="left" rowspan="1" colspan="1">Changbaishan</td><td align="center" rowspan="1" colspan="1">54</td><td align="center" rowspan="1" colspan="1"><bold>7.15</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td><td align="center" rowspan="1" colspan="1">0.69</td><td align="center" rowspan="1" colspan="1"><bold>0.10</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td></tr><tr><td align="left" rowspan="1" colspan="1">Nanjenshan</td><td align="center" rowspan="1" colspan="1">42</td><td align="center" rowspan="1" colspan="1">8.27</td><td align="center" rowspan="1" colspan="1">0.59</td><td align="center" rowspan="1" colspan="1">0.23</td></tr><tr><td align="left" rowspan="1" colspan="1">SERC</td><td align="center" rowspan="1" colspan="1">30</td><td align="center" rowspan="1" colspan="1">6.19</td><td align="center" rowspan="1" colspan="1">0.66</td><td align="center" rowspan="1" colspan="1"><bold>0.19</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td></tr><tr><td align="left" rowspan="1" colspan="1">Wabikon lake</td><td align="center" rowspan="1" colspan="1">28</td><td align="center" rowspan="1" colspan="1">5.63</td><td align="center" rowspan="1" colspan="1"><bold>0.75</bold><inline-graphic xlink:href="fgene-05-00358-i0002.jpg"/></td><td align="center" rowspan="1" colspan="1"><bold>0.18</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td></tr><tr><td align="left" rowspan="1" colspan="1">Wytham</td><td align="center" rowspan="1" colspan="1">18</td><td align="center" rowspan="1" colspan="1">4.38</td><td align="center" rowspan="1" colspan="1"><bold>0.78</bold><inline-graphic xlink:href="fgene-05-00358-i0002.jpg"/></td><td align="center" rowspan="1" colspan="1"><bold>0.17</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td></tr><tr><td align="left" rowspan="1" colspan="1">Wind river</td><td align="center" rowspan="1" colspan="1">7</td><td align="center" rowspan="1" colspan="1">3.13</td><td align="center" rowspan="1" colspan="1"><bold>0.92</bold><inline-graphic xlink:href="fgene-05-00358-i0002.jpg"/></td><td align="center" rowspan="1" colspan="1"><bold>0.31</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td></tr><tr><td align="left" rowspan="1" colspan="1">Yosemite</td><td align="center" rowspan="1" colspan="1">7</td><td align="center" rowspan="1" colspan="1">3.00</td><td align="center" rowspan="1" colspan="1"><bold>0.79</bold><inline-graphic xlink:href="fgene-05-00358-i0002.jpg"/></td><td align="center" rowspan="1" colspan="1"><bold>0.31</bold><inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/></td></tr></tbody></table><table-wrap-foot><p><italic>For each metric 999 randomizations were used to assess departure from random community structure. Significant differences from random are in bold, with pattern denoted by superscript. Standard effect sizes, Z and p-values are reported in Supplemental Tables <xref ref-type="supplementary-material" rid="SM3">S3</xref>–S5</italic>.</p><p><inline-graphic xlink:href="fgene-05-00358-i0002.jpg"/>, <italic>Significant Overdispersion;</italic> <inline-graphic xlink:href="fgene-05-00358-i0001.jpg"/>, <italic>Significant Clustering</italic>.</p></table-wrap-foot></table-wrap></sec></sec><sec sec-type="discussion" id="s4"><title>Discussion</title><p>In the field of ecology phylogenetic data have been used to understand ecological processes (Webb et al., <xref rid="B55" ref-type="bibr">2002</xref>; Cavender-Bares et al., <xref rid="B6" ref-type="bibr">2009</xref>), the roles of trait conservatism and dispersal limitation in structuring communities (Fine and Kembel, <xref rid="B12" ref-type="bibr">2011</xref>; Liu et al., <xref rid="B32" ref-type="bibr">2013</xref>), and the regulation of beta diversity (Swenson, <xref rid="B51" ref-type="bibr">2012</xref>). In addition, phylogenetic information has been applied to the identification of specific environments critical for conservation (Faith, <xref rid="B11" ref-type="bibr">1992</xref>; Forest et al., <xref rid="B13" ref-type="bibr">2007</xref>; Morlon et al., <xref rid="B37" ref-type="bibr">2011</xref>). Accordingly, the ability to generate and use phylogenetic data to address core questions in ecology and to assess conservation priorities are of increasing importance.</p><p>The results shown here demonstrate that constructing a single mega-phylogeny inclusive of many individual community plots improves the estimation of the evolutionary relationships and distances among species in each separate plot. The mega-phylogeny is also helpful in examining the patterns of phylogenetic diversity within and among plots to explore broad scale patterns that may reflect processes regulating community assembly and the maintenance of diversity. Long-term biodiversity monitoring plots, such as the ForestGEO network, provide an ideal context for investigating phylogenetic diversity and geographic structuring among plots to address questions regarding community assembly at very broad scales.</p><sec><title>Generating phylogenies</title><p>The use of a constraint tree to construct the mega-phylogeny was adopted in this study and it is recommended for use in large community phylogenies, particularly those built with rapidly evolving sequence data as found in DNA barcodes (Kress et al., <xref rid="B29" ref-type="bibr">2010</xref>). For example, the non-protein coding marker <italic>psb</italic>A<italic>-trn</italic>H has been used phylogenetically at very low taxonomic scales (e.g., within genera or families) because of the difficulty in aligning sequences among distantly related taxa. This limitation has slowed its adoption as an official DNA barcode marker (Hollingsworth et al., <xref rid="B20" ref-type="bibr">2011</xref>). However, in this study we were able to use the SATe algorithm to align <italic>psb</italic>A-t<italic>rn</italic>H across all species, including distantly related ones, in the analysis rather than as in prior studies in which the marker was aligned in a nested format within a supermatrix and did not contribute to the inferred relationships of deeper taxonomic scales (Kress et al., <xref rid="B28" ref-type="bibr">2009</xref>; Pei et al., <xref rid="B42" ref-type="bibr">2011</xref>). This marker evolves very rapidly and global alignment may have contributed to the non-constrained mega-phylogeny exhibiting differentiation from expectations in APGIII. However, the use of <italic>psb</italic>A<italic>-trn</italic>H in a global alignment produced a higher fraction of resolved nodes than the use of only <italic>rbc</italic>L+<italic>mat</italic>K, and did not negatively affect rates of family and generic monophyly (Table <xref ref-type="table" rid="T1">1</xref>). Also, a nested approach to alignment of <italic>psb</italic>A-<italic>trn</italic>H requires some subjective decisions with regards to the scale at which to group sequences, which may result in the exclusion of sequences from taxa that are not readily included in groupings. This effect in turn will result in a greater asymmetry in the aligned sequence matrix, and, therefore, will complicate model selection for different data partitions in phylogenetic inference. For these reasons we recommend a global alignment of <italic>psb</italic>A-<italic>trn</italic>H in plant DNA barcode phylogenies using SATe in conjunction with a constraint tree that will enforce higher-level taxonomic resolutions.</p><p>Even the relatively limited sequence content from DNA barcode markers, as demonstrated here, can be successfully used to the construct a highly robust phylogeny across multiple plots with high rates of resolution and monophyly. When compared with other studies of very large phylogenies, the mega-phylogeny had comparable rates of resolution among species (Smith et al., <xref rid="B48" ref-type="bibr">2009</xref>, <xref rid="B49" ref-type="bibr">2011</xref>), and an overall remarkably high rate of 78% taxonomic resolution. The 15-plot mega-phylogeny with 1347 species in 43 orders and 125 families (Table <xref ref-type="table" rid="T1">1</xref>, Figure <xref ref-type="fig" rid="F2">2</xref>) was significantly larger than the individual plots in which the average was 12 orders and 38 families (Table <xref ref-type="table" rid="T1">1</xref>). The mega-phylogeny improved resolution among species in most communities relative to constructing phylogenies for individual plots (Table <xref ref-type="table" rid="T2">2</xref>). The construction of a community phylogeny is greatly improved in the context of resolving difficult taxonomic relationships when taxon density is high (Smith et al., <xref rid="B49" ref-type="bibr">2011</xref>) and the lower level of taxonomic resolution in the mega-phylogeny as a whole does not affect the inferred rates of resolution for the included plots. The increased taxon density of the mega-phylogeny represented by a lower estimate of the MNTD was a central driver in improving rates of phylogenetic resolution (see Supplemental Table S4). As the genetic distances among species become more continuous and evenly distributed, the ability to infer phylogenetic relationships increases, which is reflected in the strong correlation between decreasing MPD and increasing phylogenetic resolution (0.73). Therefore, as ever-larger mega-phylogenies are generated to include an expanded scope of land plant diversity, then more fully resolved and well-supported community phylogenies can be pruned from them.</p></sec><sec><title>Improving phylogenetic resolution</title><p>Improving the accuracy of relationships among species in a community phylogeny is not just a methodological detail. Poorly resolved phylogenies can result in biased estimates of the diversity metrics used to infer ecological process (Davies et al., <xref rid="B8" ref-type="bibr">2012</xref>) or may lead to very different conclusions about ecological process in a particular community (Kress et al., <xref rid="B28" ref-type="bibr">2009</xref>). The low rates of taxonomic resolution in supertrees relative to molecular derived community phylogenies may adversely affect ecological inference (Kress et al., <xref rid="B28" ref-type="bibr">2009</xref>); yet with supertrees, at least all samples in a study are assembled and dated similarly, and thus results observed among communities are consistent and comparable (Fine and Kembel, <xref rid="B12" ref-type="bibr">2011</xref>). The challenge of collecting genetic data for all the members of a community has limited the use of molecular phylogeny in studies of community ecology, particularly in studies comparing across multiple communities (Swenson, <xref rid="B51" ref-type="bibr">2012</xref>). With the widespread generation of DNA barcode data across tropical plots, such as the ForestGEO network of forest dynamics plots, information on phylogenetic relationships can now be applied to many communities simultaneously. The benefits of constructing phylogenies for multiple communities concurrently as well as the advantages of increased taxonomic resolution and more accurate evolutionary distances among species and clades are many. Because evolutionary distance, or branch lengths, are necessary to infer processes of community assembly, one of our goals was to quantify the improvement of estimating evolutionary distances through the use of a mega-phylogeny of many plots to construct phylogenies of individual plots.</p><p>Nearly all studies of community phylogenetics have examined one community at a time. In most cases the community phylogenies were constructed using supertree methods, including phylomatic (Webb, <xref rid="B53" ref-type="bibr">2000</xref>; Cavender-Bares et al., <xref rid="B5" ref-type="bibr">2004</xref>; Fine and Kembel, <xref rid="B12" ref-type="bibr">2011</xref>) or direct sequence data (Kress et al., <xref rid="B28" ref-type="bibr">2009</xref>; Uriarte et al., <xref rid="B52" ref-type="bibr">2010</xref>; Pei et al., <xref rid="B42" ref-type="bibr">2011</xref>), but it is difficult to know if differences in the results are attributable to differences in the phylogeny employed or in the ecological processes themselves. We have shown here that constructing a molecular phylogeny for all communities together improves estimates of phylogenetic diversity and structure compared to estimating individual phylogenies for each community.</p></sec><sec><title>Phylogenetic diversity</title><p>A mega-phylogeny may also improve estimates of community phylogenetic diversity through the conversion of all phylogenies into molecular-clock-based ultrametric trees using the MPL adjustment (Britton et al., <xref rid="B4" ref-type="bibr">2002</xref>) and then directly estimating three commonly employed diversity metrics (Table <xref ref-type="table" rid="T3">3</xref>). Communities with the lowest species diversity showed the greatest contrast in diversity measures when estimated in the mega-phylogeny vs. the individual-plot phylogenies (Figure <xref ref-type="fig" rid="F4">4</xref>). For example, in the Yosemite and Wind-River plots (where species richness = 7), diversity estimates from individually-derived phylogenies were less than half that observed in the mega-phylogeny; whereas for the larger plots the differences were much less. For all communities, the values of PD were lower in individually-constructed community phylogenies (Figure <xref ref-type="fig" rid="F4">4</xref>). We note that this result considers only trees, and that work comparing canopy and understory diversity suggest that temperate forests may contain comparable phylogenetic diversity when all plants are considered (Halpern and Lutz, <xref rid="B16" ref-type="bibr">2013</xref>). However, for our observations, divergence between estimates were correlated with species richness of the plot (Species Richness vs. % difference in MPD = 0.68) with smaller plots showing the greatest differentiation, and suggests that the mega-phylogeny should greatly improve comparisons among plots, particularly when those communities differ in species richness.</p></sec><sec><title>Phylogenetic structure among communities</title><p>A growing, but still small, number of studies have compared phylogenetic structure across communities (Hardy et al., <xref rid="B17" ref-type="bibr">2012</xref>; Swenson, <xref rid="B51" ref-type="bibr">2012</xref>; Oliveira-Filho et al., <xref rid="B38" ref-type="bibr">2013a</xref>,<xref rid="B39" ref-type="bibr">b</xref>). However, as shown here the evolutionary structure among plots, via the inter-community measures of MPD and MNTD (Figure <xref ref-type="fig" rid="F5">5</xref>), complements similar patterns of phylogenetic structure within communities. The lack of differentiation among plots (Figure <xref ref-type="fig" rid="F5">5</xref>), with the exception of the extremely taxon-poor Yosemite and Wind-River plots in the Cascade and Sierra Nevada Mountains, is striking. The prevalence of trees in the families Fabaceae, Euphorbiaceae, and Myrtaceae in the tropical plots and their relative paucity in the plots located in temperate environments was not significant enough to differentiate these communities in most cases. The effect of latitude on measures of phylogenetic diversity was highly significant (with PD, MPD, and MNTD showing Spearman correlation coefficient of -0.905, 0.684, 0.521, respectively) and followed changes in species richness along the tropical to temperate transition. The correlation for PD was negative with latitude, whereas MPD and MNTD were positive, reflecting how the two latter metrics remove the effect of species richness on phylogenetic diversity. The reliance of MPD on the genetic distances of the most basal nodes of the phylogeny and the emphasis on the presence or absence of basal lineages suggest that substitution of one family (or order) in communities that differ in species number are equivalent. It is even more striking that the inter-community estimates of MNTD should show similarly low rates of differentiation among sites. While the differentiation in MPD can be more readily explained by the role of deeper nodes in determining differentiation, the MNTD would be inflated when comparing environments from the tropics with that of the temperate zones. The lack of differentiation among plots corresponds well to the observation that trees in these plots are in general phylogenetically clustered, and that environmental filtering is driving assembly processes. The main caveat is that we can infer a role of environmental filtering from phylogenetic clustering only when the traits that drive fitness are evolutionarily conserved.</p></sec><sec><title>Phylogenetic distance and ecological processes</title><p>A central benefit of constructing a mega-phylogeny containing many communities is our ability to more accurately contrast ecological processes operating in different communities. Therefore, phylogenetic patterns that are observed (e.g., clustering, overdispersion) are not attributable to differences in how community phylogeny are assembled, but are more directly linked to different ecological processes in those communities. We note that disentangling these processes within a community phylogenetic context remains a challenge, as we are just beginning to apply phylogenetic information to multiple communities and appropriate null models of phylogenetic pattern that incorporate explicit geographic differentiation are still being developed. The role of dispersal limitation and biogeographic vicariance in generating differences in species composition observed in different communities affect our results as would community assembly processes within sites. Yet the patterns derived with existing models can at least be viewed as having an ecological or evolutionary basis rather than a simple product of phylogeny construction.</p><p>In our study, for each of the different metrics of PD the most diverse tropical communities were composed of a set of more closely related species than expected at random in the context of the null model used (Table <xref ref-type="table" rid="T3">3</xref>). The pattern of increased relatedness was most evident for the nearest-taxon metric MNTD, which exhibited significant clustering for all but two plots, but was also true for MPD and PD for the tropical communities. This clustering of related species could be attributable to several factors. From the perspective of community ecology, these observations are consistent with local scale environmental filtering for phylogenetically conserved traits and niche conservatism. We note that with such geographically widespread communities other factors, including dispersal limitation linked with regional vicariance speciation, will play important roles and will require further investigation. Null models of no-dispersal limitation among communities will need to be explicitly re-examined in future work as we continue to construct phylogenies that encompass an increased number of communities.</p><p>With respect to environmental filtering and niche conservatism, these two processes are not mutually exclusive, although they make different assumptions regarding the role of phylogenetic conservatism and the role of dispersal. Much work has been done on the degree to which trait conservatism occurs in tropical forests (reviewed in Cavender-Bares et al., <xref rid="B6" ref-type="bibr">2009</xref>) and the role of trait conservatism on phylogenetic pattern (Kraft et al., <xref rid="B27" ref-type="bibr">2007</xref>; Crisp et al., <xref rid="B7" ref-type="bibr">2009</xref>). Kraft et al. (<xref rid="B26" ref-type="bibr">2011</xref>) demonstrated that increasing phylogenetic trait conservation will amplify phylogenetic structure, which results in communities composed of more closely related sets of species. Crisp et al. (<xref rid="B7" ref-type="bibr">2009</xref>) examined phylogenetic distribution across major South American biomes and found a high degree of constraint on the ability of related groups to invade novel biomes. These results are concordant with our observations of the tropical communities studied here, in which species in each community tended to be phylogenetically clustered. A growing number of studies (e.g., Hardy et al., <xref rid="B17" ref-type="bibr">2012</xref>; Ricklefs et al., <xref rid="B46" ref-type="bibr">2012</xref>) have found evidence for globally-scaled processes regulating species diversity in the tropics. For example, in the neotropics the number of individuals and the number of species in certain families is strongly conserved across five replicated forest plots (Ricklefs et al., <xref rid="B46" ref-type="bibr">2012</xref>). While the main objective of that particular study was an evaluation of the theory of ecological neutrality in community assembly (Hubbell, <xref rid="B21" ref-type="bibr">2001</xref>), the results are concordant with high levels of phylogenetic trait conservatism and environmental filtering (Kraft et al., <xref rid="B26" ref-type="bibr">2011</xref>). In some cases, field-based studies have shown mixed results in linking phylogenetic signal to trait dispersion in tropical forests (Liu et al., <xref rid="B32" ref-type="bibr">2013</xref>). Therefore, even though the current results are consistent with a global pattern of environmental filtering and niche conservatism as a driving force in community assembly, more work needs to be done to clarify the role of phylogenetic trait conservatism in large-scale community processes.</p></sec><sec><title>Conflict of interest statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec></sec>
Citric Acid-based Hydroxyapatite Composite Scaffolds Enhance Calvarial Regeneration	<p>Citric acid-based polymer/hydroxyapatite composites (CABP-HAs) are a novel class of biomimetic composites that have recently attracted significant attention in tissue engineering. The objective of this study was to compare the efficacy of using two different CABP-HAs, poly (1,8-octanediol citrate)-click-HA (POC-Click-HA) and crosslinked urethane-doped polyester-HA (CUPE-HA) as an alternative to autologous tissue grafts in the repair of skeletal defects. CABP-HA disc-shaped scaffolds (65 wt.-% HA with 70% porosity) were used as bare implants without the addition of growth factors or cells to renovate 4 mm diameter rat calvarial defects (n = 72, n = 18 per group). Defects were either left empty (negative control group), or treated with CUPE-HA scaffolds, POC-Click-HA scaffolds, or autologous bone grafts (AB group). Radiological and histological data showed a significant enhancement of osteogenesis in defects treated with CUPE-HA scaffolds when compared to POC-Click-HA scaffolds. Both, POC-Click-HA and CUPE-HA scaffolds, resulted in enhanced bone mineral density, trabecular thickness, and angiogenesis when compared to the control groups at 1, 3, and 6 months post-trauma. These results show the potential of CABP-HA bare implants as biocompatible, osteogenic, and off-shelf-available options in the repair of orthopedic defects.</p>	<contrib contrib-type="author"><name><surname>Sun</surname><given-names>Dawei</given-names></name><xref ref-type="aff" rid="a1">1</xref><xref ref-type="aff" rid="a2">2</xref><xref ref-type="author-notes" rid="n1"></xref></contrib><contrib contrib-type="author"><name><surname>Chen</surname><given-names>Yuhui</given-names></name><xref ref-type="aff" rid="a1">1</xref><xref ref-type="author-notes" rid="n1"></xref></contrib><contrib contrib-type="author"><name><surname>Tran</surname><given-names>Richard T.</given-names></name><xref ref-type="aff" rid="a3">3</xref><xref ref-type="author-notes" rid="n1">*</xref></contrib><contrib contrib-type="author"><name><surname>Xu</surname><given-names>Song</given-names></name><xref ref-type="aff" rid="a1">1</xref><xref ref-type="aff" rid="a4">4</xref></contrib><contrib contrib-type="author"><name><surname>Xie</surname><given-names>Denghui</given-names></name><xref ref-type="aff" rid="a1">1</xref><xref ref-type="aff" rid="a3">3</xref></contrib><contrib contrib-type="author"><name><surname>Jia</surname><given-names>Chunhong</given-names></name><xref ref-type="aff" rid="a4">4</xref></contrib><contrib contrib-type="author"><name><surname>Wang</surname><given-names>Yuchen</given-names></name><xref ref-type="aff" rid="a1">1</xref></contrib><contrib contrib-type="author"><name><surname>Guo</surname><given-names>Ying</given-names></name><xref ref-type="aff" rid="a1">1</xref></contrib><contrib contrib-type="author"><name><surname>Zhang</surname><given-names>Zhongmin</given-names></name><xref ref-type="aff" rid="a1">1</xref></contrib><contrib contrib-type="author"><name><surname>Guo</surname><given-names>Jinshan</given-names></name><xref ref-type="aff" rid="a3">3</xref></contrib><contrib contrib-type="author"><name><surname>Yang</surname><given-names>Jian</given-names></name><xref ref-type="corresp" rid="c2">b</xref><xref ref-type="aff" rid="a3">3</xref></contrib><contrib contrib-type="author"><name><surname>Jin</surname><given-names>Dadi</given-names></name><xref ref-type="corresp" rid="c3">c</xref><xref ref-type="aff" rid="a1">1</xref></contrib><contrib contrib-type="author"><name><surname>Bai</surname><given-names>Xiaochun</given-names></name><xref ref-type="corresp" rid="c1">a</xref><xref ref-type="aff" rid="a1">1</xref><xref ref-type="aff" rid="a4">4</xref></contrib><aff id="a1"><label>1</label><institution>Academy of Orthopedics, Guangdong Province, Department of Orthopedics, The Third Affiliated Hospital, Southern Medical University</institution>, Guangzhou, 510665, <country>China</country></aff><aff id="a2"><label>2</label><institution>Department of Orthopaedics & Microsurgery, Guangdong No. 2 Provincial People's Hospital</institution>, Guangzhou, 510317, <country>China</country></aff><aff id="a3"><label>3</label><institution>Department of Biomedical Engineering, Materials Research Institute, The Huck Institutes of The Life Sciences, The Pennsylvania State University, University Park</institution>, Pennsylvania, 16802, <country>U.S.A</country></aff><aff id="a4"><label>4</label><institution>State Key Laboratory of Organ Failure Research, Department of Cell Biology, School of Basic Medical Science, Southern Medical University</institution>, Guangzhou, 510515, <country>China</country></aff>	Scientific Reports	<p>Over 2.2 million bone transplantation procedures are performed annually worldwide in a variety of fields including orthopedics, neurosurgery, and dentistry<xref ref-type="bibr" rid="b1">1</xref>. Although, autologous bone grafts remain the gold standard for bone grafting procedures due to their superior osteogenic potential, their use is associated with various complications such as hematoma, soft tissue breakdown, pain, and prolonged recovery times<xref ref-type="bibr" rid="b2">2</xref><xref ref-type="bibr" rid="b3">3</xref>. Moreover, the use of bone autografts is contraindicated in osteoporotic populations due to a significant reduction in bone quality and quantity<xref ref-type="bibr" rid="b4">4</xref>. Thus, the development of a fully synthetic, readily available, and osteogenic bone substitute as an adjunct to autologous tissue grafts is strongly encouraged and considered as a great milestone in the clinical field.</p><p>For synthetic orthopedic biomaterials, research in the field has witnessed a shift from the use of permanent, inert metals towards tissue-engineered biodegradable composites designed to mimic the native composition of bone. Initially, the majority of synthetic orthopedic materials were based off calcium phosphates (CaPs), such as hydroxyapatite (HA) and beta tricalcium phosphate (TCP), because of their ability to replicate the native mineral constituent of bone tissue<xref ref-type="bibr" rid="b5">5</xref><xref ref-type="bibr" rid="b6">6</xref><xref ref-type="bibr" rid="b7">7</xref><xref ref-type="bibr" rid="b8">8</xref><xref ref-type="bibr" rid="b9">9</xref>. Although biomimetic and osteogenic, their applications are severely limited when fabricated into porous structures due to the inherent brittleness and very slow degradation rates<xref ref-type="bibr" rid="b10">10</xref><xref ref-type="bibr" rid="b11">11</xref><xref ref-type="bibr" rid="b12">12</xref>. To improve their utility, the hybridization of bioceramics and biodegradable polymers has been widely adopted to reform the mechanical properties and bioactivity of the resulting materials for orthopedic applications<xref ref-type="bibr" rid="b13">13</xref>. However, the current composite materials still suffer from several significant problems such as unsatisfactory mechanical strength, inefficient bone regeneration, poor bone integration, and the inability to mimic native bone chemical composition, which is composed of 60–65 wt.-% hydroxyapatite embedded in a collagen matrix<xref ref-type="bibr" rid="b14">14</xref><xref ref-type="bibr" rid="b15">15</xref>.</p><p>To address these limitations, our lab has focused on the development of citric acid-based materials to composite with bioceramics for orthopedic tissue engineering. Citrate, a naturally occurring Kreb's cycle product, is highly conserved in native bone with over 90% of the body's total citrate content being located in the skeletal system. Recent research has suggested that citrate plays significant roles in bone anatomy, physiology, and orthopedic biomaterial development<xref ref-type="bibr" rid="b16">16</xref><xref ref-type="bibr" rid="b17">17</xref><xref ref-type="bibr" rid="b18">18</xref>. For example, citrate is not only a dissolved calcium-solubilizing agent, but has recently been found to be an integral part of the bone nanocomposite. Previous research by Hu <italic>et al.</italic> and Davies <italic>et al.</italic> has shown that citrate molecules are strongly studded to the apatite nanocrystal surface and form bridges between mineral platelets regulating bone mineral crystallinity, respectively, which is highly related to the overall strength of bone tissue<xref ref-type="bibr" rid="b16">16</xref><xref ref-type="bibr" rid="b19">19</xref>. Along with the recent exciting and significant strides that have been made elucidating the role of citrate in bone formation and physiology, our lab has recently shown the potential of citrate as a cornerstone in orthopedic biomaterial design. Our recent exciting results showed that exogenous citrate, whether presented on a biomaterial, supplemented into culture media, or released from a biomaterial over the course of degradation can enhance alkaline phosphatase (ALP) and osterix (OSX) gene expression, osteoblast phenotype progression, implant osteoinductivity, and osteointegration both <italic>in vitro</italic> and <italic>in vivo</italic><xref ref-type="bibr" rid="b17">17</xref>. The natural abundance of citrate in bone tissue, its importance in bone physiology, and our recent findings on stem cell culture hint that citrate may have significant impacts on bone development and orthopedic biomaterial development<xref ref-type="bibr" rid="b19">19</xref><xref ref-type="bibr" rid="b20">20</xref>.</p><p>As previously mentioned, citric acid-based polymer/hydroxyapatite composites (CABP-HAs) are a novel class of orthopedic biomaterial, which offer distinct advantages for bone tissue engineering applications<xref ref-type="bibr" rid="b17">17</xref>. The first reported citric acid-based composite, poly (1,8-octanediol)-HA (POC-HA), is synthesized with non-toxic monomers using simple and cost-effective procedures. POC-HA displayed enhanced HA integration through the calcium chelating ability of free carboxylic chemistry in the bulk of the material<xref ref-type="bibr" rid="b20">20</xref>. The abundant carboxyl chemistry allowed for the incorporation of up to 65 wt.-% of HA in the composites to match the native mineral content of bone tissue and is a feature that is not possible with previous materials. Since the development of POC-HA, our laboratory has long been working on a series of citrate-based biodegradable polymers such as mechanically strong crosslinked urethane-doped polyesters (CUPEs)<xref ref-type="bibr" rid="b21">21</xref>, biodegradable photoluminescent polymers (BPLPs)<xref ref-type="bibr" rid="b22">22</xref><xref ref-type="bibr" rid="b23">23</xref>, dual-crosslinkable poly (alkylene maleate citrate) (PAMC)<xref ref-type="bibr" rid="b24">24</xref><xref ref-type="bibr" rid="b25">25</xref>, and clickable POC-based elastomers (POC-Click)<xref ref-type="bibr" rid="b26">26</xref> for various biomedical applications such as bone, vascular, and neural tissue engineering, cancer imaging, and drug delivery<xref ref-type="bibr" rid="b17">17</xref><xref ref-type="bibr" rid="b27">27</xref><xref ref-type="bibr" rid="b28">28</xref><xref ref-type="bibr" rid="b29">29</xref>. To improve upon the mechanical properties of POC-HA through urethane chemistry, we have shown that CUPE-HA composites display impressive compressive strengths of 116.23 ± 65.37 MPa, which falls within the range of native human cortical bone (100–230 MPa). Although, CUPE-HA showed minimal chronic inflammation and full osteointegration when implanted in a rabbit lateral femoral condyle defect model, the urethane-doping strategy sacrificed valuable pendant carboxyl chemistry to chelate with hydroxyapatite limiting the mechanical potential of the material<xref ref-type="bibr" rid="b17">17</xref><xref ref-type="bibr" rid="b21">21</xref><xref ref-type="bibr" rid="b24">24</xref>.</p><p>To address this situation, we have recently developed a clickable biodegradable elastomer, poly (octanediol citrate) – click (POC-Click), which employs azide-alkyne cycloaddition (click chemistry) as an additional crosslinking mechanism to improve the mechanical strength of the bulk material without sacrificing valuable pendant citric acid carboxyl chemistry for HA calcium chelation<xref ref-type="bibr" rid="b26">26</xref>. Although we have previously shown that citrate-based materials can address the challenges in designing mechanically strong, osteoconductive, and osteoinductive orthopedic biomaterials, this new class of orthopedic biomaterials has only been studied as solid, non-porous implants, and their use as a scaffold for cranial bone defect repair has not been studied yet. Therefore, in this study, CUPE-HA and POC-Click-HA porous scaffolds were fabricated and compared for their potential to repair cranial defects in rats.</p><sec disp-level="1" sec-type="methods"><title>Methods</title><p>Hydroxyapatite [Mw: 502.32, assay > 90% (as Ca<sub>3</sub>(PO<sub>4</sub>)<sub>2</sub>); particle size: > 75 µm (0.5%), 45–75 µm (1.4%), < 45 µm (98.1%)] and all remaining chemicals were purchased from Sigma-Aldrich (St. Louis, MO, USA) and used as received unless stated otherwise.</p><sec disp-level="2"><title>Scaffold fabrication</title><p>CUPE pre-polymers and POC-Click (clickable diols 30% of total diol composition) pre-polymers were synthesized according to previously published methods<xref ref-type="bibr" rid="b21">21</xref><xref ref-type="bibr" rid="b26">26</xref>. To fabricate porous CUPE-HA disc shaped scaffolds, CUPE pre-polymer was first dissolved in 1,4-dioxane and mixed with hydroxyapatite (65 wt.-%). Next, sodium chloride salt with an average size in the range of 200–400 μm was added to the mixture (70 wt.-%) and stirred in a Teflon dish until a homogenous viscous paste was formed. To fabricate disc-shaped scaffolds, the viscous paste was inserted into Teflon tubes (4 × 2 mm; inner diameter × height) purchased from McMaster-Carr (Aurora, OH, USA). Following solvent evaporation, the scaffolds were post-polymerized in an oven maintained at 100°C for 3 days. Salt was leached out from the scaffolds by immersion in deionized water under vacuum for 72 hours with water changes every 12 hours. Finally, the scaffolds were dried using lyophilization to obtain the final disc-shaped scaffolds (4 × 2 mm; diameter × height). POC-Click-HA disc shaped scaffolds were also fabricated as described above.</p></sec><sec disp-level="2"><title>Ethics Statement</title><p>All animal experiments were carried out with the approval of the Southern Medical University Animal Care and Use Committee in accordance with the guidelines for the ethical treatment of animals. All surgeries were performed under chloral hydrate and xylazine anesthesia, and all efforts were made to minimize animal suffering.</p></sec><sec disp-level="2"><title>Surgical procedure</title><p>Seventy-two male Sprague–Dawley rats (200–220 g) were used for the animal experiments. The body weights were closely monitored to confirm feeding and expected growth rates. All rats were handled regularly for at least one week prior to surgery and individually housed in cages in climate controlled rooms at 22°C with 50% humidity and 12 h light/dark cycles. Surgeries were performed under semi-sterile conditions with animals under anesthesia induced by the intraperitoneal injection of 100 mg/kg chloral hydrate and 10 mg/kg xylazine. The surgical site was shaved and prepared with a 70% ethanol solution, and a subcutaneous injection of 0.5 mL of 1% lidocaine (local anesthetic) was given at the sagittal midline of the skull. Following this injection, a sagittal incision was made over the scalp from the nasal bone to the middle sagittal crest, and the periosteum was bluntly dissected. Using a punch, a 4 mm diameter pit defect was made with a trephine drill, which was constantly cooled with sterile saline. The calvarial disk was then carefully removed to avoid tearing of the dura. Animals were randomly assigned into four groups (n = 18 per group) in which the 4 mm defect was: 1) left empty as a negative control (CON), 2) filled with autogenous bone (AB), 3) filled with CUPE-HA scaffolds (CUPE-HA), or 4) filled with POC-Click-HA scaffolds (POC-Click-HA). The skin was sutured with 6-0 vicryl, and the animals were monitored following standard post-operative animal care protocols (<xref ref-type="supplementary-material" rid="s1">Fig. S1</xref>). Animals were ultimately anesthetized and sacrificed by xylazine injection at 1, 3, 6 months post-surgery.</p></sec><sec disp-level="2"><title>Scanning electron microscope</title><p>Bare CABP-HA scaffolds and explanted scaffolds isolated 6 months after surgery were fixed with 2.5% glutaraldehyde (Sigma, USA) for 12 h and characterized using scanning electron microscopy (SEM). After tension-free drying, the samples were coated in gold and analyzed with a S-3000N SEM (Hitachi, Japan) under high-vacuum conditions (air pressure<1 × 10<sup>−7</sup> mbar), 20 kV voltage, and 18 mm working distance. A Philips XL 30 ESEM-FEG system (FEI Co, USA) was used to capture images at 50× and 500× magnification.</p></sec><sec disp-level="2"><title>Microcomputed tomography (micro CT)</title><p>Micro CT analysis was used to quantify the volume of bone formation within the defect. The tomography of fixed rat calvarial bone specimens was performed using a microtomographic imaging system (ZKKS-MCT-Sharp-III scanner, Caskaishen, China) at 50 kV, 40 W, and an exposure time of 300 ms through a 0.5-mm-thick aluminum filter. The X-ray projections were obtained at 0.72° intervals with a scanning angular rotation of 360°. The reconstructed dataset was segmented by an automated thresholding algorithm. The projection images were reconstructed into three-dimensional images using ZKKS-MicroCT software (version 4.1) from ZKKS. To calculate the bone mineral density (BMD) and trabecular thickness (Tb.Th) among the four groups, a hollow cylindrical volume of interest (VOI-1) 4 mm in external diameter, 2.8 mm in inner diameter, and 300 μm in height was selected for scanning and corrected for the CT value to evaluate the bone regeneration at the periphery of the implant areas (<xref ref-type="fig" rid="f1">Fig. 1A</xref>). To further evaluate bone regeneration between the CUPE-HA and POC-Click-HA groups in the entire defect areas, a cylindrical volume of interest (VOI-2) 4 mm in diameter and 300 μm in height was selected to analyze the BMD and Tb.Th (<xref ref-type="fig" rid="f1">Fig. 1B</xref>).</p></sec><sec disp-level="2"><title>Histological assessment</title><p>For histological assessment, the defect and surrounding tissues were fixed in 4% paraformaldehyde for 48 h, decalcified with 0.5 M ethylenediaminetetraacetic acid (EDTA) at pH 7.4 for 3 weeks, and embedded in paraffin. Tissues were longitudinally sectioned with a 4-μm thickness, deparaffinated with xylene, gradually hydrated, and stained with hematoxylin and eosin (H&E) for light microscopic analysis. Images were captured at 200× magnification using an Olympus bx51 microscope with a 20× objective lens (Olympus, Japan) and a digital camera (ProgRes C14, Jenoptik, Germany). Five random images were obtained for each rat along the length of the defect. Overall, a total of 25 to 30 images were acquired from five to six rats in each group. The number of blood vessels was determined by counting luminal structures lined by vascular endothelium and partially filled with red blood cells<xref ref-type="bibr" rid="b30">30</xref>. The deparaffinated slices were stained using Alkaline Phosphatase Kit based on naphthol AS-MX phosphate and fast blue RR salt (Sigma, MO, USA). Images were observed at 400× magnification.</p></sec><sec disp-level="2"><title>Immunohistochemical analysis</title><p>To stain for osteocalcin (OCN) and Vascular Endothelial Growth Factor-b (VEGF-b), tissue sections were treated with proteinase K (Sigma, MO, USA) for antigen recovery, washed with PBS, blocked with 5% bovine serum albumin at room temperature for 30 min, and exposed to antibodies for VEGF-b (1:100 dilution, Bioworld, CA, USA) and osteocalcin (1:100 dilution, Santa Cruz, CA, USA) overnight at 4°C. Peroxidase activity was detected using the enzyme substrate 3-amino-9-ethyl carbazole. For negative controls, sections were treated in an identical manner, except they were incubated in PBS-buffered saline without primary antibodies. Vascular numbers were determined by three single blind pathologists and calculated according to the average number of vessels in five random areas of an image at 200× magnification.</p></sec><sec disp-level="2"><title>Statistical analysis</title><p>Data are expressed as the mean ± standard deviation. The statistical significance between two sets of data was calculated using a two-tail Student's t-test. Analysis of variance (ANOVA) with Newman-Keuls multiple comparisons test post-hoc analysis was used to determine significant differences among three or more groups. Data analysis was performed using SPSS 20.0 software. Data were considered to be significant when a p-value of 0.05 or less was obtained.</p></sec></sec><sec disp-level="1" sec-type="results"><title>Results</title><sec disp-level="2"><title>SEM analysis of CABP-HA scaffolds</title><p>SEM images of the disc-shaped CUPE-HA and POC-Click-HA scaffolds fabricated in this study are shown in <xref ref-type="fig" rid="f2">Figure 2</xref>. As shown in the SEM images (<xref ref-type="fig" rid="f2">Fig 2A-B</xref>), a porous structure with an average pore size in the range of 200–400 μm can be seen. The SEM images suggest that the scaffolds present rough pore walls, which may facilitate protein adsorption and cell adhesion. Explanted samples after 6 months of implantation were also viewed under SEM (<xref ref-type="fig" rid="f2">Fig. 2C–D</xref>). As seen from the two SEM images, the overall morphology between the explanted CUPE-HA and POC-Click-HA groups were similar in appearance with dense tissue filling the scaffold pore structures.</p></sec><sec disp-level="2"><title>Radiographic assessment of new bone formation</title><p>To observe new bone formation within the defects, 3D images of the defects were reconstructed using micro-CT 1, 3, and 6 months post-surgery (<xref ref-type="fig" rid="f3">Fig. 3</xref>). Radiographic evidence of new bone formation was highly variable between non-scaffold groups (CON and AB groups) and CABP-HA scaffold groups (CUPE-HA and POC-Click-HA groups). In the untreated defects (CON group), no new bone formation was observed in the defect 6 months after surgery (<xref ref-type="fig" rid="f3">Figs. 3A–C</xref>). In contrast, dense tissue was evident inside the defect of autologous bone treated animals (<xref ref-type="fig" rid="f3">Figs. 3D–F</xref>). In two scaffold groups, radiopaque tissue was found spread over the entire defect site indicating new bone formation. The amount of radiopaque tissue appeared higher in the defects treated with CUPE-HA scaffolds (<xref ref-type="fig" rid="f3">Figs. 3G–I</xref>) compared to POC-Click-HA scaffolds (<xref ref-type="fig" rid="f3">Figs. 3J–L</xref>).</p><p>Next, micro-CT analysis was conducted to quantify the mineralized skeletal bone formation at the periphery of the implant using a hollow cylindrical volume of interest (VOI-1) (<xref ref-type="fig" rid="f4">Fig. 4A and 4B</xref>). The highest level of peripheral bone formation was observed in defects treated with AB compared to all other groups. However, compared to the negative control (CON), the defects repaired with CABP-HA scaffolds exhibited greater peripheral bone regeneration. The peripheral BMD and Tb.Th of the AB group was significantly higher than that of the other groups at 1, 3, and 6 months after surgery (p<0.001), which is evidenced by the radiopaque tissue bridging the implant and surrounding bone. The peripheral BMD and Tb.Th of the CUPE-HA and POC-Click-HA groups was significantly higher than the CON group at all time points (<xref ref-type="fig" rid="f4">Fig. 4A and B</xref>) (p<0.01). The peripheral BMD and Tb.Th observed between the CUPE-HA group and the POC-Click-HA groups showed no significant difference at all time points (p>0.05). To quantify the amount of mineralized tissue within the scaffold treated defect sites, the BMD and Tb.Th was calculated using a solid cylindrical VOI-2 (<xref ref-type="fig" rid="f1">Fig. 1B</xref>). As shown in <xref ref-type="fig" rid="f4">Figures 4C and D</xref>, no significant difference was seen at the 1 month time point for both BMD and Tb.Th (<xref ref-type="fig" rid="f4">Fig. 4C and 4D</xref>, respectively). However, the CUPE-HA treated cranial defects exhibited increased dense tissue formation within the defect sites when compared to the POC-Click-HA group with the higher BMD (p<0.05, p<0.01) (<xref ref-type="fig" rid="f4">Fig. 4C</xref>) and Tb.Th (*p<0.01) (<xref ref-type="fig" rid="f4">Fig. 4D</xref>) at 3 and 6 months post-implantation.</p></sec><sec disp-level="2"><title>Histological assessment</title><p>Histological assessment of defect sites in the CUPE-HA and POC-Click-HA groups were similar to the findings in the autologous bone treated group (AB group). Specifically, the edge of the defect site was composed of fibrous stroma and reactive bone. The fibrous stroma appeared loose around the scaffolds and exhibited a relatively high level of angiogenesis. In contrast, the defect sites in the negative control group (CON group) did not exhibit reactive bone formation and displayed less fibrous stroma compared to both the AB and CABP-HA scaffold groups (<xref ref-type="fig" rid="f5">Fig. 5</xref>). Tissue sections were stained for alkaline phosphatase (ALP) to indicate the presence of osteoblast within the defect site 1, 3, and 6 months post implantation (<xref ref-type="fig" rid="f6">Fig. 6A</xref>). Staining for osteocalcin (OCN) was also performed and revealed OCN positive cells in the defect site and were distributed within the implanted scaffolds 1, 3, and 6 months post implantation (<xref ref-type="fig" rid="f6">Fig. 6B</xref>). ALP and OCN positive cells were also quantified in the defect site of AB and CON control group (<xref ref-type="supplementary-material" rid="s1">Fig. S2ab</xref>). Immunohistochemical staining for VEGF-b was also performed after 1, 3, and 6 months of implantation (<xref ref-type="fig" rid="f7">Fig. 7</xref>). In the negative control group (without any material implantation), very few fibroblasts and inflammatory cells were concentrated in the defect space and lower levels of VEGF-b expression were observed when compared to the same area of other groups. In contrast, CUPE-HA and POC-Click-HA treated groups showed higher levels of VEGF-b expression and was similar to that of AB group, especially at the 1 month time point (<xref ref-type="fig" rid="f7">Fig. 7</xref>). Quantification of the vessel numbers found in defect sites after 1, 3, and 6 months for all experimental groups are shown in <xref ref-type="fig" rid="f8">Figure 8</xref>. Greater vessel numbers in CABP-HA scaffold treated defects were confirmed by morphometric analysis of the vessels (<xref ref-type="fig" rid="f8">Fig. 8</xref>). The vascular numbers of CUPE-HA groups were significantly higher than the AB and POC-Click-HA groups 1 month post-surgery (p<0.05) (<xref ref-type="fig" rid="f8">Fig. 8</xref>). No significant difference was observed in vascular numbers between each group at 3 and 6 months post-surgery (p<0.05).</p></sec></sec><sec disp-level="1" sec-type="discussion"><title>Discussion</title><p>In the early 1960s, it was found that citrate made up about 5 wt.-% of the organic component in bone. Recent studies have found that the surface of apatite nanocrystals is actually strongly bound with citrate molecules, which regulate and stabilize the crystallinity of apatite nanocrystals by bridging mineral platelets<xref ref-type="bibr" rid="b16">16</xref><xref ref-type="bibr" rid="b19">19</xref><xref ref-type="bibr" rid="b31">31</xref>. Our recent reports on citrate-based polymer blend composites have shown that citrate, whether presented on the biomaterial or exogenously supplemented into culture media, can promote bone development through the up regulation of alkaline phosphatase (ALP) and osterix (OSX) gene expression and acceleration of osteoblast phenotype progression<xref ref-type="bibr" rid="b17">17</xref>. These recent insights have reintroduced interest into the role of citrate in bone development and provided a new hypothesis that osteoblasts were specialized citrate producing cells providing the increased citrate levels necessary for proper bone formation<xref ref-type="bibr" rid="b18">18</xref>. It is from this rationale that citrate-based materials may enhance bone formation through the presentation of citrate chemistry on the material surface and the release of citrate during the course of material degradation. Our previous studies have shown that citrate-based materials are beneficial for orthopedic applications in that they inherently provide increased citrate levels to the newly forming tissue through material degradation, offer a wide range of controllable material properties, and the ability to be composited with up to 65 wt.-% HA through enhanced bioceramic-polymer interactions, which is a feature not possible with previous orthopedic biomaterials<xref ref-type="bibr" rid="b17">17</xref><xref ref-type="bibr" rid="b20">20</xref><xref ref-type="bibr" rid="b22">22</xref><xref ref-type="bibr" rid="b32">32</xref>.</p><p>Although, the functional bio-integration of CABP-HA solid implants has been previously demonstrated to show excellent osteointegration due to their bone-mimicking compositions <italic>in vitro</italic> and <italic>in vivo</italic>, the objective of this study was to compare the efficacy of using bare citric acid-based polymer composite scaffolds to stimulate bone tissue regeneration in a rat calvarial defect model without the use of growth factor supplementation or prior <italic>in vitro</italic> cell seeding<xref ref-type="bibr" rid="b16">16</xref><xref ref-type="bibr" rid="b17">17</xref><xref ref-type="bibr" rid="b20">20</xref><xref ref-type="bibr" rid="b33">33</xref>. The use of a cranial defect as a means to evaluate bone graft substitutes dates back to more than 100 years when demineralized bovine bone matrix were implanted into the skull of canines<xref ref-type="bibr" rid="b34">34</xref>. It was later in 1986 that Schmitz and Hollinger established a rat cranial prototype model for osseous nonunions and discontinuity defects<xref ref-type="bibr" rid="b35">35</xref>. Their results showed that although bone is a dynamic tissue undergoing constant remodeling, its innate ability to heal itself upon injury is largely dependent upon the defect size and location<xref ref-type="bibr" rid="b35">35</xref><xref ref-type="bibr" rid="b36">36</xref>. Circular defects created on the rat cranium failed to heal and were filled with fibrous tissue instead of new bone formation over a 3-month period<xref ref-type="bibr" rid="b35">35</xref><xref ref-type="bibr" rid="b36">36</xref><xref ref-type="bibr" rid="b37">37</xref>. The reduced healing response of the skull was attributed poor blood supply and relative deficiency of bone marrow when compared to other bones<xref ref-type="bibr" rid="b35">35</xref><xref ref-type="bibr" rid="b38">38</xref><xref ref-type="bibr" rid="b39">39</xref>. Unlike typical long bones, there is no primary nutrient artery present rendering the calvarium biologically inert<xref ref-type="bibr" rid="b40">40</xref>. Instead, blood is primarily supplied by the middle meningeal artery with secondary sources of perfusion being accessory arterioles<xref ref-type="bibr" rid="b41">41</xref>. These reasons make it especially difficult for even small defects in the skull to spontaneously repair causing many to consider the cranial bone defect as the most severe bone implant test<xref ref-type="bibr" rid="b39">39</xref>.</p><p>Although poor blood supply is one of the major reasons for slow bone regeneration in cranial defects, our results show that CABP-HA materials have promoted bone regeneration over time without the use of growth factors or prior cell seeding. Our citric acid based bone substitutes were compared with autogenous bone grafts, which were considered as the gold standard for bone regeneration and osteointegration in clinic due to its inherent ability to provide the ideal biocompatibility and osteogenicity<xref ref-type="bibr" rid="b1">1</xref><xref ref-type="bibr" rid="b8">8</xref><xref ref-type="bibr" rid="b42">42</xref>. Although BMD and Tb.Th values for CABP-HA scaffolds were lower than that of autogenous bone grafts, fibrous-osteotylus formation, a vital marker of intramembranous ossification, was markedly observed in both materials group showing the osteoconductivity of citric acid-based bone substitutes<xref ref-type="bibr" rid="b43">43</xref>. Alkaline phosphatase (ALP) and osteocalcin (OCN) are considered compelling bio-markers of osteogenesis<xref ref-type="bibr" rid="b44">44</xref>, and our results show the presence ALP positive osteoblasts as well as OCN positive osteoblasts in the defect area of scaffold treated animals. Interestingly, CUPE-HA scaffolds displayed enhanced bone regeneration over POC-Click-HA treated defects in regards to BMD. The increased BMD values may be attributed to the degradation rates of the two materials as previous studies have shown that POC-Click degrades much more rapidly than CUPE-HA<xref ref-type="bibr" rid="b26">26</xref>. However, the ability of citrate-based materials to stimulate bone regeneration in cranial defects as bare implant materials shows the osteogenic potential of these novel biomaterials.</p><p>The lack of a stable and integrated vascular system has been determined to be a significant cause for graft failure<xref ref-type="bibr" rid="b45">45</xref>. To address this issue, a large number of studies on biomaterials have focused on the bio-graft, which consists of a scaffold, seeded cells, and the release of bio-factors to either induce angiogenesis or recruit host vasculature to impregnate the implanted graft. Vascular endothelial growth factor (VEGF) and bone morphogenetic protein (BMP) are the most common bio-factors applied in tissue engineering and can enhance the regeneration of bone and soft tissue through stimulating angiogenesis and proliferation of seeded cells<xref ref-type="bibr" rid="b46">46</xref>. To circumvent the long <italic>in vitro</italic> cell culture times and regulatory issues involved with the use of growth factors, recent research in the field has focused on the use of off-the-shelf available acellular systems in which the implanted biomaterial/graft alone can induce tissue regeneration, angiogenesis, and host-cell recruitment to the defect site<xref ref-type="bibr" rid="b47">47</xref><xref ref-type="bibr" rid="b48">48</xref>. Surprisingly, a stimulation of soft tissue regeneration and angiogenesis was revealed in animals treated with CABP-HA scaffolds in the early time points. In our study, increased vessel numbers and a higher level of VEGF expression suggest the development of a satisfactory blood supply during bone and soft tissue regeneration. As blood supply is vital for successful bone grafting, the improved angiogenesis in the defect area after CABP-HA scaffold transplantation indicates that blood supply demands can potentially be remedied through graft transplantation<xref ref-type="bibr" rid="b43">43</xref><xref ref-type="bibr" rid="b49">49</xref><xref ref-type="bibr" rid="b50">50</xref>. In this regard, the two kinds of CABP-HA scaffolds used in this study enhanced VEGF expression in defect area, which is a crucial factor in promoting soft tissue regeneration in the bone defects<xref ref-type="bibr" rid="b51">51</xref><xref ref-type="bibr" rid="b52">52</xref><xref ref-type="bibr" rid="b53">53</xref>. Although the mechanism of the biomaterial mediated angiogenesis is not well understood at this time, taken together, these results suggest the great potential of citric acid-based hydroxyapatite biomaterials for clinical use.</p></sec><sec disp-level="1" sec-type="conclusions"><title>Conclusions</title><p>In conclusion, highly porous disc-shaped citric acid-based polymer hydroxyapatite composite scaffolds were fabricated and used to repair rat calvarial defects. Without incorporating biological growth factors or pre-seeding with stem cells or osteoblastic cells, both POC-Click-HA and CUPE-HA scaffolds displayed satisfactory host responses and osteogenic potential through the stimulation of proximal bone formation and angiogenesis in the repair of calvarial defects. Therefore, citric acid-based polymer hydroxyapatite scaffolds could serve as a promising off-the-shelf implant for the regeneration of bone defects.</p></sec><sec disp-level="1"><title>Author Contributions</title><p>D.S., Y.C. and R.T.T. wrote the manuscript; R.T.T. and J.G. made the scaffolds; D.S., S.X. and D.X. conducted animal surgery; C.J., Y.W. and Y.G. prepared figures 1–4; D.S. and Z.Z. prepared figures 5–8; J.Y., D.J. and X.B. were responsible for the research and revised the manuscript. All authors reviewed the manuscript.</p></sec><sec sec-type="supplementary-material" id="s1"><title>Supplementary Material</title><supplementary-material id="d33e27" content-type="local-data"><caption><title>Supplementary Information</title><p>Digital photographs of cranial defect procedure and repair in rats and IHC stain for ALP/OCN of cranial defect area tissue sections from AB and CON groups.</p></caption><media xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="srep06912-s1.pdf"/></supplementary-material></sec>
Neonatal Magnesium Levels Correlate with Motor Outcomes in Premature Infants: A Long-Term Retrospective Cohort Study	<p><bold>Objective:</bold> Chronic neurological deficits are a significant complication of preterm birth. Magnesium supplementation has been suggested to have neuroprotective function in the developing brain. Our objective was to determine whether higher neonatal serum magnesium levels were associated with better long-term neurodevelopmental outcomes in very-low birth weight infants.</p><p><bold>Study Design:</bold> A retrospective cohort of 75 preterm infants (<1500 g, gestational age <27 weeks) had follow-up for the outcomes of abnormal motor exam and for epilepsy. Average total serum magnesium level in the neonate during the period of prematurity was the main independent variable assessed, tested using a Wilcoxon rank-sum test.</p><p><bold>Results:</bold> Higher average serum magnesium level was associated with a statistically significant decreased risk for abnormal motor exam (<italic>p</italic> = 0.037). A lower risk for epilepsy in the group with higher magnesium level did not reach statistical significance (<italic>p</italic> = 0.06).</p><p><bold>Conclusion:</bold> This study demonstrates a correlation between higher neonatal magnesium levels and decreased risk for long-term abnormal motor exam. Larger studies are needed to evaluate the hypothesis that higher neonatal magnesium levels can improve long-term neurodevelopmental outcomes.</p>	<contrib contrib-type="author"><name><surname>Doll</surname><given-names>Elizabeth</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref></contrib><contrib contrib-type="author"><name><surname>Wilkes</surname><given-names>Jacob</given-names></name><xref ref-type="aff" rid="aff2"><sup>2</sup></xref></contrib><contrib contrib-type="author"><name><surname>Cook</surname><given-names>Lawrence J.</given-names></name><xref ref-type="aff" rid="aff3"><sup>3</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/190587"/></contrib><contrib contrib-type="author"><name><surname>Korgenski</surname><given-names>E. Kent</given-names></name><xref ref-type="aff" rid="aff2"><sup>2</sup></xref></contrib><contrib contrib-type="author"><name><surname>Faix</surname><given-names>Roger G.</given-names></name><xref ref-type="aff" rid="aff4"><sup>4</sup></xref></contrib><contrib contrib-type="author"><name><surname>Yoder</surname><given-names>Bradley A.</given-names></name><xref ref-type="aff" rid="aff4"><sup>4</sup></xref></contrib><contrib contrib-type="author"><name><surname>Srivastava</surname><given-names>Rajendu</given-names></name><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><xref ref-type="aff" rid="aff5"><sup>5</sup></xref></contrib><contrib contrib-type="author"><name><surname>Sherwin</surname><given-names>Catherine M. T.</given-names></name><xref ref-type="aff" rid="aff6"><sup>6</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/154891"/></contrib><contrib contrib-type="author"><name><surname>Spigarelli</surname><given-names>Michael G.</given-names></name><xref ref-type="aff" rid="aff6"><sup>6</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/102627"/></contrib><contrib contrib-type="author"><name><surname>Clark</surname><given-names>Erin A. S.</given-names></name><xref ref-type="aff" rid="aff7"><sup>7</sup></xref></contrib><contrib contrib-type="author"><name><surname>Bonkowsky</surname><given-names>Joshua L.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="corresp" rid="cor1">*</xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/61025"/></contrib>	Frontiers in Pediatrics	<sec sec-type="introduction" id="S1"><title>Introduction</title><p>Preterm birth can lead to a wide range of motor and intellectual disabilities affecting up to 35% of survivors (<xref rid="B1" ref-type="bibr">1</xref>–<xref rid="B4" ref-type="bibr">4</xref>). Very-low birth weight (VLBW) infants with birth weight less than 1500 g have elevated rates of cerebral palsy, epilepsy, autism, intellectual disability, and behavioral problems (<xref rid="B5" ref-type="bibr">5</xref>, <xref rid="B6" ref-type="bibr">6</xref>). While survival rates have improved dramatically for premature infants (<xref rid="B7" ref-type="bibr">7</xref>), neurodevelopmental outcomes have not (<xref rid="B8" ref-type="bibr">8</xref>). Despite efforts to reduce preterm birth, the rate has remained relatively stable over the last few decades and was 11.7% in the United States in 2011 (<xref rid="B9" ref-type="bibr">9</xref>).</p><p>The neurodevelopmental problems in prematurely born infants are caused by a variety of complex pathophysiological mechanisms (<xref rid="B10" ref-type="bibr">10</xref>, <xref rid="B11" ref-type="bibr">11</xref>) with few therapeutic options (<xref rid="B12" ref-type="bibr">12</xref>). Further, the complications of preterm birth are now also recognized to damage both gray matter and axon tracts and to lead to impaired neurodevelopment (<xref rid="B13" ref-type="bibr">13</xref>–<xref rid="B17" ref-type="bibr">17</xref>).</p><p>Magnesium sulfate administered antenatally has been found to reduce rates of cerebral palsy when given prior to preterm birth (<xref rid="B18" ref-type="bibr">18</xref>–<xref rid="B20" ref-type="bibr">20</xref>). Magnesium has also been found in small studies to improve neurodevelopmental outcomes in term infants with birth asphyxia (<xref rid="B21" ref-type="bibr">21</xref>–<xref rid="B24" ref-type="bibr">24</xref>). Further, several animal model studies suggest that magnesium could play neuroprotective roles in the developing vertebrate CNS (<xref rid="B25" ref-type="bibr">25</xref>–<xref rid="B27" ref-type="bibr">27</xref>).</p><p>Our hypothesis was that postnatal magnesium could serve a neuroprotective role in the developing premature brain. Magnetic resonance imaging (MRI) data suggest that longer exposure of premature infants to the extrauterine environment results in increasing impairments of CNS connectivity (<xref rid="B28" ref-type="bibr">28</xref>). We posited that magnesium could help protect connectivity development of premature infants, and that higher magnesium levels throughout the premature period could be neuroprotective.</p><p>To address this hypothesis, we evaluated serum magnesium levels in VLBW infants and compared these with long-term neurodevelopmental outcomes, specifically epilepsy, and abnormal motor exam. We recorded magnesium levels during the initial hospitalization, for the time period when the infants were still premature (less than 37 weeks gestation). The objective of our study was to determine whether higher neonatal magnesium levels are associated with improved long-term neurological outcomes.</p></sec><sec sec-type="materials\|methods" id="S2"><title>Materials and Methods</title><sec id="S2-1"><title>Ethics statement</title><p>This study was approved by the Institutional Review Boards at the University of Utah and Intermountain Healthcare (IH). Data were anonymously collected and analyzed with no identifying information, and a waiver of informed consent was obtained.</p></sec><sec id="S2-2"><title>Study design and data extraction</title><p>Data extraction and analysis were performed retrospectively in a cohort of premature infants born at an IH hospital and who were seen in follow-up in the Utah State Department of Health Neonatal Follow-up Program. The cohort consisted of a consecutive series of 107 infants born between 1/1/06 through 12/31/10, with birth weights <1500 g and up to 26<sup>6/7</sup> weeks gestational age (Table <xref ref-type="table" rid="T1">1</xref>). Five patients were lost to follow-up; 27 patients had incomplete or missing data and were not included in analysis (Table <xref ref-type="table" rid="T2">2</xref>). Serum magnesium levels were those drawn during the initial hospitalization up through the end of the premature period; defined as less than 37 weeks gestation. IH is a large, vertically integrated not-for-profit health care system in the Intermountain West encompassing 23 hospitals including the single children’s hospital. Antenatal, perinatal, and follow-up data were extracted for each patient in the cohort from the Enterprise Data Warehouse (EDW) maintained by IH.</p><table-wrap id="T1" position="float"><label>Table 1</label><caption><p><bold>Demographic characteristics of the study group</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="left" rowspan="1" colspan="1">Characteristic</th><th align="center" rowspan="1" colspan="1">Study cohort <italic>n</italic> (%)</th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">Gender (male)</td><td align="left" rowspan="1" colspan="1">44 (59%)</td></tr><tr><td align="left" colspan="1" rowspan="1">Ethnicity</td></tr><tr><td align="left" rowspan="1" colspan="1"> Caucasian</td><td align="left" rowspan="1" colspan="1">51 (68%)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Hispanic</td><td align="left" rowspan="1" colspan="1">8 (11%)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Pacific-Islander</td><td align="left" rowspan="1" colspan="1">2 (3%)</td></tr><tr><td align="left" rowspan="1" colspan="1"> African-American</td><td align="left" rowspan="1" colspan="1">2 (3%)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Native American</td><td align="left" rowspan="1" colspan="1">1 (1%)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Asian</td><td align="left" rowspan="1" colspan="1">3 (4%)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Unknown</td><td align="left" rowspan="1" colspan="1">8 (11%)</td></tr><tr><td align="left" rowspan="1" colspan="1">Multiple gestation</td><td align="left" rowspan="1" colspan="1">4 (5%)</td></tr><tr><td align="left" rowspan="1" colspan="1">Antenatal magnesium</td><td align="left" rowspan="1" colspan="1">20 (27%)</td></tr><tr><td align="left" colspan="1" rowspan="1">Gestational age (weeks)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Mean (SD; range)</td><td align="left" rowspan="1" colspan="1">25.8 (1.2; 22–27)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Median (Q1, Q3)</td><td align="left" rowspan="1" colspan="1">26.0 (25.0, 26.7)</td></tr><tr><td align="left" colspan="1" rowspan="1">Birth weight (g)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Mean (SD; range)</td><td align="left" rowspan="1" colspan="1">817.3 (213; 450–1410)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Median (Q1, Q3)</td><td align="left" rowspan="1" colspan="1">770 (660, 930)</td></tr><tr><td align="left" colspan="1" rowspan="1">Length of mechanical ventilation (days)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Mean (SD; range)</td><td align="left" rowspan="1" colspan="1">12.1 (21.6, 0–97)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Median (Q1, Q3)</td><td align="left" rowspan="1" colspan="1">3, (2, 6)</td></tr></tbody></table><table-wrap-foot><p><italic>Study group, <italic>n</italic> = 75</italic>.</p><p><italic>SD, standard deviation; g, grams; Q1 and Q3, first and third quartiles</italic>.</p></table-wrap-foot></table-wrap><table-wrap id="T2" position="float"><label>Table 2</label><caption><p><bold>Selected demographic, birth, laboratory, and outcome characteristics of the excluded patients <italic>n</italic> = 27 (no magnesium levels drawn)</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="left" rowspan="1" colspan="1">Characteristic</th><th align="left" rowspan="1" colspan="1">Excluded cohort <italic>n</italic> (%)</th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">Gender (male)</td><td align="left" rowspan="1" colspan="1">9 (33%)</td></tr><tr><td align="left" colspan="1" rowspan="1">Ethnicity</td></tr><tr><td align="left" rowspan="1" colspan="1"> Caucasian</td><td align="left" rowspan="1" colspan="1">20 (71.4%)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Hispanic</td><td align="left" rowspan="1" colspan="1">3 (11%)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Pacific-Islander</td><td align="left" rowspan="1" colspan="1">0 (0%)</td></tr><tr><td align="left" rowspan="1" colspan="1"> African-American</td><td align="left" rowspan="1" colspan="1">1 (3.6%)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Native American</td><td align="left" rowspan="1" colspan="1">0 (0%)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Asian</td><td align="left" rowspan="1" colspan="1">0 (0%)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Unknown</td><td align="left" rowspan="1" colspan="1">3 (14%)</td></tr><tr><td align="left" rowspan="1" colspan="1">Multiple gestation</td><td align="left" rowspan="1" colspan="1">4 (15%)</td></tr><tr><td align="left" colspan="1" rowspan="1">Gestational age (weeks)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Mean (range)</td><td align="left" rowspan="1" colspan="1">25.7 (23.8–27)</td></tr><tr><td align="left" colspan="1" rowspan="1">Birth weight (g)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Average</td><td align="left" rowspan="1" colspan="1">789</td></tr><tr><td align="left" rowspan="1" colspan="1"> Range</td><td align="left" rowspan="1" colspan="1">500–1110</td></tr><tr><td align="left" colspan="1" rowspan="1">Length of mechanical ventilation (days)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Mean (range)</td><td align="left" rowspan="1" colspan="1">8.6 (0–45)</td></tr></tbody></table></table-wrap><p>We queried the EDW using unique identifiers assigned to each of the cohort infants for the period including up to 5 years after birth. Data collected from the EDW included name; date of birth; gender; ethnicity; birth weight; birth head circumference; gestational age; presence of multiple gestation; administration of corticosteroids prior to delivery; administration of magnesium sulfate prior to delivery; mode of delivery; length of hospitalization; all neonatal total serum magnesium levels; days requiring mechanical ventilation; and the presence of seizures (ICD-9 codes 779.0 and 345.x). Of note, diagnoses of seizures at any time during the NICU hospitalization were excluded, as was the diagnosis of febrile seizures. Data extracted manually from the neurodevelopmental assessment at age 20–36 months included neurological exam for hypotonia, spasticity, and/or cerebral palsy.</p><p>Long-term follow-up for infants was assessed and recorded using two sources. First, for the outcome of the abnormal motor exam, data from the Utah State Department of Health Neonatal Follow-up Program were obtained. A standardized neurological motor exam was performed by a developmental pediatrician or pediatric neurologist when the infant was between 20 and 36 months of age. An abnormal motor exam was defined as cerebral palsy (including hypertonia/spasticity or dystonia), hypotonia, or spasticity. Second, for the outcome of epilepsy, we followed the infants for up to 5 years after birth using the EDW. We defined epilepsy as any encounter that had a record of the patient having had a seizure and for which the patient was placed on an anti-epileptic drug. The outcome of epilepsy excluded seizures that occurred solely during the NICU hospitalization; and febrile seizures.</p><p>We limited the time period of magnesium levels used in our analysis to that of prematurity only; i.e., less than 37 weeks gestation. In the IH system, the lower and upper limits of magnesium levels are defined as from 1.2 to 2.8 mg/dL, respectively.</p></sec><sec id="S2-3"><title>Statistical analysis</title><p>Statistical analyses were performed using SAS Analytics Pro version 9.3 (SAS Inc.). Descriptive statistics were used to characterize the study cohort. Wilcoxon Rank-Sum tests were used to compare magnesium levels for the outcomes of seizures and of composite abnormal motor exams. An alpha level of 0.05 was used to determine statistical significance; <italic>p</italic>-values were two-sided. For multivariate logistic regression analysis we modeled birth weight and magnesium levels by analysis of quartiles.</p></sec></sec><sec id="S3"><title>Results</title><p>We collected all total serum magnesium levels (<italic>n</italic> = 223) drawn on a cohort of 75 very-low birth weight (VLBW) infants during their initial, post-birth hospitalization (Table <xref ref-type="table" rid="T1">1</xref>). There were no deaths in the study cohort. On average infants had their magnesium levels checked three times, but the number of magnesium levels checked ranged from 1 to 17 times (Table <xref ref-type="table" rid="T3">3</xref>). Average total serum magnesium level was 2.4 mg/dL with a range of 1.1–5.8 mg/dL.</p><table-wrap id="T3" position="float"><label>Table 3</label><caption><p><bold>Characteristics of serum magnesium testing</bold>.</p></caption><table frame="hsides" rules="groups"><tbody><tr><td align="left" colspan="1" rowspan="1">Characteristic</td></tr><tr><td align="left" rowspan="1" colspan="1"> Average number of draws/infant</td><td align="left" rowspan="1" colspan="1">3</td></tr><tr><td align="left" rowspan="1" colspan="1"> Range number of draws/infant</td><td align="left" rowspan="1" colspan="1">1–17</td></tr><tr><td align="left" rowspan="1" colspan="1"> Age at first draw (DOL) (avg.)</td><td align="left" rowspan="1" colspan="1">3.5</td></tr><tr><td align="left" rowspan="1" colspan="1"> Range of age for first draw (DOL)</td><td align="left" rowspan="1" colspan="1">0–56</td></tr><tr><td align="left" rowspan="1" colspan="1"> Proportion of draws <3 DOL</td><td align="left" rowspan="1" colspan="1">36%</td></tr><tr><td align="left" colspan="1" rowspan="1">Serum levels (mg/dl)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Mean (SD; range)</td><td align="left" rowspan="1" colspan="1">2.4 (0.83; 1.1–5.8)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Median</td><td align="left" rowspan="1" colspan="1">2.2 (1.9, 2.7)</td></tr><tr><td align="left" rowspan="1" colspan="1"> Mode</td><td align="left" rowspan="1" colspan="1">2.1</td></tr></tbody></table><table-wrap-foot><p><italic>DOL, day of life; SD, standard deviation</italic>.</p></table-wrap-foot></table-wrap><p>There were 10 (13%) infants that had epilepsy, and 24 (32%) with abnormal motor exams. Abnormal motor exam, which was defined as the presence of cerebral palsy, hypotonia, or spasticity, was assessed between 20 and 36 months after birth. The outcome of epilepsy was assessed for by following patient outcomes for up to 5 years after birth. Importantly, to avoid inflating epilepsy rates, “epilepsy” was defined for the purpose of outcomes by excluding seizures which occurred in the neonatal period only, and by excluding febrile seizures.</p><p>All infants with epilepsy also had an abnormal motor exam. We found that children with abnormal motor exams had statistically significant lower magnesium levels in the neonatal period (<italic>p</italic> = 0.037) (Table <xref ref-type="table" rid="T4">4</xref>; Figure <xref ref-type="fig" rid="F1">1</xref>A). Infants who went on to develop epilepsy had lower average minimum magnesium levels in the neonatal period, but this was not statistically significant (<italic>p</italic> = 0.060) (Table <xref ref-type="table" rid="T4">4</xref>; Figure <xref ref-type="fig" rid="F1">1</xref>B). We also performed logistic regression analyses that included birth weight and magnesium level. While they showed a trend toward lower magnesium levels associated with increased risk for abnormal motor outcome or for epilepsy, the sample size was underpowered and did not show statistically significant differences (Table <xref ref-type="table" rid="T5">5</xref>).</p><table-wrap id="T4" position="float"><label>Table 4</label><caption><p><bold>Wilcoxon rank-sum tests for epilepsy and for abnormal motor exam</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="left" rowspan="1" colspan="1">Magnesium level</th><th align="center" colspan="2" rowspan="1">Outcome</th><th align="left" rowspan="1" colspan="1"><italic>p</italic>-Value</th></tr></thead><tbody><tr><td align="left" colspan="4" rowspan="1"><bold>Epilepsy</bold></td></tr><tr><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1">No (<italic>n</italic> = 65)</td><td align="left" rowspan="1" colspan="1">Yes (<italic>n</italic> = 10)</td><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1">Median</td><td align="left" rowspan="1" colspan="1">2.30</td><td align="left" rowspan="1" colspan="1">1.95</td><td align="left" rowspan="1" colspan="1">0.060</td></tr><tr><td align="left" colspan="4" rowspan="1"><bold>Abnormal motor exam</bold></td></tr><tr><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1">No (<italic>n</italic> = 51)</td><td align="left" rowspan="1" colspan="1">Yes (<italic>n</italic> = 24)</td><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1">Median</td><td align="left" rowspan="1" colspan="1">2.30</td><td align="left" rowspan="1" colspan="1">2.00</td><td align="left" rowspan="1" colspan="1">0.037</td></tr></tbody></table><table-wrap-foot><p><italic>The number of patients with the outcome is indicated in parentheses. Magnesium levels are milligrams per deciliter. Two-sided <italic>p</italic>-values are given</italic>.</p></table-wrap-foot></table-wrap><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Strip scatterplots of average magnesium levels (<italic>x</italic>-axis) and neurodevelopmental outcomes, for abnormal motor exam (A), and epilepsy (B)</bold>. Thin line is the median; dotted lines show 25th and 75th quartiles.</p></caption><graphic xlink:href="fped-02-00120-g001"/></fig><table-wrap id="T5" position="float"><label>Table 5</label><caption><p><bold>Logistic regression results; (A) Multivariable regression analysis for the outcome of seizures, analyzed for birth weight and magnesium levels; (B) Multivariable regression analysis for the outcome of abnormal motor exam, analyzed for birth weight, and magnesium levels</bold>.</p></caption><table frame="hsides" rules="groups"><tbody><tr><td align="left" colspan="5" style="background-color:Darkgray;" rowspan="1"><bold>(A)</bold></td></tr><tr><td align="left" colspan="5" rowspan="1"><bold>Average magnesium level quartile by seizures</bold></td></tr><tr><td align="left" rowspan="1" colspan="1">Quartile of Mg level</td><td align="left" rowspan="1" colspan="1">No</td><td align="left" rowspan="1" colspan="1">Yes</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1">Q1, Mg level ≤1.9</td><td align="left" rowspan="1" colspan="1">17 (77%)</td><td align="left" rowspan="1" colspan="1">5 (23%)</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1">Q2, 1.9 < Mg level ≤2.2</td><td align="left" rowspan="1" colspan="1">14 (82%)</td><td align="left" rowspan="1" colspan="1">3 (18%)</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1">Q3, 2.2 < Mg level ≤2.7</td><td align="left" rowspan="1" colspan="1">18 (95%)</td><td align="left" rowspan="1" colspan="1">1 (5%)</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1">Q4, Mg level > 2.7</td><td align="left" rowspan="1" colspan="1">16 (94%)</td><td align="left" rowspan="1" colspan="1">1 (6%)</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" colspan="5" rowspan="1"><bold>Birth weight quartile by seizures</bold></td></tr><tr><td align="left" rowspan="1" colspan="1">Quartile of Mg level</td><td align="left" rowspan="1" colspan="1">No</td><td align="left" rowspan="1" colspan="1">Yes</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1">Q1, Birth weight ≤660</td><td align="left" rowspan="1" colspan="1">19 (95%)</td><td align="left" rowspan="1" colspan="1">1 (5%)</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1">Q2, 660 < Birth weight ≤770</td><td align="left" rowspan="1" colspan="1">14 (78%)</td><td align="left" rowspan="1" colspan="1">4 (22%)</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1">Q3, 770 < Birth weight ≤930</td><td align="left" rowspan="1" colspan="1">16 (84%)</td><td align="left" rowspan="1" colspan="1">3 (16%)</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1">Q4, Birth weight > 930</td><td align="left" rowspan="1" colspan="1">16 (89%)</td><td align="left" rowspan="1" colspan="1">2 (11%)</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" colspan="5" rowspan="1"><hr/></td></tr><tr><td align="left" rowspan="1" colspan="1"><bold>Parameter</bold></td><td align="left" rowspan="1" colspan="1"><bold>Estimate</bold></td><td align="left" rowspan="1" colspan="1"><bold>LCL</bold></td><td align="left" rowspan="1" colspan="1"><bold>UCL</bold></td><td align="left" rowspan="1" colspan="1"><bold><italic>p</italic>-Value</bold></td></tr><tr><td align="left" colspan="5" rowspan="1"><hr/></td></tr><tr><td align="left" colspan="5" rowspan="1"><bold>Birth weight (Q4 = Ref)</bold></td></tr><tr><td align="left" rowspan="1" colspan="1">Q1</td><td align="left" rowspan="1" colspan="1">1.00</td><td align="left" rowspan="1" colspan="1">0.22</td><td align="left" rowspan="1" colspan="1">4.51</td><td align="left" rowspan="1" colspan="1">1.00</td></tr><tr><td align="left" rowspan="1" colspan="1">Q2</td><td align="left" rowspan="1" colspan="1">1.34</td><td align="left" rowspan="1" colspan="1">0.31</td><td align="left" rowspan="1" colspan="1">5.89</td><td align="left" rowspan="1" colspan="1">0.696</td></tr><tr><td align="left" rowspan="1" colspan="1">Q3</td><td align="left" rowspan="1" colspan="1">1.15</td><td align="left" rowspan="1" colspan="1">0.28</td><td align="left" rowspan="1" colspan="1">4.76</td><td align="left" rowspan="1" colspan="1">0.842</td></tr><tr><td align="left" colspan="5" rowspan="1"><bold>Milligrams level (Q4 = Ref)</bold></td></tr><tr><td align="left" rowspan="1" colspan="1">Q1</td><td align="left" rowspan="1" colspan="1">2.40</td><td align="left" rowspan="1" colspan="1">0.62</td><td align="left" rowspan="1" colspan="1">9.24</td><td align="left" rowspan="1" colspan="1">0.842</td></tr><tr><td align="left" rowspan="1" colspan="1">Q2</td><td align="left" rowspan="1" colspan="1">1.30</td><td align="left" rowspan="1" colspan="1">0.30</td><td align="left" rowspan="1" colspan="1">5.61</td><td align="left" rowspan="1" colspan="1">0.725</td></tr><tr><td align="left" rowspan="1" colspan="1">Q3</td><td align="left" rowspan="1" colspan="1">0.28</td><td align="left" rowspan="1" colspan="1">0.05</td><td align="left" rowspan="1" colspan="1">1.72</td><td align="left" rowspan="1" colspan="1">0.169</td></tr><tr><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" colspan="5" style="background-color:DarkGray;" rowspan="1"><bold>(B)</bold></td></tr><tr><td align="left" colspan="5" rowspan="1"><bold>Average magnesium level quartile by abnormal motor scores</bold></td></tr><tr><td align="left" rowspan="1" colspan="1">Quartile of milligram level</td><td align="left" rowspan="1" colspan="1">No</td><td align="left" rowspan="1" colspan="1">Yes</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1">Q1, Mg level ≤1.9</td><td align="left" rowspan="1" colspan="1">11 (50%)</td><td align="left" rowspan="1" colspan="1">11 (50%)</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1">Q2, 1.9 < Mg level ≤2.2</td><td align="left" rowspan="1" colspan="1">11 (65%)</td><td align="left" rowspan="1" colspan="1">6 (35%)</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1">Q3, 2.2 < Mg level ≤2.7</td><td align="left" rowspan="1" colspan="1">17 (89%)</td><td align="left" rowspan="1" colspan="1">2 (11%)</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1">Q4, Mg level >2.7</td><td align="left" rowspan="1" colspan="1">12 (71%)</td><td align="left" rowspan="1" colspan="1">5 (29%)</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" colspan="5" rowspan="1"><bold>Birth weight quartile by abnormal motor scores</bold></td></tr><tr><td align="left" rowspan="1" colspan="1">Quartile of milligrams Level</td><td align="left" rowspan="1" colspan="1">No</td><td align="left" rowspan="1" colspan="1">Yes</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1">Q1, Birth weight ≤660</td><td align="left" rowspan="1" colspan="1">17 (77%)</td><td align="left" rowspan="1" colspan="1">5 (23%)</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1">Q2, 660 < Birth weight ≤770</td><td align="left" rowspan="1" colspan="1">14 (82%)</td><td align="left" rowspan="1" colspan="1">3 (18%)</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1">Q3, 770 < Birth weight ≤930</td><td align="left" rowspan="1" colspan="1">18 (95%)</td><td align="left" rowspan="1" colspan="1">1 (5%)</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" rowspan="1" colspan="1">Q4, Birth weight >930</td><td align="left" rowspan="1" colspan="1">16 (94%)</td><td align="left" rowspan="1" colspan="1">1 (6%)</td><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"/></tr><tr><td align="left" colspan="5" rowspan="1"><hr/></td></tr><tr><td align="left" rowspan="1" colspan="1"><bold>Parameter</bold></td><td align="left" rowspan="1" colspan="1"><bold>Estimate</bold></td><td align="left" rowspan="1" colspan="1"><bold>LCL</bold></td><td align="left" rowspan="1" colspan="1"><bold>UCL</bold></td><td align="left" rowspan="1" colspan="1"><bold><italic>p</italic>-Value</bold></td></tr><tr><td align="left" colspan="5" rowspan="1"><hr/></td></tr><tr><td align="left" colspan="5" rowspan="1"><bold>Birth weight (Q4 = Ref)</bold></td></tr><tr><td align="left" rowspan="1" colspan="1">Q1</td><td align="left" rowspan="1" colspan="1">0.61</td><td align="left" rowspan="1" colspan="1">0.05</td><td align="left" rowspan="1" colspan="1">7.77</td><td align="left" rowspan="1" colspan="1">0.700</td></tr><tr><td align="left" rowspan="1" colspan="1">Q2</td><td align="left" rowspan="1" colspan="1">3.10</td><td align="left" rowspan="1" colspan="1">0.45</td><td align="left" rowspan="1" colspan="1">21.39</td><td align="left" rowspan="1" colspan="1">0.251</td></tr><tr><td align="left" rowspan="1" colspan="1">Q3</td><td align="left" rowspan="1" colspan="1">1.46</td><td align="left" rowspan="1" colspan="1">0.20</td><td align="left" rowspan="1" colspan="1">10.50</td><td align="left" rowspan="1" colspan="1">0.707</td></tr><tr><td align="left" colspan="5" rowspan="1"><bold>Milligrams level (Q4 = Ref)</bold></td></tr><tr><td align="left" rowspan="1" colspan="1">Q1</td><td align="left" rowspan="1" colspan="1">4.82</td><td align="left" rowspan="1" colspan="1">0.48</td><td align="left" rowspan="1" colspan="1">47.91</td><td align="left" rowspan="1" colspan="1">0.180</td></tr><tr><td align="left" rowspan="1" colspan="1">Q2</td><td align="left" rowspan="1" colspan="1">3.51</td><td align="left" rowspan="1" colspan="1">0.31</td><td align="left" rowspan="1" colspan="1">39.89</td><td align="left" rowspan="1" colspan="1">0.311</td></tr><tr><td align="left" rowspan="1" colspan="1">Q3</td><td align="left" rowspan="1" colspan="1">0.86</td><td align="left" rowspan="1" colspan="1">0.047</td><td align="left" rowspan="1" colspan="1">15.59</td><td align="left" rowspan="1" colspan="1">0.917</td></tr></tbody></table><table-wrap-foot><p><italic>LCL, 95% lower confidence limit; UCL, 95% upper confidence limit; Q1, first quartile</italic>.</p><p><italic>(A) Fisher’s exact <italic>p</italic>-value = 0.316; Fisher’s exact <italic>p</italic>-value = 0.891</italic>.</p><p><italic>(B) Fisher’s exact <italic>p</italic>-Value = 0.058; Fisher’s exact <italic>p</italic>-value = 0.434</italic>.</p></table-wrap-foot></table-wrap><p>We also considered whether there might be potential clinical history confounders, by examining whether there were other clinical characteristics associated with high or low magnesium levels (high defined as 2.3 mg/dL or above, low 2.2 mg/dL and below) (Table <xref ref-type="table" rid="T6">6</xref>), and by analyzing whether other clinical characteristics were associated with the outcomes of epilepsy or abnormal motor outcome (Table <xref ref-type="table" rid="T7">7</xref>). While there were some differences between the groups, there was not a consistent trend.</p><table-wrap id="T6" position="float"><label>Table 6</label><caption><p><bold>Comparison of clinical variables in “low” and “high” magnesium patient groups</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="left" rowspan="1" colspan="1">Clinical variable</th><th align="left" rowspan="1" colspan="1">Low magnesium</th><th align="left" rowspan="1" colspan="1">High magnesium</th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">Male gender</td><td align="left" rowspan="1" colspan="1">57%</td><td align="left" rowspan="1" colspan="1">59%</td></tr><tr><td align="left" rowspan="1" colspan="1">Caucasian</td><td align="left" rowspan="1" colspan="1">71%</td><td align="left" rowspan="1" colspan="1">68%</td></tr><tr><td align="left" rowspan="1" colspan="1">Birth weight</td><td align="left" rowspan="1" colspan="1">872 g</td><td align="left" rowspan="1" colspan="1">755 g</td></tr><tr><td align="left" rowspan="1" colspan="1">Gestational age</td><td align="left" rowspan="1" colspan="1">25.9 weeks</td><td align="left" rowspan="1" colspan="1">25.7 weeks</td></tr><tr><td align="left" rowspan="1" colspan="1">Maternal age</td><td align="left" rowspan="1" colspan="1">27.5 years</td><td align="left" rowspan="1" colspan="1">28.4 years</td></tr><tr><td align="left" rowspan="1" colspan="1">Maternal parity</td><td align="left" rowspan="1" colspan="1">3.1</td><td align="left" rowspan="1" colspan="1">2.6</td></tr><tr><td align="left" rowspan="1" colspan="1">Maternal smoking</td><td align="left" rowspan="1" colspan="1">29%</td><td align="left" rowspan="1" colspan="1">14%</td></tr><tr><td align="left" rowspan="1" colspan="1">Medicaid</td><td align="left" rowspan="1" colspan="1">36%</td><td align="left" rowspan="1" colspan="1">38%</td></tr><tr><td align="left" rowspan="1" colspan="1">Maternal diabetes</td><td align="left" rowspan="1" colspan="1">0</td><td align="left" rowspan="1" colspan="1">5.4%</td></tr><tr><td align="left" rowspan="1" colspan="1">Maternal hypertension</td><td align="left" rowspan="1" colspan="1">7.1%</td><td align="left" rowspan="1" colspan="1">32%</td></tr><tr><td align="left" rowspan="1" colspan="1">Maternal drug use</td><td align="left" rowspan="1" colspan="1">7.1%</td><td align="left" rowspan="1" colspan="1">0</td></tr><tr><td align="left" rowspan="1" colspan="1">Multiple gestation</td><td align="left" rowspan="1" colspan="1">0</td><td align="left" rowspan="1" colspan="1">10.8%</td></tr><tr><td align="left" rowspan="1" colspan="1">Antepartum hemorrhage</td><td align="left" rowspan="1" colspan="1">29%</td><td align="left" rowspan="1" colspan="1">16%</td></tr><tr><td align="left" rowspan="1" colspan="1">Chorioamnionitis</td><td align="left" rowspan="1" colspan="1">4.8%</td><td align="left" rowspan="1" colspan="1">2.7%</td></tr><tr><td align="left" rowspan="1" colspan="1">Steroids pre-delivery</td><td align="left" rowspan="1" colspan="1">83%</td><td align="left" rowspan="1" colspan="1">73%</td></tr><tr><td align="left" rowspan="1" colspan="1">Ventilator days</td><td align="left" rowspan="1" colspan="1">12.3</td><td align="left" rowspan="1" colspan="1">10.9</td></tr><tr><td align="left" rowspan="1" colspan="1">ECMO</td><td align="left" rowspan="1" colspan="1">0%</td><td align="left" rowspan="1" colspan="1">0%</td></tr><tr><td align="left" rowspan="1" colspan="1">Hydrocephalus</td><td align="left" rowspan="1" colspan="1">7.1%</td><td align="left" rowspan="1" colspan="1">8.1%</td></tr><tr><td align="left" rowspan="1" colspan="1">NEC</td><td align="left" rowspan="1" colspan="1">17%</td><td align="left" rowspan="1" colspan="1">11%</td></tr><tr><td align="left" rowspan="1" colspan="1">IVH (Grade II, III, IV)</td><td align="left" rowspan="1" colspan="1">40%</td><td align="left" rowspan="1" colspan="1">27%</td></tr><tr><td align="left" rowspan="1" colspan="1">Length of stay</td><td align="left" rowspan="1" colspan="1">109 days</td><td align="left" rowspan="1" colspan="1">104 days</td></tr></tbody></table><table-wrap-foot><p><italic>“Low” magnesium was defined as average magnesium level of 2.2 mg/dL or lower; “high” was defined as 2.3 mg/dL or higher</italic>.</p><p><italic>ECMO, extra-corporeal membrane oxygenation; IVH, intraventricular hemorrhage; NEC, necrotizing enterocolitis</italic>.</p></table-wrap-foot></table-wrap><table-wrap id="T7" position="float"><label>Table 7</label><caption><p><bold>Comparison of clinical variables and the outcomes for epilepsy or abnormal motor exam</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="left" rowspan="1" colspan="1">Clinical variable</th><th align="left" rowspan="1" colspan="1">Epilepsy</th><th align="left" rowspan="1" colspan="1">Abnormal motor</th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1">+/−</td><td align="left" rowspan="1" colspan="1">+/−</td></tr><tr><td align="left" rowspan="1" colspan="1">Male gender</td><td align="left" rowspan="1" colspan="1">90/57%</td><td align="left" rowspan="1" colspan="1">65/55%</td></tr><tr><td align="left" rowspan="1" colspan="1">Caucasian</td><td align="left" rowspan="1" colspan="1">60/71%</td><td align="left" rowspan="1" colspan="1">69/70%</td></tr><tr><td align="left" rowspan="1" colspan="1">Birth weight</td><td align="left" rowspan="1" colspan="1">832/815 g</td><td align="left" rowspan="1" colspan="1">833/810 g</td></tr><tr><td align="left" rowspan="1" colspan="1">Gestational age (days)</td><td align="left" rowspan="1" colspan="1">179/181</td><td align="left" rowspan="1" colspan="1">180/181</td></tr><tr><td align="left" rowspan="1" colspan="1">Maternal age</td><td align="left" rowspan="1" colspan="1">25.5/28.3</td><td align="left" rowspan="1" colspan="1">25.7/29</td></tr><tr><td align="left" rowspan="1" colspan="1">Maternal parity</td><td align="left" rowspan="1" colspan="1">4.2/2.7</td><td align="left" rowspan="1" colspan="1">2.47/1.74</td></tr><tr><td align="left" rowspan="1" colspan="1">Maternal smoking</td><td align="left" rowspan="1" colspan="1">10/23%</td><td align="left" rowspan="1" colspan="1">12/26%</td></tr><tr><td align="left" rowspan="1" colspan="1">Medicaid</td><td align="left" rowspan="1" colspan="1">60/33%</td><td align="left" rowspan="1" colspan="1">46/32%</td></tr><tr><td align="left" rowspan="1" colspan="1">Maternal diabetes</td><td align="left" rowspan="1" colspan="1">0/2.9%</td><td align="left" rowspan="1" colspan="1">0/3.8%</td></tr><tr><td align="left" rowspan="1" colspan="1">Maternal hypertension</td><td align="left" rowspan="1" colspan="1">0/21.7%</td><td align="left" rowspan="1" colspan="1">12/23%</td></tr><tr><td align="left" rowspan="1" colspan="1">Maternal drug use</td><td align="left" rowspan="1" colspan="1">0/4.3%</td><td align="left" rowspan="1" colspan="1">0/5.7%</td></tr><tr><td align="left" rowspan="1" colspan="1">Multiple gestation</td><td align="left" rowspan="1" colspan="1">0/5.8%</td><td align="left" rowspan="1" colspan="1">0/7.5%</td></tr><tr><td align="left" rowspan="1" colspan="1">Antepartum hemorrhage</td><td align="left" rowspan="1" colspan="1">50/19%</td><td align="left" rowspan="1" colspan="1">31/19%</td></tr><tr><td align="left" rowspan="1" colspan="1">Chorioamnionitis</td><td align="left" rowspan="1" colspan="1">0/4.3%</td><td align="left" rowspan="1" colspan="1">3.8/5.7%</td></tr><tr><td align="left" rowspan="1" colspan="1">Mg pre-delivery</td><td align="left" rowspan="1" colspan="1">20/35%</td><td align="left" rowspan="1" colspan="1">19/28%</td></tr><tr><td align="left" rowspan="1" colspan="1">Steroids pre-delivery</td><td align="left" rowspan="1" colspan="1">90/77%</td><td align="left" rowspan="1" colspan="1">88/74%</td></tr><tr><td align="left" rowspan="1" colspan="1">Ventilator days</td><td align="left" rowspan="1" colspan="1">19/11</td><td align="left" rowspan="1" colspan="1">17/9.2</td></tr><tr><td align="left" rowspan="1" colspan="1">ECMO</td><td align="left" rowspan="1" colspan="1">0/0%</td><td align="left" rowspan="1" colspan="1">0/0%</td></tr><tr><td align="left" rowspan="1" colspan="1">Hydrocephalus</td><td align="left" rowspan="1" colspan="1">20/5.8%</td><td align="left" rowspan="1" colspan="1">15/3.8%</td></tr><tr><td align="left" rowspan="1" colspan="1">NEC</td><td align="left" rowspan="1" colspan="1">10/14%</td><td align="left" rowspan="1" colspan="1">12/15%</td></tr><tr><td align="left" rowspan="1" colspan="1">IVH (Grade II, III, IV)</td><td align="left" rowspan="1" colspan="1">50/32</td><td align="left" rowspan="1" colspan="1">50/26</td></tr><tr><td align="left" rowspan="1" colspan="1">Length of stay (days)</td><td align="left" rowspan="1" colspan="1">132/103</td><td align="left" rowspan="1" colspan="1">123/98</td></tr></tbody></table><table-wrap-foot><p><italic>+, diagnosis present; −, diagnosis absent (e.g., normal)</italic>.</p><p><italic>ECMO, extra-corporeal membrane oxygenation; IVH, intraventricular hemorrhage; NEC, necrotizing enterocolitis</italic>.</p></table-wrap-foot></table-wrap></sec><sec sec-type="discussion" id="S4"><title>Discussion</title><p>We examined the correlation between postnatal magnesium levels and neurodevelopmental outcomes in premature infants. While prenatal magnesium administration has been associated with a decreased risk of cerebral palsy (<xref rid="B18" ref-type="bibr">18</xref>–<xref rid="B20" ref-type="bibr">20</xref>), our study demonstrates that postnatal serum magnesium levels in VLBW infants may be associated with improved neurodevelopmental outcomes. Specifically, higher serum magnesium levels in preterm infants were associated with lower rates of an abnormal motor exam (spasticity, cerebral palsy, or hypotonia). There was a lower risk for epilepsy, but this finding was not statistically significant.</p><p>Our study raises the possibility that there may be a broader window of opportunity for magnesium administration in premature infants to help improve neurodevelopmental outcomes. That is, perhaps magnesium supplementation could also be considered in premature infants and not only in mothers at risk for preterm birth. Use of magnesium sulfate would have to be balanced with concerns for potential adverse side-effects in premature infants (<xref rid="B29" ref-type="bibr">29</xref>, <xref rid="B30" ref-type="bibr">30</xref>). Another issue requiring further study will be research on the mechanism(s) of potential neuroprotection. This is because most infants had a magnesium level in the “normal” range, and because there was significant overlap in the magnesium levels of infants with normal outcomes compared to infants who developed epilepsy or abnormal motor outcomes (Figure <xref ref-type="fig" rid="F1">1</xref>).</p><p>Animal model data show a neuroprotective role for magnesium against injury in the developing CNS (<xref rid="B25" ref-type="bibr">25</xref>–<xref rid="B27" ref-type="bibr">27</xref>). Further, in term infants with birth asphyxia a potentially protective role has been suggested for magnesium (<xref rid="B21" ref-type="bibr">21</xref>–<xref rid="B24" ref-type="bibr">24</xref>). There are multiple potential mechanism(s) by which magnesium could exert neuroprotective effects. Magnesium can block calcium influx through the <italic>N</italic>-methyl-<sc>d</sc>-aspartate (NMDA) receptor channel and thereby reduce glutamate excitotoxicity; can reduce inflammatory cytokine and free radical production; can stabilize membranes; and can normalize blood pressure fluctuations (<xref rid="B31" ref-type="bibr">31</xref>–<xref rid="B33" ref-type="bibr">33</xref>). Magnesium can also prevent activation of the hypoxia inducible factor 1α (HIF1α) pathway that leads to axon pathfinding errors (<xref rid="B27" ref-type="bibr">27</xref>).</p><p>Limitations of this study included sample size, retrospective data collection, and missing follow-up for some infants. While the presence of an abnormal neuro-motor exam can be determined by age 20 months as was done in this study, a more extensive longitudinal study with objective scoring, such as using the Bailey Scale of Infant Development, would provide more reliable outcomes data. We excluded 32 infants from our analysis because of incomplete data, including a lack of magnesium levels. The small sample size also led to limitations on performing multivariate regression analyzes (<xref rid="B34" ref-type="bibr">34</xref>). Another bias could arise because timing of magnesium level blood draws were not evenly distributed in the different infants; and were not distributed evenly across the hospitalization. Because of the sample size, we were not able to control for multiple factors that could play an important role in the outcomes, such as intraventricular hemorrhage; maternal steroid or magnesium administration; or birth weight, among others. Pharmacokinetics and pharmacodynamics of serum magnesium can be affected by both endogenous and iatrogenic factors, including medications and calcium metabolism. Our study did not address these issues. In fact, normative premature infant magnesium levels, and effects of maternal magnesium administration on neonatal levels, are under active study (<xref rid="B35" ref-type="bibr">35</xref>, <xref rid="B36" ref-type="bibr">36</xref>), and are important subjects for future research.</p><p>Our pilot findings raise the possibility that magnesium levels during a critical developmental time window could affect neurological outcome. In the U.S. 500,000 births each year are premature, while worldwide it is estimated that 12.9 million infants yearly are born before 37 weeks gestation (<xref rid="B9" ref-type="bibr">9</xref>, <xref rid="B37" ref-type="bibr">37</xref>, <xref rid="B38" ref-type="bibr">38</xref>). This significant burden of prematurity, with its attendant risks for adverse neurodevelopmental outcomes, warrants further investigations into potential neuroprotective roles for magnesium after preterm birth.</p></sec><sec id="S5"><title>Conflict of Interest Statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec>
Immediate Effects of Maternal Deprivation on the (Re)Activity of the HPA-Axis Differ in CD1 and C57Bl/6J Mouse Pups	<p>The postnatal development of the mouse is characterized by a period of hypo-responsiveness of the hypothalamic–pituitary–adrenal (HPA) axis to mild stressors. Maternal deprivation (MD) during this period can disrupt the quiescence of the HPA-axis. The present study examined the influence of strain (outbred CD1 vs. inbred C57BL/6J mice) on some central and peripheral components of the HPA-axis in neonatal mice (5-day-old) in the presence of their mother or after 24 h MD (on postnatal day 4) under basal or mild stressful conditions. In the presence of the dam, adrenal corticosterone (CORT) secretion was low in both mouse strains. Compared to CD1 mice, C57BL/6J had lower CORT levels associated with higher ACTH levels and ACTH/CORT ratio (i.e., lower adrenal sensitivity to ACTH), and higher glucocorticoid receptor (GR) mRNA expression in the paraventricular nucleus. Although MD disinhibited the HPA-axis in both strains as reflected by increased basal CORT and ACTH, we found a strain-dependent pattern. MD increased CORT more in C57BL/6J compared to CD1 mice together with a lower ACTH/CORT ratio (i.e., higher adrenal sensitivity to ACTH), while GR mRNA was no longer different in the two strains. However, this increased adrenal sensitivity in maternally deprived C57BL/6J mice was not reflected in their CORT response to a subsequent novelty stressor, possibly due to an MD-induced ceiling effect in their steroidogenic capacity. In conclusion, the immediate outcome of MD depends on the genetic background of the mother–infant dyad, suggesting that maybe also the outcome in later-life cannot be generalized.</p>	<contrib contrib-type="author"><name><surname>Daskalakis</surname><given-names>Nikolaos P.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><xref ref-type="aff" rid="aff3"><sup>3</sup></xref><xref ref-type="aff" rid="aff4"><sup>4</sup></xref><xref ref-type="aff" rid="aff5"><sup>5</sup></xref><xref ref-type="corresp" rid="cor1">*</xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/5716"/></contrib><contrib contrib-type="author"><name><surname>Enthoven</surname><given-names>Leo</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref></contrib><contrib contrib-type="author"><name><surname>Schoonheere</surname><given-names>Edwige</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref></contrib><contrib contrib-type="author"><name><surname>de Kloet</surname><given-names>Edo Ronald</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/613"/></contrib><contrib contrib-type="author"><name><surname>Oitzl</surname><given-names>Melly S.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="aff" rid="aff6"><sup>6</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/389"/></contrib>	Frontiers in Endocrinology	<sec sec-type="introduction" id="S1"><title>Introduction</title><p>Maternal stimuli play a central role in the postnatal development of the hypothalamic–pituitary–adrenal axis (HPA-axis) in rodents (<xref rid="B1" ref-type="bibr">1</xref>, <xref rid="B2" ref-type="bibr">2</xref>) especially during the stress-hyporesponsive period (SHRP). The SHRP lasts from postnatal day (pnd) 1–12 in mice, (pnd 3–14 in rats) and is characterized by low basal levels of corticosterone (CORT) and an inability to elicit a CORT response to mild stress (<xref rid="B3" ref-type="bibr">3</xref>, <xref rid="B4" ref-type="bibr">4</xref>). Rodent dams do not leave often their nest for longer than 15–20 min during the SHRP (<xref rid="B5" ref-type="bibr">5</xref>). Removal of the mother for prolonged time periods (>3–8 to 24 h) has been shown to activate the HPA-axis, while the axis also becomes responsive to mild stressors, which may modulate ongoing developmental programs (<xref rid="B6" ref-type="bibr">6</xref>, <xref rid="B7" ref-type="bibr">7</xref>).</p><p>Large individual variations in the long-term biobehavioral outcome of early-life traumatic experiences have been reported in humans (<xref rid="B8" ref-type="bibr">8</xref>) and rodents (<xref rid="B9" ref-type="bibr">9</xref>, <xref rid="B10" ref-type="bibr">10</xref>). This raised the idea that early-life trauma might shape pre-existing genetic vulnerability to certain stressful conditions in later life (<xref rid="B11" ref-type="bibr">11</xref>). Maternal deprivation (MD) is a commonly used animal paradigm to study the consequences of early-life trauma on adult stress–responses and related behaviors (<xref rid="B12" ref-type="bibr">12</xref>). The MD paradigm has been applied in various designs ranging from single 24 h deprivations to repeated daily separations in time periods ranging between 3 and 8 h (<xref rid="B9" ref-type="bibr">9</xref>, <xref rid="B11" ref-type="bibr">11</xref>).</p><p>Most of our knowledge on the effects of MD on HPA-axis and stress-related behaviors is based on research in outbred rodent strains. Although it is known that genetically selected lines of rats display differential sensitivity to the long-term effects of MD (<xref rid="B13" ref-type="bibr">13</xref>–<xref rid="B15" ref-type="bibr">15</xref>), the aspect of genetic predisposition has been given little attention. In recent experiments, we showed that responsiveness to mild stressors following prolonged maternal absence is strain-dependent (<xref rid="B16" ref-type="bibr">16</xref>). We actually observed that while maternally separated pups (i.e., repeatedly for 8 h) habituate toward the maternal absence <italic>per se</italic> by displaying low basal CORT levels (<xref rid="B16" ref-type="bibr">16</xref>–<xref rid="B18" ref-type="bibr">18</xref>), their CORT response toward a subsequent heterotypic stressor sensitizes in a strain-dependent fashion: deprived Long Evans pups were more re-active to the subsequent stressor than similarly deprived Wistar rats (<xref rid="B16" ref-type="bibr">16</xref>).</p><p>The inbred C57BL/6J mouse strain is most widely used as genetic background strain for engineering genetic mouse models for human diseases. A few studies compared C57BL/6J mice with common outbred mice strains (e.g., CD1 mice) on stress-related physiology and behavior. C57BL/6J and CD1 mice have differences in their circadian pattern of the stress–response (<xref rid="B19" ref-type="bibr">19</xref>). C57BL/6J mice have lower stress responsiveness in a light/dark exploration test for anxiety (<xref rid="B20" ref-type="bibr">20</xref>) and display a reduced exploration in a novel environment (<xref rid="B21" ref-type="bibr">21</xref>). Furthermore, CD1 mice showed better avoidance learning in a Y-maze task (<xref rid="B22" ref-type="bibr">22</xref>). Interestingly, C57BL/6J and CD1 mice seem to display differences on the long-term outcome of maternal separation on the stress–response, cognitive performance, anxiety/depression-like, or schizophrenia-like behaviors (<xref rid="B23" ref-type="bibr">23</xref>–<xref rid="B33" ref-type="bibr">33</xref>). Generally, the reported effects indicate more often significant effects in C57BL/6J than in CD1 mice.</p><p>Studying the immediate effects of MD on the development of the stress system responsiveness might give insights on the salient factors that influence the long-term outcome. This is an approach proven to be successful using a variety of early-life stress paradigms (<xref rid="B18" ref-type="bibr">18</xref>, <xref rid="B34" ref-type="bibr">34</xref>). In the current study, we compared C57BL/6J with CD1 mouse pups with regard to the immediate effects of pnd 4 MD on HPA-axis stress reactivity.</p></sec><sec sec-type="materials\|methods" id="S2"><title>Materials and Methods</title><sec id="S2-1"><title>Animals</title><p>Offspring of CD1 and C57BL/6J mice were used in this study. All mice were housed under a 12:12 h light/dark cycle (lights on at 07:00 hours) and constant temperature (23 ± 2°C) and humidity (55 ± 5%) conditions. Food (SRM-A; Hope Farms, Woerden, The Netherlands) and water (172 ml HCl/200l H<sub>2</sub>O) was provided <italic>ad libitum</italic>. Three females were mated with one male in polycarbonate boxes (820 cm<sup>3</sup>) containing sawdust bedding. Pregnant females were transferred to clean polycarbonate cages containing sawdust and two sheets of paper towels for nesting material. Pregnant females were checked for litters daily between 09:00 and 10:00 hours. If litters were found, the day of birth was defined as day 0 for that litter. On the day after parturition, day 1, each litter was culled to eight healthy pups (four males and four females) for the CD1 strain and to six healthy pups (three males and three females) for the C57BL/6J strain and then remained undisturbed until used in the experiment. A total of four CD1 litters and six C57BL/6J litters were used in the study. Animal experiments were approved by the Local Committee for Animal Health, Ethics, and Research of Leiden University and carried out in accordance with European Communities Council Directive 86/609/EEC.</p></sec><sec id="S2-2"><title>Experimental design</title><p>At postnatal day 4, mothers from nests randomly selected for MD (two CD1 and three C57BL/6J nests) were removed from the home cage. The home cage containing the pups was transferred to an adjacent room with similar light and temperature conditions and placed on a heating pad set at 30–33°C. Neither food nor water was available during MD. At pnd 5, half of the non-deprived (NON-DEP) and half of the deprived (DEP) pups were decapitated immediately to provide a basal sample for measurements in blood and brain. The remaining NON-DEP and DEP pups were placed individually in novel cages containing clean sawdust bedding on heating pads set at 30–33°C and decapitated after 30 min to provide a novelty stress sample.</p></sec><sec id="S2-3"><title>Blood processing and hormone determination</title><p>Trunk blood from all decapitated pups was collected in EDTA-coated microcentrifuge tubes; plasma was extracted and stored frozen at −20°C until hormone determination. ACTH was measured by radioimmunoassay (RIA; MP Biomedicals, LLC, NY, USA; sensitivity 10 pg/ml, intra-assay variation 4.1%, inter-assay variation 4.4%) as described before (<xref rid="B16" ref-type="bibr">16</xref>). CORT was measured by RIA (MP Biomedicals, LLC, NY, USA; sensitivity 1.25 ng/ml, intra-assay variation, 4.4%, interassay variation 6.5%) as described before (<xref rid="B16" ref-type="bibr">16</xref>). We calculated the ratio of ACTH to CORT as an indirect measure of adrenal sensitivity to ACTH (<xref rid="B18" ref-type="bibr">18</xref>).</p></sec><sec id="S2-4"><title><italic>In situ</italic> hybridization</title><p>Frozen brains and pituitaries were sectioned at −20°C in a cryostat microtome at 16 μm in the coronal plane. Sections were thaw-mounted on poly-<sc>l</sc>-lysine coated slides, air-dried, and kept at −80°C. <italic>In situ</italic> hybridization using 35S-UTP labeled ribonucleotide probes [CRH and glucocorticoid receptor (GR)] was performed as described previously (<xref rid="B17" ref-type="bibr">17</xref>, <xref rid="B18" ref-type="bibr">18</xref>). The slides were opposed to Kodak Biomax MR film (Eastman Kodak Co., Rochester, NY, USA) and developed. Autoradiographs were digitized, and relative expression of CRH and GR mRNA was determined by computer-assisted optical densitometry (analysis 3.1, Soft Imaging System GmbH, Münster, Germany). The mean of four to six measurements was calculated for each mouse.</p></sec><sec id="S2-5"><title>Statistics</title><p>Data were analyzed by analysis of variance (ANOVA) using strain (CD1 or C57BL/6J), treatment (NON-DEP or DEP), and time (basal or novelty) as fixed factors. When appropriate, <italic>post hoc</italic> Tukey test was performed. The initial analysis included sex as a factor; once it was determined that sex was not a significant factor, the data were collapsed across this variable. The level of significance was set at <italic>P</italic> < 0.05. Data are presented as mean ± SEM.</p></sec></sec><sec id="S3"><title>Results</title><sec id="S3-6"><title>Weight</title><p>Two-way ANOVA revealed main effects of strain (<italic>F</italic><sub>1,64</sub> = 141.34; <italic>p</italic> < 0.001) and treatment (<italic>F</italic><sub>1,64</sub> = 141.34; <italic>p</italic> < 0.001). C57BL/6J were lighter than CD1 mice (<italic>p</italic> < 0.001) in both treatment conditions (Table <xref ref-type="table" rid="T1">1</xref>). DEP pups were lighter than NON-DEP pups (<italic>p</italic> < 0.001 for both strains).</p><table-wrap id="T1" position="float"><label>Table 1</label><caption><p><bold>Body weight (grams) of non-deprived (NON-DEP) and 24 h deprived (DEP) pups in CD1 and C57BL/6J mice at postnatal day 5</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="left" rowspan="1" colspan="1">Strain</th><th align="center" colspan="3" rowspan="1">NON-DEP<hr/></th><th align="center" colspan="3" rowspan="1">DEP<hr/></th><th align="left" rowspan="1" colspan="1">% Change</th></tr><tr><th align="left" rowspan="1" colspan="1"/><th align="left" rowspan="1" colspan="1">Mean</th><th align="left" rowspan="1" colspan="1"><italic>N</italic></th><th align="left" rowspan="1" colspan="1">SEM</th><th align="left" rowspan="1" colspan="1">Mean</th><th align="left" rowspan="1" colspan="1"><italic>N</italic></th><th align="left" rowspan="1" colspan="1">SEM</th><th align="left" rowspan="1" colspan="1"/></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">CD1</td><td align="left" rowspan="1" colspan="1">3.15</td><td align="left" rowspan="1" colspan="1">16</td><td align="left" rowspan="1" colspan="1">0.06</td><td align="left" rowspan="1" colspan="1">2.51<sup>#</sup></td><td align="left" rowspan="1" colspan="1">16</td><td align="left" rowspan="1" colspan="1">0.08</td><td align="left" rowspan="1" colspan="1">↓20</td></tr><tr><td align="left" rowspan="1" colspan="1">C57BL/6J</td><td align="left" rowspan="1" colspan="1">2.51<sup>$</sup></td><td align="left" rowspan="1" colspan="1">16</td><td align="left" rowspan="1" colspan="1">0.10</td><td align="left" rowspan="1" colspan="1">1.78<sup>#$</sup></td><td align="left" rowspan="1" colspan="1">17</td><td align="left" rowspan="1" colspan="1">0.11</td><td align="left" rowspan="1" colspan="1">↓29</td></tr></tbody></table><table-wrap-foot><p><italic>Data represent mean ± SEM. # vs. NON-DEP, $ vs. CD1</italic>.</p><p><italic>Significance level was set at <italic>p</italic> = 0.05</italic>.</p></table-wrap-foot></table-wrap></sec><sec id="S3-7"><title>ACTH</title><p>Three-way ANOVA revealed main effects of strain (<italic>F</italic><sub>1,42</sub> = 10.79; <italic>p</italic> < 0.001), treatment (<italic>F</italic><sub>1,42</sub> = 53.65; <italic>p</italic> < 0.001), and interaction of treatment and time (<italic>F</italic><sub>1,42</sub> = 4.41; <italic>p</italic> = 0.043) (Figure <xref ref-type="fig" rid="F1">1</xref>A). Strain differences were found at NON-DEP basal levels (<italic>p</italic> = 0.005). Novelty exposure increased ACTH levels in CD1 (↑41%, <italic>p</italic> = 0.025) but not in C57BL/6J mice. After 24 h MD, ACTH basal levels were elevated (↑156%, <italic>p</italic> = 0.001 for CD1; ↑100%, <italic>p</italic> = 0.006 for C57BL/6J). Subsequent novelty exposure did not produce further increase in ACTH in either CD1 or C57BL/6J mice, while in both strains ACTH levels were higher than the respective NON-DEP levels (<italic>p</italic> = 0.010 for CD1, <italic>p</italic> = 0.040 for C57BL/6J).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>ACTH [(A); picogram/milliliter], corticosterone [(B); CORT in nanogram/milliliter] blood plasma levels, and their ratio [(C); ACTH/CORT] of non-deprived (NON-DEP) and 24 h deprived (DEP) pups measured at basal conditions (basal; white bars) or after 30 min of novelty exposure (novelty; black bars) at postnatal day (pnd) 5</bold>. Data represent mean ± SEM. * vs. basal, # vs. NON-DEP, $ vs. CD1. Significance level was set at <italic>p</italic> = 0.05.</p></caption><graphic xlink:href="fendo-05-00190-g001"/></fig></sec><sec id="S3-8"><title>Corticosterone</title><p>Three-way ANOVA revealed main effects of strain (<italic>F</italic><sub>1,59</sub> = 59.86; <italic>p</italic> < 0.001), treatment (<italic>F</italic><sub>1,59</sub> = 248.76; <italic>p</italic> < 0.001), interaction strain and treatment (<italic>F</italic><sub>1,59</sub> = 83.52; <italic>p</italic> < 0.001), interaction strain and time (<italic>F</italic><sub>1,59</sub> = 6.38; <italic>p</italic> = 0.015), and interaction of strain, treatment, and time (<italic>F</italic><sub>1,59</sub> = 4.50; <italic>p</italic> = 0.039) (Figure <xref ref-type="fig" rid="F1">1</xref>B). Novelty exposure increased CORT levels in CD1 (↑50%, <italic>p</italic> = 0.002) but not in C57BL/6J mice. After 24 h MD, CORT basal levels were elevated in both strains (↑191%, <italic>p</italic> < 0.001 for CD1; ↑4099%, <italic>p</italic> < 0.001 for C57BL/6J). Subsequent novelty exposure further increased CORT only in CD1 mice (additional ↑167%, <italic>p</italic> < 0.001), while in both strains CORT levels were higher than the respective NON-DEP levels (<italic>p</italic> < 0.001). Strain differences were found at all four conditions: C57BL/6J CORT levels being lower than in CD1 at NON-DEP conditions (<italic>p</italic> < 0.001 for both basal and novelty), and higher at DEP conditions (for basal: <italic>p</italic> < 0.001, for novelty: <italic>p</italic> = 0.003).</p></sec><sec id="S3-9"><title>ACTH/CORT ratio</title><p>Three-way ANOVA revealed main effects of strain (<italic>F</italic><sub>1,40</sub> = 126.05; <italic>p</italic> < 0.001), treatment (<italic>F</italic><sub>1,40</sub> = 290.46; <italic>p</italic> < 0.001), time (<italic>F</italic><sub>1,40</sub> = 6.24; <italic>p</italic> = 0.018), interaction strain and treatment (<italic>F</italic><sub>1,40</sub> = 196.03; <italic>p</italic> < 0.001), and interaction of strain, treatment, and time (<italic>F</italic><sub>1,40</sub> = 6.12; <italic>p</italic> = 0.019) (Figure <xref ref-type="fig" rid="F1">1</xref>C). At NON-DEP basal conditions, C57BL/6J displayed much higher ACTH/CORT than CD1 mice (↑393%, <italic>p</italic> < 0.001). Novelty exposure decreased the ratio in C57BL/6J (↓19%, <italic>p</italic> = 0.048) but not in CD1 mice. After 24 h MD, ACTH/CORT ratio decreased in C57BL/6J (↓96%, <italic>p</italic> < 0.001) but not in CD1 mice in such an extent that the C57BL/6J displayed even less ratio than CD1 mice (<italic>p</italic> = 0.029). For both strains, ACTH/CORT ratios after subsequent novelty exposure were lower than the respective NON-DEP levels (CD1: <italic>p</italic> = 0.004, C57BL/6J: <italic>p</italic> < 0.001).</p></sec><sec id="S3-10"><title>CRH mRNA expression in the PVN</title><p>Two-way ANOVA revealed main effects of treatment (<italic>F</italic><sub>1,30</sub> = 5.41; <italic>p</italic> = 0.028) (Figure <xref ref-type="fig" rid="F2">2</xref>A). Twenty-four hours of MD downregulated CRH mRNA (<italic>p</italic> = 0.036) in CD1 mice but not in C57BL/6J.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>CRH (A) and GR (B) mRNA expression in the paraventricular nucleus of the hypothalamus measured in non-deprived (NON-DEP) and deprived (DEP) mice at postnatal day (pnd) 5</bold>. Data represent mean ± SEM. # vs. NON-DEP, $ vs. CD1. Significance level was set at<italic>p</italic> = 0.05.</p></caption><graphic xlink:href="fendo-05-00190-g002"/></fig></sec><sec id="S3-11"><title>GR mRNA expression in the PVN</title><p>Two-way ANOVA revealed main effects strain (<italic>F</italic><sub>1,27</sub> = 10.77; <italic>p</italic> = 0.003), treatment (<italic>F</italic><sub>1,27</sub> = 17.97; <italic>p</italic> < 0.001), and interaction of strain and treatment (<italic>F</italic><sub>1,27</sub> = 5.02; <italic>p</italic> = 0.035) (Figure <xref ref-type="fig" rid="F2">2</xref>B). At basal conditions, C57BL/6J displayed higher levels of GR mRNA than CD1 mice (<italic>p</italic> = 0.002). Twenty-four hours of MD downregulated GR mRNA in C57BL/6J (<italic>p</italic> < 0.001).</p></sec><sec id="S3-12"><title>GR mRNA expression in pituitary (data not shown)</title><p>There were no main effects of strain or treatment on GR mRNA in pituitary.</p></sec></sec><sec sec-type="discussion" id="S4"><title>Discussion</title><p>Our data show that the two mouse strains, CD1 and C57BL/6J mice, differ in the neonatal HPA-axis activity at basal conditions as well as after a 24 h MD period.</p><p>Regarding basal HPA-axis activity, C57BL/6J displayed higher ACTH and lower CORT than CD1 mice, indicating lower basal adrenal sensitivity to ACTH as reflected by a higher ACTH/CORT ratio. Additionally, basal GR mRNA expression in the PVN is higher than in CD1 mice. We propose that this increased GR mRNA expression might be a result of the lower CORT production. The higher GR mRNA is not likely to be an indication of stronger negative feedback capacity because there was no strain difference in basal CRH mRNA expression in the PVN or GR mRNA expression in the pituitary. Exposure of NON-DEP pups to novelty resulted in a subtle statistically significant rise in both ACTH and CORT in CD1 mice only. This finding underlines strain-dependent effects and confirms that the SHRP is a period of stress-hypo-responsiveness (<xref rid="B3" ref-type="bibr">3</xref>).</p><p>Maternal deprivation elicited in both strains the expected increase of ACTH (<xref rid="B35" ref-type="bibr">35</xref>) and CORT (<xref rid="B35" ref-type="bibr">35</xref>, <xref rid="B36" ref-type="bibr">36</xref>). ACTH rose at a similar extent in both strains. CORT levels were dramatically increased in C57BL/6J compared to a more moderate increase in CD1 pups. Previous findings in rats showed that during the time-course of 24 h maternal separation, adrenal sensitivity to stress increased (<xref rid="B37" ref-type="bibr">37</xref>) through increases in melanocortin type 2 receptors for ACTH (<xref rid="B16" ref-type="bibr">16</xref>) or other mechanisms (<xref rid="B38" ref-type="bibr">38</xref>) in a strain-dependent manner (<xref rid="B16" ref-type="bibr">16</xref>). The decrease in ACTH/CORT ratio in C57BL/6J compared to CD1 pups (from higher ACTH/lower CORT to comparable ACTH/higher CORT) indicates that, C57BL/6J after MD are no longer less sensitive to ACTH than CD1 mice at the adrenal level, but actually they display increased adrenal sensitivity compared to CD1 mice. In that, CORT secretion may be influenced also by factors other than ACTH, direct measures of neonatal adrenal sensitivity to ACTH need to be undertaken in future experiments. Only CD1 mice displayed a CORT response to novelty stress after MD. The absence of an additional novelty-induced CORT increase in C57BL/6J might be related to a ceiling effect in their steroidogenic capacity.</p><p>It is interesting that C57BL/6J do not show the expected reduction in CRH mRNA expression following MD (<xref rid="B35" ref-type="bibr">35</xref>, <xref rid="B39" ref-type="bibr">39</xref>) that was seen in the CD1 pups. This might be associated with the reduction in GR mRNA expression in the PVN and, thus, with potentially less efficient negative feedback actions of CORT at the cells that produce and release CRH. This might be an indication that, in C57BL/6J, MD causes a greater disruption of SHRP, which is characterized by enhanced negative feedback (<xref rid="B40" ref-type="bibr">40</xref>).</p><p>Another contributing factor to the strain differences here might be the transcortin levels and ultimately the free (biologically active) CORT, which is the HPA-axis feedback signal. RIA does not distinguish between free and transcortin-bound cortisol. Transcortin levels are low during SHRP (<xref rid="B41" ref-type="bibr">41</xref>) and strain differences are possible. Peripheral and central metabolic factors (e.g., blood glucose, arcuate nucleus NPY) can also mediate the activation of the HPA-axis induced by maternal separation (<xref rid="B42" ref-type="bibr">42</xref>, <xref rid="B43" ref-type="bibr">43</xref>). Indeed, in terms of body weight changes, MD caused the greatest metabolic challenge in C57BL/6J pups, which also displayed the highest activation of the HPA-axis expressed by CORT. Other factors not related to feeding might be also involved. Actually feeding is more related with the adrenal CORT secretion and tactile stimulation more related to pituitary ACTH release (<xref rid="B1" ref-type="bibr">1</xref>).</p><p>We have to acknowledge some limitations of the study. Pre-weaning pups from small litters (<5 pups) have higher body weight and higher basal CORT levels than pups from large litters (>15 pups) (<xref rid="B44" ref-type="bibr">44</xref>). The C57BL/6J litters are naturally smaller in size than the CD1 litters. This has created an unavoidable, without cross-fostering, confound that might have interfered with the strain differences reported. We opted for an equal sex-ratio (1:1) that removed the sex-ratio bias. Nevertheless, the litter-size difference between the strains was small (two pups) and did not seem to have a noticeable effect; in this experiment, the pups of the C57BL/6J strain (with the smaller litter size of six pups) displayed lower body weight and lower basal CORT than the pups of the CD1 strain (with the larger litter size of eight pups). Future studies could illuminate the role of litter-size, but also of the basal mother–pup interactions and other related epigenetically mediated mechanisms (<xref rid="B45" ref-type="bibr">45</xref>) on the neonatal basal and post-MD HPA-axis activity.</p><p>Specific genetic contributions could be clarified in the future with the use of transgenic mice, but the strain differences in immediate effects of MD observed, here, in mice and, previously, in rats (<xref rid="B16" ref-type="bibr">16</xref>) emphasize the importance of genetic background on the effects of early maternal environment on the development of the stress system. Late-life consequences may also depend on genetic background, but this remains to be tested.</p></sec><sec id="S5"><title>Conflict of Interest Statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec>
Thyroid Hormone Function in the Rat Testis	<p>Thyroid hormones are emerging regulators of testicular function since Sertoli, germ, and Leydig cells are found to express thyroid hormone receptors (TRs). These testicular cells also express deiodinases, which are capable of converting the pro-hormone T4 to the active thyroid hormone T3, or inactivating T3 or T4 to a non-biologically active form. Furthermore, thyroid hormone transporters are also found in the testis. Thus, the testis is equipped with the transporters and the enzymes necessary to maintain the optimal level of thyroid hormone in the seminiferous epithelium, as well as the specific TRs to execute thyroid hormone action in response to different stages of the epithelial cycle of spermatogenesis. Studies using genetic models and/or goitrogens (e.g., propylthiouracil) have illustrated a tight physiological relationship between thyroid hormone and testicular function, in particular, Sertoli cell differentiation status, mitotic activity, gap junction function, and blood–testis barrier assembly. These findings are briefly summarized and discussed herein.</p>	<contrib contrib-type="author"><name><surname>Gao</surname><given-names>Ying</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/189821"/></contrib><contrib contrib-type="author"><name><surname>Lee</surname><given-names>Will M.</given-names></name><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/190030"/></contrib><contrib contrib-type="author"><name><surname>Cheng</surname><given-names>C. Yan</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="corresp" rid="cor1">*</xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/184084"/></contrib>	Frontiers in Endocrinology	<sec sec-type="intro" id="S1"><title>Introduction</title><p>Thyroid hormones play a crucial role in regulating development, differentiation, and metabolism in multiple mammalian tissues. Testis was regarded as a thyroid hormone unresponsive organ for many years. In the past two decades, however, mounting evidence has emerged demonstrating the presence of functional thyroid hormone receptors (TRs) in the testis (<xref rid="B1" ref-type="bibr">1</xref>, <xref rid="B2" ref-type="bibr">2</xref>). These findings illustrate that thyroid hormones likely play an important role in testis function. Studies have demonstrated that thyroid hormones most notably T3 (3,5,3′-tri-iodothyronine) regulates Sertoli cell proliferation and differentiation during testis development including the assembly of the blood–testis barrier (BTB) (<xref rid="B3" ref-type="bibr">3</xref>–<xref rid="B5" ref-type="bibr">5</xref>). Moreover, it also induces Leydig cell differentiation and stimulates steroidogenesis in the rat testis (<xref rid="B6" ref-type="bibr">6</xref>). Several idothyronine deiodinases and thyroid hormone transporters have been identified in the testis (<xref rid="B7" ref-type="bibr">7</xref>–<xref rid="B10" ref-type="bibr">10</xref>), illustrating that these enzymes and transporters necessary to maintain the homeostasis of thyroid hormone are present in the testis. It is generally accepted that thyroid hormone acts as an important regulator in testis development. However, few studies focused on the role of thyroid hormone in regulating spermatogenesis in adult testis. Studies in recent years have suggested that altered thyroid status in adult males is associated with abnormal spermatogenesis, reducing sexual activity and impeding fertility (<xref rid="B11" ref-type="bibr">11</xref>–<xref rid="B15" ref-type="bibr">15</xref>), illustrating the crucial relationship between thyroid hormones and maturation status of Sertoli cells. In fact, TRα1 is a reliable marker of Sertoli cell maturation because its expression is considerably down-regulated in adult testes, and continual expression of TRα1 illustrates delayed Sertoli cell maturation in adult mice (<xref rid="B16" ref-type="bibr">16</xref>, <xref rid="B17" ref-type="bibr">17</xref>). There are also reports in recent years that thyroid hormone is crucial to maintain gap junction (GJ) and BTB function, as well as BTB maturation during postnatal development. Our goal in this mini-review is to focus on the role of thyroid hormone and junction dynamics, in particular, the BTB function during spermatogenesis, providing an update on the current status of research in this area. We also highlight research areas that deserve attention in future studies. We first provide a brief outline regarding the role of thyroid hormone in testis development and testicular function since this information is closely related to the emerging field in which thyroid is a major player in junction dynamics during spermatogenesis.</p></sec><sec id="S2"><title>Thyroid Hormone Action</title><p>Thyroxin (3, 5, 3′, 5′-tetraiodothyronine, T4) is the major form of thyroid hormones released by the thyroid gland into the systemic circulation. Thyroxin, however, is a pro-hormone, which must be converted to tri-iodothyronine (3, 5, 3′-tri-iodothyronine, T3), which takes place primarily in the liver and kidney. T3 is the bioactive form of thyroid hormone that has high affinity for nuclear TRs (<xref rid="B18" ref-type="bibr">18</xref>, <xref rid="B19" ref-type="bibr">19</xref>). A small amount of T3 and reverse T3 (rT3), however, is also produced by the thyroid gland (<xref rid="B20" ref-type="bibr">20</xref>). T3 mediates its effects via genomic and also non-genomic pathways. For the classical genomic pathway, T3 mediates its effects by TRs. In the nucleus, TRs usually forms heterodimers with retinoid X receptor (RXR), and this complex further binds to thyroid response elements (TRE) in the promoter region of a target gene to regulate gene transcription (<xref rid="B21" ref-type="bibr">21</xref>). In addition, thyroid hormone also regulates the release of thyrotrophin-releasing hormone (TRH) by the hypothalamus and of thyroid-stimulating hormone (TSH) by the pituitary gland (<xref rid="B21" ref-type="bibr">21</xref>) to serve as a feedback loop in the hypothalamic–pituitary–thyroid axis to maintain the physiological level of thyroid hormone in the systemic circulation. In contrast to the genomic pathway, which has a relatively long response time, ranging from hours to days, non-genomic pathways have short latency and are not affected by transcription or translation inhibitors. Thyroid hormone binds to the binding elements such as integrin αvβ3 located at the plasma membrane or within a cell to exert its effects. These non-gemonic effects include the regulation of ion influxes, kinase signaling pathways, amino acid accumulation, extracellular nucleotide levels, and vimentin phosphorylation via non-receptor protein kinases downstream (<xref rid="B10" ref-type="bibr">10</xref>). While T4 is a pro-hormone, it can bind to TRs but with low affinity, and the T4 liganded-TR is less stable versus the T3-liganded-TR. Nonetheless, T4 serves as an agonist to TRs at appropriate concentration (<xref rid="B22" ref-type="bibr">22</xref>), which also depends on receptor isoform and the presence of cellular cofactors (e.g., thyroid hormone receptor-associated protein 220, TRAP200) (<xref rid="B23" ref-type="bibr">23</xref>). In addition to the genomic pathway, T4 also initiates rapid non-genomic response by binding to integrin αvβ3 in the plasma membrane, leading to an increase in cellular amino acid accumulation (<xref rid="B24" ref-type="bibr">24</xref>–<xref rid="B26" ref-type="bibr">26</xref>). Collectively, these findings illustrate T4 has a limited functional role in mammalian cells.</p></sec><sec id="S3"><title>Thyroid Hormone Receptors in Testicular Cells</title><p>Thyroid hormone receptors (TRs) are able to mediate the effects of thyroid hormone via classical genomic pathway via two genes, <italic>THRA</italic> (TRα) and <italic>THRB</italic> (TRβ). Alternative splicing gives rise to several TR isoforms: TRα1, α2, α3, and β1, β2, β3 (<xref rid="B21" ref-type="bibr">21</xref>). It is known that TRβ2 is restricted to the anterior pituitary and hypothalamus (<xref rid="B27" ref-type="bibr">27</xref>), and TRβ3 is highly expressed in liver, kidney, and lung (<xref rid="B28" ref-type="bibr">28</xref>). Although TRα2 and TRα3 mRNA are detected in Sertoli cells, these receptors do not have T3-binding capacity (<xref rid="B2" ref-type="bibr">2</xref>, <xref rid="B29" ref-type="bibr">29</xref>–<xref rid="B31" ref-type="bibr">31</xref>). But they may exert dominant negative effects by binding to TRE to suppress gene transcription (<xref rid="B32" ref-type="bibr">32</xref>, <xref rid="B33" ref-type="bibr">33</xref>). More important, TRα1 and TRβ1 are the functional TR isoforms by mediating T3 signaling, and also T4 but to a lesser extent. Both TRα1 and TRβ1 were shown to be expressed by Sertoli and germ cells throughout development in the rat testis (<xref rid="B1" ref-type="bibr">1</xref>). These two TR isoforms are abundantly expressed in neonatal Sertoli cells, suggesting that Sertoli cells might be the target cell type for T3 in the developing testis. A study using TRαKO or TRβKO mice has demonstrated that TRα1 is the crucial TR isoform, which mediates T3 effects in neonatal Sertoli cells (<xref rid="B34" ref-type="bibr">34</xref>). In fact, TRα serves as a reliable marker of Sertoli cell maturity. Persistent expression of TRα in adult testes is a reliable indicator of undifferentiated Sertoli cells, such as in neonatal mice (<xref rid="B4" ref-type="bibr">4</xref>, <xref rid="B35" ref-type="bibr">35</xref>, <xref rid="B36" ref-type="bibr">36</xref>) and in mice following deletion of A-kinase anchoring protein 9 (AKAP9) that impedes Sertoli cell differentiation (<xref rid="B37" ref-type="bibr">37</xref>). Recently, a transgenic model in which mice expressed a dominant negative TRα1 only in Sertoli cells was generated. By using TRα<sup>AMI</sup>-SC mice, T3 was shown to be a potent regulator to arrest Sertoli cell mitotic proliferation, which is mediated by an activation of TRα1 via the Cdk4/JunD/c-myc pathway (<xref rid="B38" ref-type="bibr">38</xref>). This finding is consistent with earlier reports that neonatal hypothyroidism induced in mice or rats by neonatal treatment with a goitrogen leads to an increase in Sertoli cell number and daily sperm production, concomitant with an increase in testis weight, due to a failure of Sertoli cell differentiation, making them mitotically active (<xref rid="B4" ref-type="bibr">4</xref>, <xref rid="B36" ref-type="bibr">36</xref>, <xref rid="B39" ref-type="bibr">39</xref>, <xref rid="B40" ref-type="bibr">40</xref>). Also, in rodents when Sertoli cells cease to divide at age ~15- to 17-day postpartum (dpp) to become fully differentiated, this event coincides with a surge in T3 that peaks in the systemic circulation (<xref rid="B41" ref-type="bibr">41</xref>), illustrating a reciprocal relationship between T3 and Sertoli cell mitotic activity and differentiation status. Collectively, these findings illustrate T3 is a regulator of Sertoli cell mitotic function and differentiation status in the testis. Furthermore, TRs are detected in germ cells by immunohistochemistry (<xref rid="B1" ref-type="bibr">1</xref>). For instance, TRα1 is expressed by spermatogenic cells from intermediate spermatogonia to mid-cycle pachytene spermatocytes (<xref rid="B1" ref-type="bibr">1</xref>), suggesting that T3 may also play a role in germ cell meiotic development. Additionally, TRs are also expressed by Leydig cells in the interstitial compartment of immature testes (<xref rid="B1" ref-type="bibr">1</xref>). In fact, it was reported that Leydig cell differentiation and steroidogenesis in postnatal rat testes were affected by T3 (<xref rid="B42" ref-type="bibr">42</xref>).</p></sec><sec id="S4"><title>Iodothyronine Deiodinases in Testis</title><p>T4 released by the thyroid gland is the pro-hormone, which is converted to bioactive T3 by deiodination of T4 catalyzed by type 1 and type 2 deiodinase (D1 and D2; deiodinase is also known as iodide peroxidase), usually takes place in the liver and kidney (<xref rid="B43" ref-type="bibr">43</xref>) (Figure <xref ref-type="fig" rid="F1">1</xref>). Both the active hormone T3 and pro-hormone T4, however, can also be inactivated via deiodination by type 3 deiodinase (D3), converting into biologically inactive metabolites 3,3′-diiodothyronine (T2) and 3,3′,5′-tri-iodothyronine (reverse T3 or rT3) (<xref rid="B43" ref-type="bibr">43</xref>, <xref rid="B44" ref-type="bibr">44</xref>), respectively (Figure <xref ref-type="fig" rid="F1">1</xref>). Thus, unlike D1 and D2 that activates thyroid hormones, D3 is an inactivator of thyroid hormones, serving as a modulator of intracellular thyroid hormone levels and action. All three deiodinases are detected in developing and adult testes (<xref rid="B45" ref-type="bibr">45</xref>). In developing testis, D3 is the predominant deiodinase and then its activity declines in adult testes (<xref rid="B45" ref-type="bibr">45</xref>), whereas D2 is the predominant activating deiodinase in the testis (<xref rid="B42" ref-type="bibr">42</xref>). D2 is abundantly expressed in elongated spermatids, whereas its expression could not be detected in Sertoli cells or other germ cells, suggesting that thyroid hormones might play a role in regulating spermatogenesis, specifically on spermiogenesis (<xref rid="B9" ref-type="bibr">9</xref>). However, the precise cellular localizations of D1 and D3 in the testis remain unclear. Earlier study has demonstrated that severe hypothyroidism may affect fertility in both sexes (<xref rid="B46" ref-type="bibr">46</xref>). Unexpectedly, mice lack either D1, D2 or both D1 and D2 are fertile and display normal serum T3 level (<xref rid="B47" ref-type="bibr">47</xref>–<xref rid="B49" ref-type="bibr">49</xref>). These findings indicate that in mice, D1 or D2 is not indispensable for maintaining serum T3 level, and D1 or D2-mediated local production of T3 is not likely to be the only source of T3 in the testis. Interestingly, knockout (KO) of D3 cause impaired fertility in mice, suggesting that D3 may play a more important physiological role in the testis (<xref rid="B50" ref-type="bibr">50</xref>). Thus, further studies are necessary to investigate the precise role of deiodinases in the testis.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>A schematic drawing to illustrate the physiological role of thyroid hormone T3 on testis function</bold>. This schematic drawing was prepared based on findings in the literature as discussed herein (see text for details). In short, T4 is the principal thyroid hormone produced by the thyroid gland and released into the systemic circulation. T4, however, is a pro-hormone, which is being activated via the action of deiodinases D1 or D2, mostly in the liver and kidney but also the testis, to form T3, the activated thyroid hormone. However, D1, D2, and D3 deiodinases are also found in the testis. The use of goitrogen [e.g., propylthiouracil (PTU)] can block the production of T4 by thyroid gland, which was used to examine the effects of thyroid hormones on testicular function. D3 deiodinase, unlike D1 and D2 deiodinases that activates T4 to T3, de-activates T4 or T3 to rT3 or T2, respectively, which are inactivated thyroid hormones, providing a crucial mechanism to regulate intracellular thyroid hormone action in cells, such as in Sertoli and/or germ cells in the testis. It is known that high level of T3 inhibits Sertoli cell proliferation and promotes Sertoli cell differentiation, whereas low level of T3 causes delayed Sertoli cell proliferation and differentiation. It is noted that at puberty (∼12 years of age) in men or by day ∼15-17 day in rodents, there is a surge in T3 level in systemic circulation, coinciding with Sertoli cell differentiation when Sertoli cells cease to divide (see text for details). T4, 3,5,3′,5′-tetraiodothyronine; T3, 3,5,3′-tri-iodothyronine; rT3, 3,3′, 5′-tri-iodothyronine; T2, 3,3′-diiodothyronine; PTU, propylthiouracil; SC, Sertoli cell; BTB, blood-testis barrier.</p></caption><graphic xlink:href="fendo-05-00188-g001"/></fig></sec><sec id="S5"><title>Thyroid Hormone Transporters in Testicular Cells</title><p>Since TRs and deiodinases are located intracellularly in mammalian tissues including the seminiferous epithelium in testes, thyroid hormones have to be transported across cell membranes before they can be activated by deiodinases, such as from T4 to T3, to mediate the effects via TRs or be inactivated, such as from T3 to T2 or T4 to rT3. While there is no specific membrane bound TRs, several membrane bound drug transporters are putative transporters of thyroid hormones that include monocarboxylate transporter (MCT) 8, MCT10, and organic anion-transporting polypeptides (OATPs) (<xref rid="B51" ref-type="bibr">51</xref>, <xref rid="B52" ref-type="bibr">52</xref>). MCT8 is a specific thyroid hormone transporter. Unlike MCT8, MCT10 not only transports thyroid hormones but also aromatic amino acid. Both of MCTs prefer T3 over T4, and MCT10 is even more efficient than MCT8 in transporting T3 across plasma membranes (<xref rid="B53" ref-type="bibr">53</xref>). However, studies have shown that MCT8 KO, MCT10 KO, and MCT8/MCT10 double KO mice are all fertile in both sexes, supporting the notion that other thyroid hormone transporters may compensate the loss of MCT8 and MCT10 (<xref rid="B54" ref-type="bibr">54</xref>). Additionally, OATPs are able to transport steroid conjugates, prostaglandins, bile acids, drugs, and thyroid hormones (<xref rid="B55" ref-type="bibr">55</xref>). Several OATP family members have been detected in the testis (<xref rid="B56" ref-type="bibr">56</xref>). For instance, OATP-F, a homolog of OATP1C1, displaying high affinity for T4 and rT3, has been detected in human Leydig cells (<xref rid="B57" ref-type="bibr">57</xref>). OATP6A1, originally identified as a cancer/testis antigen also called SLCO6A1, is predominantly expressed in normal testes (<xref rid="B58" ref-type="bibr">58</xref>). In addition, two spliced variants of OATP3A1 called OATP3A1-V1 and OATP3A1-V2 have also been detected in germ cells and Sertoli cells, respectively (<xref rid="B59" ref-type="bibr">59</xref>). The rat gonad-specific transporter (GST)-1 and GST-2, which are members of OATPs family are highly expressed in Sertoli cells, spermatogonia, and Leydig cells (<xref rid="B60" ref-type="bibr">60</xref>), which may also be involved in T3 and T4 transport across the plasma membrane. A recent study has demonstrated that MCT8 and OATP1C1 are crucial to maintain the thyroid hormone homeostasis in the mouse brain (<xref rid="B61" ref-type="bibr">61</xref>), and OATP14 is a high affinity transporter for T4 at the blood–brain barrier (<xref rid="B62" ref-type="bibr">62</xref>). Much research is needed to delineate the physiological role of OATPs and MCTs in regulating thyroid hormone transport across the BTB.</p></sec><sec id="S6"><title>Effects of Thyroid Hormones on Sertoli Cell Proliferation, Differentiation, and BTB Assembly</title><p>Propylthiouracil is a goitrogen that inhibits the enzyme thyroperoxidase by blocking the production of T4 from thyroglobulin in the thyroid, causing hypothyroidism. It also inhibits 5′-deiodinase that converts T4 to T3. Thus, PTU is a widely used thiouracil-derived drug used to treat hyperthyroidism (<xref rid="B63" ref-type="bibr">63</xref>, <xref rid="B64" ref-type="bibr">64</xref>). PTU-induced neonatal hypothyroidism by treating neonatal rats from birth was shown to increase rat testis weight and daily sperm production of up to 80 and 140%, respectively (<xref rid="B35" ref-type="bibr">35</xref>, <xref rid="B36" ref-type="bibr">36</xref>). Further studies demonstrated that this was the result of Sertoli cell proliferation and a delay of Sertoli cell maturation (<xref rid="B5" ref-type="bibr">5</xref>). Furthermore, the Sertoli cell BTB failed to assemble by 15–25 dpp even though some tight junction (TJ) structures were detected by electron microscopy at these ages, but extensive network of TJ ultrastructure and basal ectoplasmic specialization (ES) analogous to age-matched control rats was not found in these rats treated with PTU from birth to age 25 dpp (<xref rid="B5" ref-type="bibr">5</xref>). Conversely, neonatal hyperthyroidism was found to stimulate Sertoli cell differentiation, rendering Sertoli cells ceased to proliferate by age 12 versus ∼15–17 dpp in normal rats, thereby reducing the testis weight in adult animals at age 100 dpp by almost 50% (<xref rid="B3" ref-type="bibr">3</xref>). These findings suggest that thyroid hormone regulates testis development by modulating Sertoli cells mitotic activity, differentiation status, and the BTB assembly. Table <xref ref-type="table" rid="T1">1</xref> summarizes some of the known effects of thyroid hormone T3 on Sertoli and Leydig cell function in the testis.</p><table-wrap id="T1" position="float"><label>Table 1</label><caption><p><bold>Effect of thyroid hormone T3 on testes</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th valign="top" align="left" rowspan="1" colspan="1">Cell type</th><th valign="top" align="left" rowspan="1" colspan="1">Effects: stimulation (+), inhibition (−)</th><th valign="top" align="left" rowspan="1" colspan="1">Reference</th></tr></thead><tbody><tr><td valign="top" align="left" rowspan="1" colspan="1">Sertoli cell</td><td valign="top" align="left" rowspan="1" colspan="1">Proliferation (−)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B3" ref-type="bibr">3</xref>, <xref rid="B4" ref-type="bibr">4</xref>)</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">Differentiation (+)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B3" ref-type="bibr">3</xref>, <xref rid="B4" ref-type="bibr">4</xref>, <xref rid="B39" ref-type="bibr">39</xref>)</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">ABP production (−)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B65" ref-type="bibr">65</xref>)</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">AR (+)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B30" ref-type="bibr">30</xref>)</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">Aromatase (−)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B66" ref-type="bibr">66</xref>, <xref rid="B67" ref-type="bibr">67</xref>)</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">Connexin 43 (+)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B68" ref-type="bibr">68</xref>)</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">ER (−)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B69" ref-type="bibr">69</xref>)</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">GLUT1 (+)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B70" ref-type="bibr">70</xref>)</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">IGF-1 (+)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B71" ref-type="bibr">71</xref>)</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">Inhibin (+)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B3" ref-type="bibr">3</xref>)</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">Lactate (+)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B39" ref-type="bibr">39</xref>)</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">NCAM (−)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B72" ref-type="bibr">72</xref>)</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">Nidogen (+)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B73" ref-type="bibr">73</xref>)</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">p21<sup>Cip1</sup> (+)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B74" ref-type="bibr">74</xref>, <xref rid="B75" ref-type="bibr">75</xref>)</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">p27<sup>Kip1</sup> (+)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B74" ref-type="bibr">74</xref>, <xref rid="B75" ref-type="bibr">75</xref>)</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">Testosterone metabolism aromatization (−)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B39" ref-type="bibr">39</xref>)</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">Type IV collagen (−)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B73" ref-type="bibr">73</xref>)</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">Vimentin phosphorylation (+)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B76" ref-type="bibr">76</xref>)</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">Leydig cell</td><td valign="top" align="left" rowspan="1" colspan="1">Differentiation (+)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B77" ref-type="bibr">77</xref>, <xref rid="B78" ref-type="bibr">78</xref>)</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">Steroidogenesis (+)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B79" ref-type="bibr">79</xref>)</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">StAR protein (+)</td><td valign="top" align="left" rowspan="1" colspan="1">(<xref rid="B79" ref-type="bibr">79</xref>–<xref rid="B81" ref-type="bibr">81</xref>)</td></tr></tbody></table><table-wrap-foot><p><italic>ABP, androgen binding protein; AR, androgen receptor; ER, estrogen receptor; GLUT1, glucose transporter-1; IGF-1, insulin-like growth factor-1; NCAM, neural cell adhesion molecule</italic>.</p></table-wrap-foot></table-wrap></sec><sec id="S7"><title>Thyroid Hormones, Gap Junction, and Epithelial/Endothelial Barrier Function</title><p>Gap junctions are intercellular channels, which mediate direct communication between neighboring cells. These channels allow passage of ions and small molecules, usually <1–1.5 kDa, and are involved in several physiological processes, such as cell growth, apoptosis, and differentiation (<xref rid="B82" ref-type="bibr">82</xref>–<xref rid="B85" ref-type="bibr">85</xref>). Connexin 43 (Cx43) is the predominant GJ protein in the testis (<xref rid="B84" ref-type="bibr">84</xref>, <xref rid="B86" ref-type="bibr">86</xref>), it is expressed by Sertoli cells, germ cells, as well as Leydig cells in the testis and found at the Sertoli cell–cell and Sertoli–germ cell interface (<xref rid="B87" ref-type="bibr">87</xref>, <xref rid="B88" ref-type="bibr">88</xref>). Although the Cx43 germ line KO mice died shortly after birth due to heart defects, deletion of Cx43 was shown to induce germ cell deficiency in the testis of developing embryo (<xref rid="B89" ref-type="bibr">89</xref>). Interestingly, Sertoli cell-specific Cx43 KO (SC-Cx43 KO) mice have smaller testes, and the seminiferous tubules of these KO mice contain mitotically active Sertoli cells and early spermatogonia but not any other germ cell types since spermatogonia failed to differentiate into spermatocytes beyond type A to enter meiosis (<xref rid="B90" ref-type="bibr">90</xref>). It is noteworthy that Sertoli cells of SC-Cx43 KO mice remained proliferative in adult mutant mice (<xref rid="B16" ref-type="bibr">16</xref>, <xref rid="B90" ref-type="bibr">90</xref>), analogous to the phenotypes of Sertoli cells in the goitrogen-induced hypothyroidism model. These findings also illustrate that Sertoli cell maturation is perturbed following deletion of Cx43 in these mutant mice. TRα1 mRNA expression was also found to be up-regulated by 20- and 60-dpp in the testis of SC-Cx43 KO mice versus the age-matched control (<xref rid="B16" ref-type="bibr">16</xref>). It is noted that TRα1 is abundantly expressed in the testis during neonatal period but rapidly declines in adulthood in normal rats (<xref rid="B21" ref-type="bibr">21</xref>). These findings thus illustrate an inactivation/deletion of Cx43 causes an upregulation of TRα1, which may mediate thyroid hormone action on Sertoli cell differentiation. Taken collectively, these data thus demonstrate unequivocally that Cx43 plays a crucial role in spermatogenesis and testis development, which is also involved in thyroid hormone action in the testis. In fact, studies have shown that thyroid hormone may inhibit Sertoli cell proliferation by up-regulating Cx43 expression (<xref rid="B68" ref-type="bibr">68</xref>, <xref rid="B91" ref-type="bibr">91</xref>). However, the precise mechanism remains unknown. In tumor cells, overexpression of Cx43 induces cyclin-dependent kinase inhibitor (CDKI) p27<sup>Kip1</sup> level (<xref rid="B92" ref-type="bibr">92</xref>). Consistent with this finding, <italic>in vitro</italic> studies have shown that T3 up-regulates p27<sup>Kip1</sup> and p21<sup>Cip1</sup>, which, in turn, may play a role in down-regulating Sertoli cell proliferation (<xref rid="B74" ref-type="bibr">74</xref>, <xref rid="B75" ref-type="bibr">75</xref>, <xref rid="B93" ref-type="bibr">93</xref>). It is also likely that thyroid hormone regulates Cx43 expression, which in turn induces the expression of maturation/differentiation markers p27<sup>Kip1</sup> and p21<sup>Cip1</sup> via a yet-to-be defined signaling pathway, leading to an arrest of Sertoli cell proliferation. This possibility must be carefully evaluated in future studies to define the physiological relationship between Cx43 and thyroid hormone action in the testis as well as the involving signaling molecules.</p><p>While studies using goitrogen and Sertoli cell-specific Cx43 KO models have demonstrated the physiological relationship between thyroid hormone action, Cx43-based GJ function and spermatogenesis (e.g., differentiation of spermatogonia to spermatocytes and the onset of meiosis), in particular, the impact of T3 on Sertoli cell BTB assembly, the molecular mechanism(s) underlying these observations remain unknown. An early report has demonstrated that treatment of chick with thiouracil that inhibits T3 production also delays the development of interdigitation of the lateral plasma membrane between adjacent corneal endothelial cells whereas thyroxine treatment accelerates development of endothelial cell lateral borders (<xref rid="B94" ref-type="bibr">94</xref>). These findings are physiologically important to studies in the testis since Sertoli cell cytoplasmic processes create interdigital association with different germ cell types at a Sertoli:germ cell ratio of ∼1:30-1:50 during spermatogenesis, requiring extensive interactions between Sertoli and germ cells at the plasma membranes, supporting the notion that T3 may play a role in junction dynamics in the seminiferous epithelium. It is likely that T3-mediated Cx43-based GJ function may be crucial to these events. It is logical to use the goitrogen-induced hypothyroidism model in both neonatal and adult rats to examine changes in junction dynamics at the BTB and also Sertoli–germ cell interface during spermatogenesis in future studies.</p></sec><sec id="S8"><title>Concluding Remarks and Future Perspectives</title><p>Herein, we provide an update on the role of T3 on Sertoli cell maturation, differentiation and BTB assembly during development. Figure <xref ref-type="fig" rid="F1">1</xref> summarizes the latest findings regarding the role of thyroid hormones in Sertoli cell proliferation, differentiation, and BTB assembly based on several reports in the last two decades investigating the role of thyroid hormones on testis function. However, there is a lack of data regarding the mechanism(s) by which T3 affects BTB developing at ~15- to 21-dpp in rats. Does this involve changes in the spatiotemporal expression, localization, and/or intrinsic activity of actin regulatory proteins, such as Arp2/3 (actin-related protein 2/3) complex (a branched actin polymerization inducing protein), palladin (an actin bundling/cross-linking protein), Eps8 (epidermal growth factor receptor pathway substrate 8, an actin barbed end capping, and bundling protein), which affect organization of actin microfilaments at the BTB? Does this involve changes in the endocytic vesicle-mediated protein trafficking, thereby impeding localization of adhesion protein complexes at the Sertoli cell–cell interface? What is the effect on the actin microfilament organization at the ectoplasmic specialization following knockdown of D1, D2, and/or D3 in Sertoli cells? Many of these questions will need to be addressed before we can gain some insightful information on the role of thyroid hormone on junction dynamics in the testis. Furthermore, selenium, a key element to maintain spermatogenesis and male fertility (<xref rid="B95" ref-type="bibr">95</xref>), is the prosthetic group of deiodinases, as such selenocysteine that plays an important role in determining the free circulating level of T3 in the mammalian body. As such, the involvement of selenium in thyroid hormone action should also be considered in future studies.</p></sec><sec id="S9"><title>Conflict of Interest Statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec>
Role of Protein Kinase C in Podocytes and Development of Glomerular Damage in Diabetic Nephropathy	<p>The early glomerular changes in diabetes include a podocyte phenotype with loss of slit diaphragm proteins, changes in the actin cytoskeleton and foot process architecture. This review focuses on the role of the protein kinase C (PKC) family in podocytes and points out the differential roles of classical, novel, and atypical PKCs in podocytes. Some PKC isoforms are indispensable for proper glomerular development and slit diaphragm maintenance, whereas others might be harmful when activated in the diabetic milieu. Therefore, some might be interesting treatment targets in the early phase of diabetes.</p>	<contrib contrib-type="author"><name><surname>Teng</surname><given-names>Beina</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/175636"/></contrib><contrib contrib-type="author"><name><surname>Duong</surname><given-names>Michelle</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/187137"/></contrib><contrib contrib-type="author"><name><surname>Tossidou</surname><given-names>Irini</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/191195"/></contrib><contrib contrib-type="author"><name><surname>Yu</surname><given-names>Xuejiao</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/191209"/></contrib><contrib contrib-type="author"><name><surname>Schiffer</surname><given-names>Mario</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="corresp" rid="cor1">*</xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/169136"/></contrib>	Frontiers in Endocrinology	<sec sec-type="intro" id="S1"><title>Introduction</title><p>Diabetic nephropathy (DN) is the most common cause of the end-stage renal disease (ESRD) (<xref rid="B1" ref-type="bibr">1</xref>). The glomerular changes in DN are characterized by excessive accumulation of extracellular matrix (ECM) with thickening of glomerular and tubular basement membranes and mesangial expansion, which ultimately progress primarily to glomerulosclerosis and secondarily to tubulointerstitial fibrosis. The various cell types involved include glomerular endothelial cells, mesangial cells, podocytes, and tubular epithelia, which are all targets of hyperglycemic injury. However, accumulating evidence suggests that the extent of injury and loss of podocytes is a major prognostic determinant in both, type I and type II DN (<xref rid="B2" ref-type="bibr">2</xref>–<xref rid="B6" ref-type="bibr">6</xref>). As the terminally differentiated podocytes are believed to play a critical role in maintaining the integrity of the glomerular filtration barrier, effaced podocytes may contribute to the development of albuminuria, a hallmark of DN. Although primarily structure proteins were thought to be the key elements that compose the slit diaphragm initially, it has become clear that the slit diaphragm protein complex is a highly dynamic functional protein complex and is able to initiate cascades of signaling pathways that affect podocyte function (<xref rid="B7" ref-type="bibr">7</xref>). More recent data indicate that podocytes express receptors for many circulating hormones and growth factors, which also suggest that a more complex cross-talk between the kidney and other organs affected by diabetes may occur in health and disease (<xref rid="B8" ref-type="bibr">8</xref>).</p><p>Among various signaling kinases, the protein kinase C (PKC) family seems to play a critical role in the pathogenesis of DN (<xref rid="B9" ref-type="bibr">9</xref>). The activation of PKCs in the kidney is a well-known pathway of the diabetic milieu. The PKC family is involved in a variety of signal transduction pathways, cell proliferation, differentiation, cell cycle, and apoptosis. However, the role of PKCs on podocytes in DN has not yet been fully defined. This present review will give an overview of the general role of PKCs and summarize the recent research into the regulatory role of PKCs in podocytes under diabetic conditions.</p></sec><sec id="S2"><title>PKC Subfamilies and Isoforms</title><p>All PKC isoforms contain a highly conserved catalytic domain and a regulatory domain. The catalytic domain consists of several motifs and is essential for the ATP/substrate-binding and catalysis, whereas the N-terminal regulatory domain contains an auto inhibitory pseudo-substrate domain and two discrete membrane targeting modules, C1 and C2. The sequence of pseudo-substrate contains an alanine in place of the serine/threonine phospho-acceptor site, but otherwise resembles a PKC substrate (<xref rid="B10" ref-type="bibr">10</xref>).</p><p>Protein kinase C isoforms are subdivided into three subfamilies based on differences of structure in their N-terminal regulatory domain. The isoforms α, βI, βII, and γ belong to conventional PKCs (cPKCs). The regulatory domains of cPKCs contain a C1 domain that functions as diacylglycerol (DAG)/phorbol 12-myristat 13-acetat (PMA) binding motif and a C2 domain that binds anionic phospholipids in a calcium-dependent manner (<xref rid="B10" ref-type="bibr">10</xref>). The novel PKCs (nPKCsδ, ε, η, and θ) also have two C1 domains and a C2 domain. The nPKC C2 domains lack the critical calcium-coordinating acidic residues, which is the distinct difference between cPKC and nPKC. The nPKCs can be maximally activated by agonists that promote DAG accumulation or by PMA, but they are insensitive to calcium. Atypical PKCs (aPKCs, ζ, and λ/ι) lack a calcium-sensitive C2 domain; however, they contain an atypical C1 domain, which binds PIP3 or ceramide, but not DAG or PMA. The activity of aPKCs is primarily regulated by protein–protein interaction and phosphorylation catalyzed by phosphoinositide-dependent kinase-1 (PDK-1). Although a few PKC isoforms are expressed in a tissue-specific manner, most are ubiquitously expressed.</p></sec><sec id="S3"><title>Regulation of PKCs</title><p>Unless it is post-translationally or co-translationally phosphorylated, PKCs are incapable of being activated by DAG or other cofactors (<xref rid="B11" ref-type="bibr">11</xref>, <xref rid="B12" ref-type="bibr">12</xref>). PKC is translated as a catalytically inactive protein, is converted into an active enzyme by an initial phosphate addition and then into a mature form by further phosphorylation (<xref rid="B11" ref-type="bibr">11</xref>, <xref rid="B13" ref-type="bibr">13</xref>, <xref rid="B14" ref-type="bibr">14</xref>). In addition to phosphorylation on serine and threonine residues, PKCs also undergo phosphorylation by tyrosine kinases. PKCs are regulated by two sequential and equally critical mechanisms: phosphorylation triggered by PDK-1 and binding to DAG and/or other cofactors such as phosphatidylserine (PS) or phorbol ester (PE). Each mechanism regulates the structure, subcellular localization, and function of PKCs (<xref rid="B15" ref-type="bibr">15</xref>).</p><p>As a consequence of increased glycolytic flux, chronic hyperglycemia elevates the <italic>de novo</italic> synthesis of DAG and thus leads to an increased activation of DAG dependent classic and novel PKC isoforms in cultured bovine aortic endothelial cells and smooth muscle cells (<xref rid="B16" ref-type="bibr">16</xref>). Furthermore, high-glucose induced cellular levels lead to an increased generation of advanced glycation end products (AGEs), which initiate several signaling events by activating PKC, MAP kinase, and transcription factors such as nuclear factor-κB (NF-κB). This would increase the activity of various growth factors, such as TGF-β, and thereby alter expression of ECM proteins (<xref rid="B17" ref-type="bibr">17</xref>). In addition, under high-glucose conditions PKC is activated by higher concentrations of reactive oxygen species (ROS) generated following AGE:RAGE (AGE receptor) interactions (<xref rid="B15" ref-type="bibr">15</xref>, <xref rid="B18" ref-type="bibr">18</xref>). In turn, the ROS are generated via NADPH-oxidase activated by PKC, ROS/PKC therefore can act in a cyclical manner to activate one another (<xref rid="B19" ref-type="bibr">19</xref>).</p></sec><sec id="S4"><title>Conventional PKCs in Diabetic Nephropathy</title><p>Among all the PKC isoforms, the role of PKCα in the pathogenesis of DN has been investigated intensively, and several studies have demonstrated that PKCα-deficient mice show a better outcome after streptozotocin (STZ) induced diabetes with less proteinuria and preserved nephrin expression (<xref rid="B20" ref-type="bibr">20</xref>, <xref rid="B21" ref-type="bibr">21</xref>). Studies from our group underline the involvement of PKCα in proteinuria development in DN. The expression of PKCα in podocytes of patients with DN was increased. Mice were treated after STZ-induced diabetes with the synthetic PKCα inhibitor (GÖ6976), which prevented proteinuria development and led to preserved nephrin expression. Furthermore, we could show a central role for PKCα in endocytosis of the slit diaphragm component nephrin (<xref rid="B22" ref-type="bibr">22</xref>, <xref rid="B23" ref-type="bibr">23</xref>). Quack et al. further concluded that proteinuria of diabetic mice is as a result of increased endocytosis of nephrin, which is mediated by a complex consisting of PKCα, protein interacting with c kinase-1 (PICK1), and beta-arrestin2. They found rising glucose levels go along with increased binding of beta-arrestin to nephrin <italic>in vitro</italic> as well as in <italic>in vivo</italic>, but only with preceding PKCα phosphorylating activity on nephrin. This fact and the indispensability of PICK1 suggest PKCα and PICK1 as possible drug targets in early stages of DN (<xref rid="B23" ref-type="bibr">23</xref>). These studies show that expression of PKCα is regulated by glucose concentration in the external milieu of the podocyte and that PKCα is directly involved in the maintenance of the slit diaphragm. High levels of PKCα in podocytes led to enhanced endocytosis of nephrin and instability of the slit diaphragm.</p><p>In addition, Menne and colleagues demonstrated that the high-glucose induced downregulation of nephrin is probably caused by the PKCα mediated reduction of the transcriptional factor Wilms tumor 1 (WT-1), which has previously been described as a directly interacting binding partner of the nephrin promoter (<xref rid="B24" ref-type="bibr">24</xref>, <xref rid="B25" ref-type="bibr">25</xref>). These findings are consistent with earlier studies, suggesting one of the PKC isoforms might be pivotal for the regulation of nephrin transcription and expression in podocytes (<xref rid="B26" ref-type="bibr">26</xref>), but not for CD2AP and Podocin as they remain at levels similar to those in non-diabetic kidneys (<xref rid="B27" ref-type="bibr">27</xref>).</p><p>Langham et al. showed that in diabetic patients treated with perindopril, an ACE inhibitor, nephrin levels preserved resembling those of non-diabetic subjects (<xref rid="B28" ref-type="bibr">28</xref>) and subsequently could be one factor contributing to the anti-proteinuric effects of ACE inhibitors. Another study suggests that the AGEs, which can be inhibited by aminoguanidine, are also implicated in the downregulation of nephrin in diabetes (<xref rid="B29" ref-type="bibr">29</xref>). Interestingly, several groups could demonstrate the reduction of PKC activity under the treatment with ACE inhibitors and aminoguanidine in diabetic subjects that could explain the nephrin protective effects (<xref rid="B30" ref-type="bibr">30</xref>, <xref rid="B31" ref-type="bibr">31</xref>).</p><p>Most previous studies with specific PKCβ inhibitor ruboxistaurin (LX333531) <italic>in vivo</italic> and <italic>in vitro</italic> indicated that PKCβ isoform is primarily responsible for the high-glucose-induced renal effects in diabetes (<xref rid="B32" ref-type="bibr">32</xref>–<xref rid="B36" ref-type="bibr">36</xref>). Meier et al. tested this hypothesis by inducing DN in PKCβ deficient mice and did not find a significant preventive effect of PKCβ deficiency on albuminuria. In contrast to non-albuminuric diabetic PKCα<sup>−/−</sup> mice, the loss of the basement membrane proteoglycan perlecan and the podocyte slit diaphragm protein nephrin were not prevented in the PKCβ<sup>−/−</sup> mice under diabetic conditions (<xref rid="B20" ref-type="bibr">20</xref>, <xref rid="B37" ref-type="bibr">37</xref>). However, the hyperglycemia-induced renal and glomerular hypertrophy as well as increased expression of ECM proteins was reduced in PKCβ deficiency diabetic mice.</p><p>In summary, the two important physiological features of DN, renal hypertrophy and albuminuria, are regulated through different PKC isoforms; PKCα is involved in the development of albuminuria and maintenance the glomerular filtration barrier structure, whereas the PKCβ-isoform contributes to hyperglycemia-induced renal fibrosis.</p><p>Another study by Menne et al. combined the findings about PKCα and PKCβ and demonstrated that a dual inhibition of both isoforms has a synergistic effect and is capable of preventing the development of experimental DN in streptozocin-induced diabetic mice (<xref rid="B38" ref-type="bibr">38</xref>). Blocking both isoforms has a beneficial effect on the development of renal hypertrophy and albuminuria in mice after 8 weeks of diabetes. A pharmacological approach with CGP41252, an inhibitor of PKCα and PKCβ showed that the occurrence of albuminuria could be avoided and preexisting albuminuria could be diminished in both type I and type II diabetic mice (<xref rid="B38" ref-type="bibr">38</xref>). But the treatment had little impact on the development of renal hypertrophy. Higher doses treatment also increased mortality.</p></sec><sec id="S5"><title>Essential Role of Novel and Atypical PKCs</title><p>Only little is known about PKCε in renal function, especially about its role in podocytes, although several previous studies showed increased expression and activation of PKCε isoform in experimental DN (<xref rid="B39" ref-type="bibr">39</xref>, <xref rid="B40" ref-type="bibr">40</xref>). Meier et al. investigated the functional role of PKCε in renal physiology using PKCε-knockout mice and found a renal phenotype with an elevated occurrence of tubulointerstitial fibrosis and glomerulosclerosis, whereas a systemic profibrotic phenotype was not observed (<xref rid="B41" ref-type="bibr">41</xref>). Moreover, they demonstrated an increased level of albuminuria in knockout mice whereas the kidney/body ratio remained normal in comparison to non-diabetic wild type mice, indicating that PKCε is probably less implicated in development of renal hypertrophy. Experiments on diabetic mice further showed that knockout of PKCε can exacerbate the renal phenotype with a significantly increased urinary albumin/creatinine ratio and expression of ECM proteins. These data suggest that a rising level of PKCε has a protective function in kidney injury, rather than inducing profibrotic changes.</p><p>The atypical isoforms PKCλ and PKCι are both highly expressed in podocytes. Although it is not clear whether both isotypes perform distinct roles in the cell, aPKCs are well-known to be important in the establishment of cell polarity (<xref rid="B42" ref-type="bibr">42</xref>, <xref rid="B43" ref-type="bibr">43</xref>). They form an evolutionarily conserved complex consisting of aPKC and PAR3 and PAR6, two PDZ domain containing scaffold proteins (<xref rid="B44" ref-type="bibr">44</xref>). Studies showed that the Par3–Par6–aPKC complex interacts with nephrin–podocin through the direct connection of Par3 to nephrin, an essential structural component for the maintenance of integrity of the glomerular filter as well as for the signal transduction (<xref rid="B45" ref-type="bibr">45</xref>). Interestingly, our own studies of the glomerular development in regard to the Par polarity complex and slit diaphragm molecules show that the PAR3–PAR6–aPKCλ complex translocates together with other tight junction proteins like ZO-1 from the apical to the basolateral side of the cell preceding the targeting of slit diaphragm components such as nephrin and podocin to the basal membrane, development of foot processes, and the construction of slit diaphragms (<xref rid="B46" ref-type="bibr">46</xref>).</p><p>The very recent study of Satoh and colleagues shows the necessity of aPKC in the exocytosis of newly produced nephrin and its localization on the surface of podocytes (<xref rid="B47" ref-type="bibr">47</xref>). This finding could also partially explain the diminished nephrin expression in proteinuria patients with diabetes (<xref rid="B48" ref-type="bibr">48</xref>). Under high-glucose conditions, activation of aPKC presents a protective effect on the nephrin expression (<xref rid="B48" ref-type="bibr">48</xref>).</p><p>The activation of aPKC is required for the insulin-induced glucose transport and thus defective aPKC activation in muscle and adipocytes has been shown previously in type II diabetic rats, monkeys, and human beings, which leads to a disturbed glucose uptake into muscle and the whole body glucose transfer (<xref rid="B49" ref-type="bibr">49</xref>). The knockout of the other isoform PKCζ has no specific renal phenotype but seems to be able compensate for partial functions of aPKCλ loss as the double-knockout leads to glomerular developmental defects with no secondary foot processes (<xref rid="B50" ref-type="bibr">50</xref>). This is most likely because of the regulatory cytoskeletal functions of aPKCs with direct influence on small GTPase activation of aPKCs in podocytes (<xref rid="B51" ref-type="bibr">51</xref>). Nevertheless, the role of aPKC in DN still remains unclear.</p></sec><sec id="S6"><title>PKC Up-Regulated Growth Factor Expression in Diabetic Nephropathy</title><p>Podocytes are the major site of vascular endothelial growth factor (VEGF) production in the human kidney (<xref rid="B52" ref-type="bibr">52</xref>), and the expression of VEGF is increased in podocytes in diabetic rats and human beings (<xref rid="B53" ref-type="bibr">53</xref>, <xref rid="B54" ref-type="bibr">54</xref>). Thus, the up-regulation of VEGF plays a critical role in the progression of DN. Hoshi et al. have shown that under high-glucose conditions, VEGF expression was up-regulated mediated through activation of PKC and extracellular signal-regulated kinase (ERK) in podocytes (<xref rid="B55" ref-type="bibr">55</xref>). PKC and ERK are known to regulate activator protein-1 (AP-1) activation (<xref rid="B56" ref-type="bibr">56</xref>–<xref rid="B58" ref-type="bibr">58</xref>), which promotes the binding of AP-1 to the promoter region of the VEGF gene (<xref rid="B59" ref-type="bibr">59</xref>, <xref rid="B60" ref-type="bibr">60</xref>). In addition, AGEs upregulate VEGF mRNA levels through transcription factors such as NF-κB and AP-1. AGEs are also able to activate PKCs, which further increase the expression of VEGF. Although VEGF is required for normal glomerulogenesis and essential for maintenance of glomerular filtration barrier, podocyte-specific overexpression of VEGF<sub>164</sub> or VEGF<sub>165</sub> isoform in animals leads to structural and functional renal changes similar to those abnormalities seen in DN, including proteinuria, glomerular hypertrophy, glomerular basement membrane thickening, mesangial expansion, loss of slit diaphragms, and podocyte effacement (<xref rid="B61" ref-type="bibr">61</xref>, <xref rid="B62" ref-type="bibr">62</xref>). In DN, the activation of TGF-β1 has been demonstrated to promote podocyte apoptosis and the development of glomerulosclerosis. A reduced expression level of the profibrotic cytokine TGF-β1 was detected in diabetic PKCβ<sup>−/−</sup> mice, while the alteration was not observed in diabetic PKCα<sup>−/−</sup> mice (<xref rid="B20" ref-type="bibr">20</xref>). This observation suggested that PKCβ isoform is more important in the up-regulation of TGF-β and for the development of glomerular fibrosis under diabetic conditions, whereas PKCα shows its critical role in the integrity of glomerular filtration barrier (<xref rid="B37" ref-type="bibr">37</xref>). However, PKCα plays a key role in the signaling response after stimulation with TGF-β1. An enhanced and prolonged activation of PI3K/AKT and ERK1/2 as well as a reduced proapoptotic signaling via p38MAPK in PKCα-knockout podocytes compared to wild type podocytes was detected after TGF-β1 treatment, which indicated the involvement of PKCα in the TGF-β mediated apoptosis (<xref rid="B63" ref-type="bibr">63</xref>).</p><p>Of note, deletion of PKCε signaling not only leads to increased expression of TGF-β1 but also induces activation of the TGF-β1 signaling pathway in glomeruli (<xref rid="B41" ref-type="bibr">41</xref>). PKCι might also mediate a high-glucose-induced increase in TGF-β receptor II (TGF-βRII) promoter activity, which leads to the up-regulation of TGF-βRII and fibronectin (<xref rid="B64" ref-type="bibr">64</xref>).</p></sec><sec id="S7"><title>PKC Regulated Structure Proteins in Podocytes</title><p>P-cadherin as member of the classical cadherins, a superfamily of glycoproteins, is known to be a basic scaffold for the glomerular slit diaphragm (<xref rid="B65" ref-type="bibr">65</xref>). Xu et al. have first demonstrated decreased expression of P-cadherin mRNA and protein in experimental diabetic glomeruli and in high-glucose stimulated podocytes, which suggests that a potential role for P-cadherin loss in the development of proteinuria in early DN (<xref rid="B66" ref-type="bibr">66</xref>). They also found that PKC inhibitors could ameliorate the decrement of P-cadherin in podocytes under high-glucose conditions. Thus, activation of PKC regulated pathways seems to be involved in the regulation of P-cadherin expression and contributes to the disruption of podocyte integrity (<xref rid="B66" ref-type="bibr">66</xref>).</p><p>Nevertheless, it seems likely that the molecular changes of the slit diaphragm complex, but not one single slit diaphragm-associated molecule contribute to the pathogenesis of glomerular filtration barrier in DN. P-cadherin, α-, β-, and γ-catenin, and ZO-1 are described to compose adherens junctions at the slit diaphragm and establish the link with the actin cytoskeleton (<xref rid="B65" ref-type="bibr">65</xref>, <xref rid="B67" ref-type="bibr">67</xref>). However, the role and regulation of β-catenin with P-cadherin and actin cytoskeletal proteins have not been thoroughly explored. In the situation of the podocyte injury induced by high glucose, β-catenin will be released from the destruction complex because of the activation of the Wnt-pathway. Afterwards β-catenin translocates into the nucleus to activate the downstream genes via the aggregation of a transcriptional complex with TCF/LEF, which leads to proteinuria and glomerulosclerosis (<xref rid="B68" ref-type="bibr">68</xref>, <xref rid="B69" ref-type="bibr">69</xref>). Several groups demonstrated that PKC activation phosphorylated N-terminal serine residues of β-catenin, which promoted β-catenin degradation (<xref rid="B70" ref-type="bibr">70</xref>, <xref rid="B71" ref-type="bibr">71</xref>). In contrast, our ongoing work was focused on several PKC isoforms, whose activation is able to dephosphorylate β-catenin to prevent β-catenin from degradation. It is still unknown, whether this process is protective for the high-glucose-induced cell adherens junction reduction. Nevertheless, PKC activation may be a novel mechanism in regulating β-catenin in glomerular injury, which will be further investigated by our laboratory.</p></sec><sec id="S8"><title>Conclusion</title><p>Podocyte injury or podocyte loss is a hallmark for the pathogenesis of DN. Based on the above described findings, PKC activation seems to be the most critical pathway involved in the progression of glomerular injury. Obviously, there is a fine balance between activation and inactivation of the different PKC isoforms and their cross-talk involving foot process cytoskeletal architecture, cellular junction formation, and orchestration of turnover and surface expression of slit diaphragm components (Figure <xref ref-type="fig" rid="F1">1</xref>). Therefore, PKCs and their involved pathways are potential therapeutic targets in podocytes to prevent the progression of diabetic glomerulopathy.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Schematic overview of PKC-isoform functions in podocytes</bold>. DAG or calcium activated PKCα induces nephrin endocytosis via the adaptor molecule Pick1. PKCλ/ι is required for foot process formation, cell polarity, and nephrin exocytosis. PKCε might be involved in actin remodeling via small GTPases and might orchestrate cell adhesion and cell–cell contact formation.</p></caption><graphic xlink:href="fendo-05-00179-g001"/></fig></sec><sec id="S9"><title>Conflict of Interest Statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec>
Excitatory and inhibitory projections in parallel pathways from the inferior colliculus to the auditory thalamus	<p>Individual subdivisions of the medial geniculate body (MG) receive a majority of their ascending inputs from 1 or 2 subdivisions of the inferior colliculus (IC). This establishes parallel pathways that provide a model for understanding auditory projections from the IC through the MG and on to auditory cortex. A striking discovery about the tectothalamic circuit was identification of a substantial GABAergic component. Whether GABAergic projections match the parallel pathway organization has not been examined. We asked whether the parallel pathway concept is reflected in guinea pig tectothalamic pathways and to what degree GABAergic cells contribute to each pathway. We deposited retrograde tracers into individual MG subdivisions (ventral, MGv; medial, MGm; dorsal, MGd; suprageniculate, MGsg) to label tectothalamic cells and used immunochemistry to identify GABAergic cells. The MGv receives most of its IC input (~75%) from the IC central nucleus (ICc); MGd and MGsg receive most of their input (~70%) from IC dorsal cortex (ICd); and MGm receives substantial input from both ICc (~40%) and IC lateral cortex (~40%). Each MG subdivision receives additional input (up to 32%) from non-dominant IC subdivisions, suggesting cross-talk between the pathways. The proportion of GABAergic cells in each pathway depended on the MG subdivision. GABAergic cells formed ~20% of IC inputs to MGv or MGm, ~11% of inputs to MGd, and 4% of inputs to MGsg. Thus, non-GABAergic (i.e., glutamatergic) cells are most numerous in each pathway with GABAergic cells contributing to different extents. Despite smaller numbers of GABAergic cells, their distributions across IC subdivisions mimicked the parallel pathways. Projections outside the dominant pathways suggest opportunities for excitatory and inhibitory crosstalk. The results demonstrate parallel tectothalamic pathways in guinea pigs and suggest numerous opportunities for excitatory and inhibitory interactions within and between pathways.</p>	<contrib contrib-type="author"><name><surname>Mellott</surname><given-names>Jeffrey G.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/86039"/></contrib><contrib contrib-type="author"><name><surname>Foster</surname><given-names>Nichole L.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/121652"/></contrib><contrib contrib-type="author"><name><surname>Ohl</surname><given-names>Andrew P.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref></contrib><contrib contrib-type="author"><name><surname>Schofield</surname><given-names>Brett R.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><xref ref-type="author-notes" rid="fn001"><sup>*</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/70780"/></contrib>	Frontiers in Neuroanatomy	<sec sec-type="introduction" id="s1"><title>Introduction</title><p>The projections from the inferior colliculus (IC) to the medial geniculate body (MG) have been described as 3 parallel pathways: (1) a “lemniscal” or “tonotopic” pathway; (2) a “polysensory” pathway; and (3) a “diffuse” pathway (Calford and Aitkin, <xref rid="B8" ref-type="bibr">1983</xref>; Rouiller, <xref rid="B33" ref-type="bibr">1997</xref>). The pathways reflect subdivision-specific connections from the IC to the MG and from the MG to auditory cortex (and other forebrain targets) and are considered to serve different functions in hearing (Oliver and Hall, <xref rid="B28" ref-type="bibr">1978a</xref>,<xref rid="B29" ref-type="bibr">b</xref>; Calford and Aitkin, <xref rid="B8" ref-type="bibr">1983</xref>; Redies et al., <xref rid="B32" ref-type="bibr">1989</xref>; Redies and Brandner, <xref rid="B31" ref-type="bibr">1991</xref>; Hu et al., <xref rid="B18" ref-type="bibr">1994</xref>; de Ribaupierre, <xref rid="B10" ref-type="bibr">1997</xref>; Bartlett and Smith, <xref rid="B6" ref-type="bibr">1999</xref>, <xref rid="B7" ref-type="bibr">2002</xref>; Edeline et al., <xref rid="B12" ref-type="bibr">1999</xref>; He, <xref rid="B13" ref-type="bibr">2001</xref>; Hu, <xref rid="B17" ref-type="bibr">2003</xref>; Smith et al., <xref rid="B35" ref-type="bibr">2007</xref>; Anderson et al., <xref rid="B3" ref-type="bibr">2009</xref>; Lee and Sherman, <xref rid="B21" ref-type="bibr">2010</xref>; Anderson and Linden, <xref rid="B2" ref-type="bibr">2011</xref>; Edeline, <xref rid="B11" ref-type="bibr">2011</xref>; Venkataraman and Bartlett, <xref rid="B37" ref-type="bibr">2013</xref>). The lemniscal pathway has been associated with primary-like representation of sound. It is formed primarily by projections from central IC (ICc) to ventral MG (MGv) and from there to tonotopically organized parts of the auditory cortex (de Ribaupierre, <xref rid="B10" ref-type="bibr">1997</xref>). The diffuse pathway has been associated with complex sounds and detecting change in context-dependent signals (de Ribaupierre, <xref rid="B10" ref-type="bibr">1997</xref>). The diffuse pathway involves projections from IC dorsal cortex (ICd) to dorsal MG (MGd) and from there to non-tonotopic secondary and temporal auditory cortical areas (de Ribaupierre, <xref rid="B10" ref-type="bibr">1997</xref>). Finally, the polysensory pathway has been associated with multimodal processing, reflecting inputs from auditory as well as other sensory systems (Love and Scott, <xref rid="B23" ref-type="bibr">1969</xref>). The polysensory pathway is unique among the three pathways in several ways. While it is closely associated with a single MG subdivision (the medial MG, MGm), it receives substantial inputs from 2 IC subdivisions (the ICc and the IC lateral cortex, IClc). The polysensory pathway also differs from the other pathways in having much broader projections to the forebrain, terminating widely across all areas of auditory cortex. Moreover, these thalamocortical projections terminate most heavily in cortical layer I, whereas the thalamocortical projections in the lemniscal and diffuse pathways terminate most heavily in the middle cortical layers (III–IV). While tectothalamic projections show some overlap (e.g., the ICc contributes to both the lemniscal and polysensory pathways), the general segregation is assumed to underlie substantial physiological and functional differences between these pathways.</p><p>One of the most striking discoveries about the tectothalamic pathways has been the detection of an inhibitory component arising from GABAergic IC cells (Winer et al., <xref rid="B39" ref-type="bibr">1996</xref>; Peruzzi et al., <xref rid="B30" ref-type="bibr">1997</xref>; Bartlett and Smith, <xref rid="B6" ref-type="bibr">1999</xref>; Smith et al., <xref rid="B35" ref-type="bibr">2007</xref>; Mellott et al., <xref rid="B24" ref-type="bibr">2014</xref>). GABAergic tectothalamic cells are found throughout the IC and, depending on the species, constitute 20–50% of the tectothalamic cells (cats: 20%, Winer et al., <xref rid="B39" ref-type="bibr">1996</xref>; rats: 40%, Peruzzi et al., <xref rid="B30" ref-type="bibr">1997</xref>; guinea pigs: 22%; Mellott et al., <xref rid="B24" ref-type="bibr">2014</xref>). The remaining tectothalamic cells are glutamatergic, providing ascending excitation to the MG. Physiological studies have shown that the ascending excitatory and inhibitory inputs are integrated in different ways by neurons in different MG subdivisions, supporting the proposed functional distinctions between the MG subdivisions and the associated parallel pathways (Smith et al., <xref rid="B35" ref-type="bibr">2007</xref>). However, previous anatomical studies of the GABAergic projections were based on tracer injections that included two or more subdivisions of the MG and thus did not address whether the GABAergic projections targeted a specific subdivision in the MG (Winer et al., <xref rid="B39" ref-type="bibr">1996</xref>; Peruzzi et al., <xref rid="B30" ref-type="bibr">1997</xref>; Mellott et al., <xref rid="B24" ref-type="bibr">2014</xref>). An understanding of the various functions of the MG subdivisions and their ascending projections will require a clear delineation of both the excitatory and inhibitory projections they receive from the IC.</p><p>In the present study, we combine retrograde transport from individual MG subdivisions with immunochemistry to distinguish GABAergic from non-GABAergic tectothalamic cells. We completed the studies in guinea pigs, which have recently been subjects of both anatomical and physiological studies of the MG subdivisions but which have not been examined with respect to the parallel tectothalamic pathways (Anderson et al., <xref rid="B4" ref-type="bibr">2006</xref>, <xref rid="B5" ref-type="bibr">2007</xref>). The more recent study distinguished a “suprageniculate” subdivision (MGsg), that has been described in some other species but is often included with the MGd or the MGm. Support for distinguishing this subdivision in the context of tectothalamic projections comes from preliminary studies suggesting that the MGsg differs from the other subdivisions (MGv, MGd and MGm) in receiving very little GABAergic input from the IC (Mellott and Schofield, <xref rid="B25" ref-type="bibr">2011</xref>). Our findings suggest that the IC projections to individual MG subdivisions in guinea pigs are similar to those described in other species. In addition, GABAergic cells contribute to each of these pathways. In general, both the GABAergic (presumed inhibitory) and the non-GABAergic (presumed excitatory) projections from a particular IC subdivision have strong projections to specific MG subdivisions and smaller projections to other regions of the MG. These latter projections could provide for both excitatory and inhibitory cross-talk between the parallel pathways.</p></sec><sec sec-type="materials\|methods" id="s2"><title>Materials and methods</title><p>All procedures were conducted in accordance with the Northeast Ohio Medical University Institutional Animal Care and Use Committee and NIH guidelines. Results are described from ten adult pigmented guinea pigs (Elm Hill Labs; Chelmsford, MA, USA) of either gender weighing 317–1000 g (most animals were age 5 weeks to 4 months; 1 animal was ~2 years old). Efforts were made to minimize the number of animals and their suffering.</p><sec id="s2-1"><title>Surgery</title><p>Each animal was anesthetized with isoflurane (4–5% for induction, 1.75–3% for maintenance) in oxygen. The head was shaved and disinfected with Betadine (Purdue Products L.P., Stamford, CT, USA). Atropine sulfate (0.08 mg/kg i.m.) was given to minimize respiratory secretions and Ketofen (ketoprofen, 3 mg/kg i.m.; Henry Schein, Melville, NY 11747, USA) was given for post-operative pain management. Moisture Eyes PM ophthalmic ointment (Bausch, Lomb, Rochester, NY, USA) was applied to each eye to protect the cornea. The animal’s head was positioned in a stereotaxic frame. Body temperature was maintained with a feedback-controlled heating pad. Sterile instruments and aseptic techniques were used for all surgical procedures. An incision was made in the scalp and the surrounding skin was injected with Marcaine (0.25% bupivacaine with epinephrine 1:200,000; Hospira, Inc., Lake Forest, IL, USA), a long-lasting local anesthetic. A craniotomy was made over the desired target coordinates using a dental drill. Following the tracer injection, Gelfoam (Harvard Apparatus, Holliston, MA, USA) was placed in the craniotomy site and the scalp was sutured. The animal was then removed from the stereotaxic frame and placed in a clean cage. The animal was monitored until it could walk, eat and drink without difficulty.</p></sec><sec id="s2-2"><title>Retrograde tracers</title><p>Fluorescent tracers (red fluorescent RetroBeads [“red beads”] and green fluorescent RetroBeads [“green beads”], Luma-Fluor, Inc., Naples, FL, USA; FluoroGold, FluoroChrome, Inc., Englewood, CO, USA) were deposited into the MG via stereotaxic coordinates. For most experiments, a Hamilton microsyringe (1 µl; Hamilton, Reno, NV, USA) or a micropipette (tip diameter 25–35 μm) attached to a Nanoliter Injector (World Precision Instruments, Sarasota, FL, USA) was used to deposit one of the tracers into the MG (Table <xref ref-type="table" rid="T1">1</xref>). Each syringe was dedicated to a single tracer. Injections were small in volume, < 70 nl, to better ensure the deposit was contained primarily or exclusively within one subdivision of the MG. In order to limit the spread of tracer into neighboring nuclei, the volume injected at each site was designed to account for the diffusibility of each tracer (Schofield, <xref rid="B34" ref-type="bibr">2008</xref>). In one animal, FluoroGold was deposited by iontophoresis through a micropipette (tip diameter 20 μm, +1.5 μA current, 15 min, 50% duty cycle) (Table <xref ref-type="table" rid="T1">1</xref>).</p><table-wrap id="T1" position="float"><label>Table 1</label><caption><p><bold>Summary of the tracers, volumes injected, and spread of injection sites into MG subdivisions after injections into left (L) and/or right (R) MG</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th rowspan="1" colspan="1"/><th rowspan="1" colspan="1"/><th rowspan="1" colspan="1"/><th rowspan="1" colspan="1"/><th align="center" colspan="4" rowspan="1">Extent of injection site</th></tr><tr><th align="left" rowspan="1" colspan="1">Case</th><th align="center" rowspan="1" colspan="1">Side</th><th align="center" rowspan="1" colspan="1">Tracer</th><th align="center" rowspan="1" colspan="1">Total volume</th><th align="center" rowspan="1" colspan="1">MGv</th><th align="center" rowspan="1" colspan="1">MGd</th><th align="center" rowspan="1" colspan="1">MGm</th><th align="center" rowspan="1" colspan="1">MGsg</th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">GP689</td><td align="center" rowspan="1" colspan="1">L</td><td align="center" rowspan="1" colspan="1">RB</td><td align="center" rowspan="1" colspan="1">69 nl</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">X</td><td align="center" rowspan="1" colspan="1">(x)</td></tr><tr><td align="left" rowspan="1" colspan="1">GP689</td><td align="center" rowspan="1" colspan="1">R</td><td align="center" rowspan="1" colspan="1">GB</td><td align="center" rowspan="1" colspan="1">69 nl</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">X</td><td align="center" rowspan="1" colspan="1">-</td></tr><tr><td align="left" rowspan="1" colspan="1">GP693</td><td align="center" rowspan="1" colspan="1">L</td><td align="center" rowspan="1" colspan="1">RB</td><td align="center" rowspan="1" colspan="1">46 nl</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">X</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">-</td></tr><tr><td align="left" rowspan="1" colspan="1">GP695</td><td align="center" rowspan="1" colspan="1">L</td><td align="center" rowspan="1" colspan="1">RB</td><td align="center" rowspan="1" colspan="1">27.6 nl</td><td align="center" rowspan="1" colspan="1">X</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">-</td></tr><tr><td align="left" rowspan="1" colspan="1">GP696</td><td align="center" rowspan="1" colspan="1">L</td><td align="center" rowspan="1" colspan="1">RB</td><td align="center" rowspan="1" colspan="1">27.6 nl</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">X</td></tr><tr><td align="left" rowspan="1" colspan="1">GP698</td><td align="center" rowspan="1" colspan="1">L</td><td align="center" rowspan="1" colspan="1">RB</td><td align="center" rowspan="1" colspan="1">18.4 nl</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">X</td></tr><tr><td align="left" rowspan="1" colspan="1">GP702</td><td align="center" rowspan="1" colspan="1">R</td><td align="center" rowspan="1" colspan="1">FG</td><td align="center" rowspan="1" colspan="1">ion<sup>#</sup></td><td align="center" rowspan="1" colspan="1">X</td><td align="center" rowspan="1" colspan="1">(x)</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">-</td></tr><tr><td align="left" rowspan="1" colspan="1">GP712</td><td align="center" rowspan="1" colspan="1">L</td><td align="center" rowspan="1" colspan="1">RB</td><td align="center" rowspan="1" colspan="1">50 nl</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">X</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">(x)</td></tr><tr><td align="left" rowspan="1" colspan="1">GP718</td><td align="center" rowspan="1" colspan="1">L</td><td align="center" rowspan="1" colspan="1">FG</td><td align="center" rowspan="1" colspan="1">50 nl</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">X</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">-</td></tr><tr><td align="left" rowspan="1" colspan="1">GP719</td><td align="center" rowspan="1" colspan="1">L</td><td align="center" rowspan="1" colspan="1">FG</td><td align="center" rowspan="1" colspan="1">50 nl</td><td align="center" rowspan="1" colspan="1">X</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">-</td></tr><tr><td align="left" rowspan="1" colspan="1">GP723</td><td align="center" rowspan="1" colspan="1">L</td><td align="center" rowspan="1" colspan="1">FB</td><td align="center" rowspan="1" colspan="1">50 nl</td><td align="center" rowspan="1" colspan="1">(x)</td><td align="center" rowspan="1" colspan="1">X</td><td align="center" rowspan="1" colspan="1">-</td><td align="center" rowspan="1" colspan="1">-</td></tr></tbody></table><table-wrap-foot><p><italic>X = significant involvement of the deposit. (x) = indicates minor involvement of the listed MG subdivision. – = no involvement of the listed MG subdivision. ion<sup>#</sup> = iontophoretic injection. nl = nanoliter. indicates cases used for quantitative analysis</italic>.</p></table-wrap-foot></table-wrap></sec><sec id="s2-3"><title>Perfusion and tissue processing</title><p>Five to thirteen days after surgery, the animal was deeply anesthetized with isoflurane and perfused transcardially with Tyrode’s solution, followed by 250 ml of 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4 and then by 250 ml of the same fixative with 10% sucrose. The brain was removed and stored at 4°C in fixative with 25–30% sucrose for cryoprotection. The following day the brain was prepared for processing by removing the cerebellum and blocking the remaining piece with transverse cuts posterior to the superior olive and anterior to the auditory cortex. Each piece of tissue was frozen and cut on a sliding microtome into 40 or 50 µm thick transverse sections that were collected serially in six sets.</p><p>Putative GABAergic cells were stained with immunochemistry for glutamic acid decarboxylase (GAD; Nakamoto et al., <xref rid="B27" ref-type="bibr">2013</xref>). Briefly, the sections were pretreated with normal goat serum to limit non-specific labeling, then exposed (1–2 days at 4°C) to mouse anti-GAD monoclonal antibody (GAD67; #MAB5406 Millipore, diluted 1:1000 to 1:100). The sections were treated with 1% biotinylated goat anti-mouse antibody (Vector Laboratories, Burlingame, CA, USA: BA-9200) and labeled with streptavidin conjugated to a fluorescent marker (AlexaFluor 488 [green] or AlexaFluor 647 [near-infrared], Invitrogen, Carlsbad, CA, USA). For transversely cut cases, a series of sections adjacent to the one used for tracer analysis was stained to facilitate identification of IC and MG subdivisions. The IC and the MG do no coexist in the transverse plane so sections with IC tissue could be separated from sections with MG tissue. The method of Coote and Rees (<xref rid="B9" ref-type="bibr">2008</xref>) was used to stain IC sections with antibodies to brain nitric oxide synthase (bNOS) and then identify IC subdivisions. The method of Anderson et al. (<xref rid="B5" ref-type="bibr">2007</xref>) was used on MG sections to reveal cytochrome oxidase activity and to identify subdivisions of the MG. In one case (GP723) the tissue was cut in the sagittal plane. Because the IC and the MG coexist in the sagittal plane, one series was stained with bNOS to identify the IC subdivisions and the other series was stained with cytochrome oxidase to identify the MG subdivisions. These series were on either side of the tracer-analyzed series. Stained sections were mounted on gelatin-coated slides, allowed to dry and coverslipped with DPX (Sigma).</p></sec><sec id="s2-4"><title>Data analysis</title><p>Subdivisions of the MG were identified by their patterns of staining with cytochrome oxidase (Anderson et al., <xref rid="B5" ref-type="bibr">2007</xref>). IC subdivisions were identified by the differential expression of bNOS, as detailed in Coote and Rees (<xref rid="B9" ref-type="bibr">2008</xref>). The borders of the ICc were clarified by observation at high power to identify disc-shaped cells that stain for bNOS and that are characteristic of the ICc (Coote and Rees, <xref rid="B9" ref-type="bibr">2008</xref>). Immunostaining revealed GAD-immunoreactive (GAD+) cells and boutons throughout the IC. Immunopositive cells were labeled intensely and were readily distinguished from immunonegative cells. The GAD immunostain was also readily visible in tracer-labeled cells, making it straightforward to distinguish GAD+ vs. GAD-negative staining in the retrogradely-labeled cells, including cells that contained two different retrograde tracers.</p><p>The location and extent of each injection site was determined by comparison of the tracer deposit with borders of MG subdivisions identified in sections stained for cytochrome oxidase (Anderson et al., <xref rid="B5" ref-type="bibr">2007</xref>). Results from seven injections (4 RB; 1 GB; 2 FG) that also had robust immunostaining were used for quantitative analysis (Table <xref ref-type="table" rid="T1">1</xref>). Labeled cells in the IC were plotted with a Neurolucida reconstruction system (MBF Bioscience, Williston, VT, USA) attached to a Zeiss Axioplan II microscope (Carl Zeiss MicroImaging, Inc., Thornwood, NY, USA) or a Zeiss AxioImager Z2 with an attached Apotome II (Zeiss). For each case, every labeled cell was plotted in the ipsilateral IC across a series of transverse sections (every sixth section). Each combination of tracer and immunolabel was plotted with a unique marker. The results of these plots were used for a quantitative summary of the distributions of the labeled cells.</p><p>In some cases, the anti-GAD staining did not fully penetrate the tissue, resulting in a central layer in the section where GAD staining was absent. Sections cut at 40–50 µm thickness typically shrink to 20–30 µm thickness due to tissue processing and dehydration prior to mounting on slides. In some of our cases, the GAD staining was robust only 5–10 µm from each surface, leaving an unstained or poorly stained central layer typically 10–15 µm thick. Data from these cases were plotted with the Neurolucida system and a 63X objective (NA = 1.4), with special attention to focusing on the center of the soma when plotting the symbol for a particular cell. This approach provides sufficient resolution in the z plane (section depth) to allow subsequent filtering of the data by depth. After the data were plotted, the X, Y, and Z coordinates of all markers from each subdivision of each tissue section were exported from Neurolucida to Microsoft Excel and sorted based on the Z coordinate. The depth of penetration of the GAD labeling was assessed under the 63X objective for each subdivision of each section to determine the range of depths (measured from the top surface of the section) where GAD staining was robust. This yielded 2 zones of data from each section (1 associated with each surface), and a central zone that was not stained with GAD. All markers in the central, unstained zone were excluded from further analyses.</p><p>Figures showing the distribution of labeled cells were created with Neurolucida software (MBF Bioscience) and refined with Adobe Illustrator (Adobe Systems, Inc., San Jose, CA, USA). Photomicrographs were captured using either a Zeiss AxioImager Z1 fluorescence microscope and AxioCam HRm or HRc cameras (Zeiss) or a Zeiss Axioskop fluorescence microscope and Magnafire camera (Optronics, Goleta, CA, USA). Adobe Photoshop (Adobe Systems) was used to add scale bars, crop images, erase background around tissue sections, adjust intensity levels and colorize monochrome images.</p></sec></sec><sec sec-type="results" id="s3"><title>Results</title><p>We combined retrograde tracing and immunolabeling for GAD to identify GABAergic IC cells that project to individual subdivisions of the MG. While our main objective was to distinguish the GABAergic vs. non-GABAergic components of the tectothalamic pathways, it was necessary to first establish the overall patterns of connections between the IC subdivisions and the MG subdivisions. As described in the Introduction, a suprageniculate subdivision has been distinguished in several species, including guinea pigs. We continue this distinction and, for the sake of discussion, group the MGsg with the MGd as part of the diffuse pathway. We first describe the injection sites and evidence for parallel pathways without regard to GAD-immunoreactivity. We then describe the same experimental cases with attention to the presence or absence of GAD immunostaining in the retrogradely labeled cells.</p><sec id="s3-1"><title>Injection sites and evidence for parallel tectothalamic pathways</title><p>The results are based on tracer deposits in 11 MGs (Table <xref ref-type="table" rid="T1">1</xref>). Most of the injections were isolated to one subdivision of the MG (Figure <xref ref-type="fig" rid="F1">1</xref>). Quantitative data were derived from 7 cases. Four cases had deposits that were centered in a particular MG subdivision but spread slightly into an adjacent subdivision. The number of cells labeled by the encroachment was probably very small, but because the exact number could not be determined, these cases were excluded from quantitative analysis. Nonetheless, the overall labeling patterns in these cases were very similar to those with more restricted injections, and thus serve to confirm the findings. The number of labeled cells varied between cases, but the overall patterns and percentages of cells in IC subdivisions were consistent between animals and between different tracers for injections in a given subdivision.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Photomicrographs showing representative deposits of Red Beads (RB) or Green Beads (GB) in four subdivisions of the medial geniculate body (MG)</bold>. <bold>(A)</bold> A deposit of RB contained within the left MGv. The bright fluorescence on the dorsolateral edge of the section is imaging artifact resulting from tissue damage during sectioning; the tracer deposit is the bright spot within the MGv. <bold>(B)</bold> A deposit of GB contained within the right MGm. Additional green fluorescence is seen around the margins of a blood vessel along the dorsomedial border of the ventral MG (v); this represents spread of beads that does not result in retrogradely labeled cells. The tracer was deposited in the right MG; the image is reversed left to right to facilitate comparisons with the other panels. <bold>(C)</bold> A deposit of RB contained within the MGd. <bold>(D)</bold> A deposit of RB contained within the left MGsg. Experiment numbers (e.g., GP695) are shown in each panel (cf. Table <xref ref-type="table" rid="T1">1</xref>). Scale bar = 0.5 mm. D—dorsal; MGd—dorsal division of the MG; M—medial; MGm—medial division of the MG; s—shell of the MG; L—lateral; SC—superior colliculus; MGsg—suprageniculate division of the MG; MGv—ventral division of the MG.</p></caption><graphic xlink:href="fnana-08-00124-g0001"/></fig><p>Injections into different MG subdivisions yielded distinct distributions of labeled cells in the IC, supporting the idea of parallel but different pathways to each MG subdivision (Figure <xref ref-type="fig" rid="F2">2</xref>). Tracer injections into the MGv labeled cells across the IC, with a majority (three-fourths) located in the ICc (Figure <xref ref-type="fig" rid="F2">2A</xref>). A very different pattern followed injections into the MGm, where the majority of labeled cells were split nearly equally between two IC subdivisions (40% in the ICc and 39% in the IClc (Figure <xref ref-type="fig" rid="F2">2B</xref>). Injections into the MGd or the MGsg produced very similar results, with a majority of labeled cells located in the ICd (Figures <xref ref-type="fig" rid="F2">2C,D</xref>). Also in both situations, there were very few labeled cells in the ICc (Figures <xref ref-type="fig" rid="F2">2C,D</xref>). The similarities in these distributions (and their distinct difference from results of injections into the other 2 MG subdivisions) provides the rationale for grouping the MGd and the MGsg results together (associated with the “diffuse” pathway as described in the Introduction). The results of GAD immunochemistry, described below, will provide the basis for distinguishing the MGd from the MGsg.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Histograms showing the distribution of cells in the various subdivisions of the IC that project to the MGv (<italic>n</italic> = 2) (A), MGm (<italic>n</italic> = 1) (B), MGd (<italic>n</italic> = 2) (C), or MGsg (<italic>n</italic> = 2) (D)</bold>. The <italic>y</italic> axis reflects the proportion of labeled cells in each IC subdivision as a percentage of all the labeled cells in the IC. <italic>n</italic> = the # of IC cells counted for injections into the indicated MG subdivision. ICc—IC central nucleus of the IC; ICd—IC dorsal cortex; IClc—IC lateral cortex; MGd—dorsal division of the MG; MGm—medial division of the MG; MGsg—suprageniculate division of the MG; MGv—ventral division of the MG.</p></caption><graphic xlink:href="fnana-08-00124-g0002"/></fig></sec><sec id="s3-2"><title>GAD-positive (GAD+) and GAD-negative tectothalamic cells</title><p>Tracer injections into any of the MG subdivisions labeled GAD+ and GAD-negative IC cells. The GAD+ cells (Figure <xref ref-type="fig" rid="F3">3</xref>, arrows) were interpreted as GABAergic cells that project to the injection site. GAD-negative retrogradely-labeled cells (Figure <xref ref-type="fig" rid="F3">3</xref>, arrowheads) were often in close proximity to GAD+ cells. As described in Methods, our quantitative analyses included only those retrogradely-labeled cells at tissue depths that were successfully stained with anti-GAD immunostaining. Consequently, we interpreted these immunonegative cells as non-GABAergic and not the result of inadequate GAD staining.</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Paired photomicrographs showing retrogradely-labeled cells in the inferior colliculus (IC) that are GAD-immunopositive (GAD+; arrows) or GAD-negative (arrowheads)</bold>. The top row in each pair shows cells retrogradely labeled by either Red Beads or Green Beads. The bottom row in each pair shows the same field viewed for immunoreactivity to GAD (cyan). <bold>(A)</bold> Cells that project to the ipsilateral ventral division of the medial geniculate body (MGv). Images are taken from the IC central nucleus (ICc, left column); the IC dorsal cortex (ICd; middle column) and the IC lateral cortex (IClc; right column). GP695. <bold>(B)</bold> Cells that project to the ipsilateral medial division of the medial geniculate body (MGm). Images in the left, middle and right columns are from the ICc, ICd and IClc, respectively. GP689. <bold>(C)</bold> Cells that project to the ipsilateral dorsal division of the medial geniculate body (MGd). Images are from the ICd (left column) and IClc (right column). GP693. <bold>(D)</bold> Cells that project to the ipsilateral suprageniculate division of the medial geniculate body (MGsg). Images are from the ICd (left column) and IClc (right column). GP698. Scale bar = 50 µm.</p></caption><graphic xlink:href="fnana-08-00124-g0003"/></fig><p>Although all our injections labeled both GAD+ and GAD-negative IC cells, their proportions relative to one another and their distribution among the IC subdivisions varied according to the MG subdivision that was injected. The following sections describe the distributions of GAD+ and GAD-negative cells across the IC subdivisions following injections into each of the 4 MG subdivisions investigated.</p></sec><sec id="s3-3"><title>GAD-negative and GAD+ projections to individual MG subdivisions</title><p>Tracer injections restricted to the MGv labeled GAD-negative and GAD+ cells in each IC subdivision (Figure <xref ref-type="fig" rid="F4">4A</xref>). Overall, 79% of the tracer-labeled cells were GAD-negative. These GAD-negative cells were most numerous in the ICc, with the remaining cells split nearly evenly between the IClc and the ICd (Figure <xref ref-type="fig" rid="F4">4B</xref>, top graph). The GAD+ population constituted 21% of the retrogradely labeled cells. This population was also most prominent in the ICc with very few cells in the ICd and IClc (Figure <xref ref-type="fig" rid="F4">4B</xref>, bottom graph). Thus, both GAD-negative and GAD+ projections to the MGv originate primarily from the ICc.</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>(A)</bold> Plots of transverse sections of the inferior colliculus (IC) illustrating the distribution of GAD+ (blue circles) and GAD-negative (red triangles) cells that were labeled by an injection of Red Beads into the ipsilateral ventral division of the medial geniculate body (MGv). Each symbol represents one retrogradely-labeled cell. Dorsal is up; 1 is the most caudal section; 5 is the most rostral section. Only the left IC is shown from each section except for section 3, which is accompanied by a full drawing of the brainstem cross section. Case GP695. <bold>(B)</bold> Histograms summarizing the distribution of GAD-negative and GAD+ IC cells that project to the MGv (data from GP695 and GP719; total <italic>n</italic> = 652 cells). Aq—aqueduct; ICc—IC central nucleus of the IC; ICd—IC dorsal cortex; IClc—lateral cortex of the IC; ll—lateral lemniscus; scp—superior cerebellar peduncle; V—motor nucleus of V.</p></caption><graphic xlink:href="fnana-08-00124-g0004"/></fig><p>Tracer injections restricted to the MGm labeled GAD-negative and GAD+ cells in each IC subdivision (Figure <xref ref-type="fig" rid="F5">5A</xref>). Overall, GAD-negative cells constituted 80% and GAD+ cells 20% of the tracer-labeled IC cells. As described above, the MGm was unique in receiving substantial inputs from two IC subdivisions (namely, the ICc and IClc). This pattern applied to both the GAD-negative and GAD+ populations of projecting cells (Figure <xref ref-type="fig" rid="F5">5B</xref>). The ICd contained the fewest cells in each population, with GAD+ cells particularly limited. Thus, the MGm receives substantial GAD+ and GAD-negative projections from both the ICc and the IClc, and a small, primarily GAD-negative projection from the ICd.</p><fig id="F5" position="float"><label>Figure 5</label><caption><p><bold>(A)</bold> Plots of transverse sections illustrating the distribution of GAD+ (blue circles) and GAD-negative (red triangles) inferior colliculus (IC) cells that were labeled by an injection of Green Beads into the ipsilateral MGm. Each symbol represents one retrogradely-labeled cell. Dorsal is up; 1 is the most caudal section; 6 is the most rostral section. Case GP689. <bold>(B)</bold> Histograms summarizing the distribution of GAD-negative and GAD+ IC cells that project to the MGm (data from GP689R; total <italic>n</italic> = 703 cells). ICc—IC central nucleus; ICd—IC dorsal cortex; IClc—lateral cortex of the IC.</p></caption><graphic xlink:href="fnana-08-00124-g0005"/></fig><p>Tracer injections restricted to the MGd labeled GAD-negative and GAD+ cells in each IC subdivision (Figure <xref ref-type="fig" rid="F6">6A</xref>). Overall, GAD-negative cells constituted 89% and GAD+ cells 11% of the tracer-labeled IC cells (Figure <xref ref-type="fig" rid="F6">6B</xref>). Both the GAD-negative and GAD+ populations were most prominent in the ICd (Figure <xref ref-type="fig" rid="F6">6B</xref>). Smaller subsets of both groups were located in the IClc, and the smallest proportions of each group were found in the ICc (Figure <xref ref-type="fig" rid="F6">6B</xref>). Thus, the MGd receives substantial GAD-negative and GAD+ projections primarily from the ICd and much less so from the IClc.</p><fig id="F6" position="float"><label>Figure 6</label><caption><p><bold>(A)</bold> Plots of transverse sections illustrating the distribution of GAD+ (blue circles) and GAD-negative (red triangles) inferior colliculus (IC) cells that were labeled by an injection of Red Beads into the ipsilateral MGd. Each symbol represents one retrogradely-labeled cell. Dorsal is up; 1 is the most caudal section; 6 is the most rostral section. Case GP693. <bold>(B)</bold> Histograms summarizing the distribution of GAD-negative and GAD+ IC cells that project to the MGd (data from GP693 and GP718; total <italic>n</italic> = 545 cells). ICc—IC central nucleus; ICd—IC dorsal cortex; IClc—lateral cortex of the IC.</p></caption><graphic xlink:href="fnana-08-00124-g0006"/></fig><p>Tracer injections restricted to the MGsg labeled GAD-negative cells in each IC subdivision and GAD+ cells in the ICd and IClc (Figure <xref ref-type="fig" rid="F7">7A</xref>). Overall, GAD-negative cells constituted 96% and GAD+ cells 4% of the tracer-labeled cells. The GAD-negative cells were most numerous in the ICd, with most of the remainder in the IClc (Figure <xref ref-type="fig" rid="F7">7B</xref>, top graph). The rare GAD+ cells were located in the IClc and, less often, in the ICd (Figure <xref ref-type="fig" rid="F7">7B</xref>, bottom graph). Thus the MGsg is unique in receiving an extremely limited GABAergic projection from the IC. The prominent GAD-negative projection originates primarily from the ICd. We employed chi-square independence tests along with <italic>post hoc</italic> pairwise chi-square tests to determine the likelihood that a given IC cell projecting to the MGsg was GAD+. Results showed that a cell in the IC projecting to the MGsg was significantly less likely to be GAD+ than if it were projecting to MGv, MGm or MGd (Figure <xref ref-type="fig" rid="F8">8</xref>; <italic>p</italic> < 0.001).</p><fig id="F7" position="float"><label>Figure 7</label><caption><p><bold>(A)</bold> Plots of transverse sections illustrating the distribution of GAD+ (blue circles) and GAD-negative (red triangles) inferior colliculus (IC) cells that were labeled by an injection of Red Beads into the ipsilateral MGsg. Each symbol represents one retrogradely-labeled cell. Dorsal is up; 1 is the most caudal section; 6 is the most rostral section. GP698. <bold>(B)</bold> Histograms summarizing the distribution of GAD-negative and GAD+ IC cells that project to the MGsg (data from GP696 and GP698; total <italic>n</italic> = 687 cells). ICc—IC central nucleus; ICd—IC dorsal cortex; IClc—lateral cortex of the IC.</p></caption><graphic xlink:href="fnana-08-00124-g0007"/></fig><fig id="F8" position="float"><label>Figure 8</label><caption><p><bold>Histogram showing the percentage of GAD+ cells in pathways from the IC to specific MG subdivisions</bold>. Statistical comparisons were generated by chi-square independence tests along with <italic>post hoc</italic> pairwise chi-square tests to determine the likelihood that a cell in the IC that projected to the MGsg was significantly less likely to be GAD+ than if it were projecting to MGv, MGm or MGd. Statistical significance: *<italic>p</italic> < 0.001. MGd—dorsal division of the MG; MGm—medial division of the MG; MGsg—suprageniculate division of the MG; MGv—ventral division of the MG.</p></caption><graphic xlink:href="fnana-08-00124-g0008"/></fig></sec></sec><sec sec-type="discussion" id="s4"><title>Discussion</title><p>Despite the common usage of guinea pigs in auditory research, there is little information about the organization of tectothalamic projections in this species. The current study examines the projections from specific IC subdivisions to 4 subdivisions of the MG in guinea pigs. Our first finding is that each MG subdivision receives input predominantly from one (or, for the MGm, two) IC subdivisions. These dominant connections closely reflect the parallel pathways described in several other species. Inputs from the non-dominant IC subdivisions represented 21–32% of the tectothalamic inputs to a particular MG subdivision. These inputs represent a possibility of cross-talk between the parallel pathways that could underlie functional integration. Our second objective focused on the presence of GAD-negative and GAD-positive tectothalamic cells. We asked whether these two subpopulations share similar patterns of projection to the MG. For all the MG subdivisions examined, the majority of IC inputs come from GAD-negative, presumptively glutamatergic, cells. GAD-positive, presumptive GABAergic, cells constituted 4–21% of the inputs to individual MG subdivisions. For projections to 3 of the MG subdivisions, the glutamatergic and GABAergic projections showed a similar distribution of inputs from the different IC subdivisions, differing only in the greater overall number of glutamatergic cells. For projections to the MGsg, GABAergic cells make only a minimal contribution, suggesting that the tectothalamic projections to this part of the MG is almost exclusively excitatory. In the following sections, we discuss technical aspects of our analysis, compare our findings to previous descriptions of the parallel tectothalamic pathways, and then consider some functional implications of these pathways and the differential contributions of GABAergic projections.</p><sec id="s4-1"><title>Technical Considerations</title><p>Small volumes of tracer were used to ensure that deposits were located within a single MG subdivision. Such confined injections allow analysis without contamination by labeling cells that project to a neighboring subdivision. Of course, none of the restricted injections filled an MG subdivision entirely, so there is a risk that projections that terminate in only a portion of a subdivision could go undetected. The small injection volumes may also risk incomplete labeling because of limits to the sensitivity of tracers. We tried to minimize these limitations by using multiple tracers (green beads, red beads and FluoroGold). The beads are particularly valuable in this regard because of their high sensitivity and limited diffusibility in the tissue (Schofield, <xref rid="B34" ref-type="bibr">2008</xref>). These characteristics allow relatively large amounts of tracer to be restricted to a small volume of tissue, resulting in many labeled cells. The fact that we obtained similar results across animals and across tracers, and that larger injections (involving multiple MG subdivisions) were consistent with the results from the small injections, suggests that our results are generally valid.</p><p>The GAD antibody used here has been validated in previous studies in guinea pigs (Xiong et al., <xref rid="B40" ref-type="bibr">2008</xref>; Nakamoto et al., <xref rid="B27" ref-type="bibr">2013</xref>; Mellott et al., <xref rid="B24" ref-type="bibr">2014</xref>) and we believe that our tissue contained few false positive cells. Incomplete penetration of immunoreagents can lead to false negative staining, which could substantially affect quantitative analyses. We systematically limited our analysis such that, for each tissue section, labeled structures were analyzed only at tissue depths that included robust immunostaining. We conclude that GAD+ cells are GABAergic and that the GAD-negative cells are almost certainly non-GABAergic. Both anatomical and physiological data argue that the GAD-negative cells are glutamatergic. First, nearly all IC cells appear to be GABAergic or glutamatergic (Ito et al., <xref rid="B19" ref-type="bibr">2011</xref>; Ito and Oliver, <xref rid="B20" ref-type="bibr">2012</xref>). Second, stimulation of IC inputs to the MG can be blocked completely by pharmacological blockade of glutamate and GABA (Peruzzi et al., <xref rid="B30" ref-type="bibr">1997</xref>). We conclude that most or all of the GAD-negative tectothalamic cells are glutamatergic.</p></sec><sec id="s4-2"><title>Parallel pathways in guinea pigs</title><p>The concept of parallel pathways in the upper auditory system is often traced to a seminal report by Calford and Aitkin (<xref rid="B8" ref-type="bibr">1983</xref>). These authors based their findings on the patterns of tectothalamic connections in cats, which they related to thalamocortical projections described in earlier studies. The concept thus encompasses pathways from midbrain to forebrain and has proven attractive and widely accepted (e.g., de Ribaupierre, <xref rid="B10" ref-type="bibr">1997</xref>; Rouiller, <xref rid="B33" ref-type="bibr">1997</xref>; Hu, <xref rid="B17" ref-type="bibr">2003</xref>; Wenstrup, <xref rid="B38" ref-type="bibr">2005</xref>). Extension of this concept to other species has often been based on thalamocortical (and corticothalamic) relationships (reviewed by, Rouiller, <xref rid="B33" ref-type="bibr">1997</xref>), with relatively less information on the tectothalamic projections. Support for the existence of parallel pathways in guinea pigs comes from evidence for anatomical and physiological differences between the MG subdivisions and for differences in thalamocortical connections (Redies et al., <xref rid="B32" ref-type="bibr">1989</xref>; Redies and Brandner, <xref rid="B31" ref-type="bibr">1991</xref>; Edeline et al., <xref rid="B12" ref-type="bibr">1999</xref>; He, <xref rid="B13" ref-type="bibr">2001</xref>, <xref rid="B14" ref-type="bibr">2003</xref>; Anderson et al., <xref rid="B5" ref-type="bibr">2007</xref>). The present data indicate that tectothalamic projections in guinea pigs reflect the organization described in other species. The MGv and lemniscal pathway receive majority input from the ICc, the MGd/MGsg and the diffuse pathway get inputs mostly from the ICd, and the MGm and associated polysensory pathway get substantial inputs from both the IClc and the ICc. In all cases, smaller projections arise from the non-dominant IC subdivisions, providing potential opportunities for cross-talk between the parallel pathways (Figure <xref ref-type="fig" rid="F2">2</xref>).</p></sec><sec id="s4-3"><title>GABAergic and glutamatergic components of the parallel pathways</title><p>Our analysis of glutamatergic vs. GABAergic components of the tectothalamic pathway lead to a few additional conclusions. First, the glutamatergic projections are numerically dominant and, not surprisingly, closely match the overall projection patterns (i.e., the subdivision-specific connections). The GABAergic projections are smaller, comprising 4–21% of the projections to a given MG subdivision. These projections largely reflect the overall parallel pathways, with the notable difference that very few GABAergic cells project to the MGsg (this difference is one basis for distinguishing the MGsg from the MGd and the other MG subdivisions; this is discussed in more detail below). Except for the MGsg, each MG subdivision receives both glutamatergic and GABAergic projections from the same subset of IC subdivisions. What function(s) are served by this convergence of excitation and inhibition?</p><p>Inhibitory inputs to the MG, like those to other regions of the auditory system, are considered critical for temporal processing of acoustic signals (Venkataraman and Bartlett, <xref rid="B37" ref-type="bibr">2013</xref>). <italic>In vitro</italic> studies have demonstrated that ascending excitatory and inhibitory projections (presumed tectothalamic inputs) converge on cells in multiple MG subdivisions (Bartlett and Smith, <xref rid="B6" ref-type="bibr">1999</xref>, <xref rid="B7" ref-type="bibr">2002</xref>; Smith et al., <xref rid="B35" ref-type="bibr">2007</xref>; Venkataraman and Bartlett, <xref rid="B37" ref-type="bibr">2013</xref>). In general, excitation or inhibition can arrive first, suggesting that the inhibitory inputs could influence both the onset and sustained portions of neuronal responses to sound. Moreover, different patterns of convergence occur, with some cells dominated by excitation, others dominated by inhibition (without any sign of ascending excitation) and the remaining cells showing more evenly mixed interactions. Temporal processing could be expected to play critical roles in all the parallel auditory pathways (Lennartz and Weinberger, <xref rid="B22" ref-type="bibr">1992</xref>; Abrams et al., <xref rid="B1" ref-type="bibr">2011</xref>), and in fact inhibitory/excitatory convergence has been seen in all MG subdivisions. However, the MG subdivisions differ in the relative numbers of cells that show the different patterns of excitatory and inhibitory interaction (discussed in Smith et al., <xref rid="B35" ref-type="bibr">2007</xref>). The present results show differences in the excitatory and inhibitory projections from specific IC subdivisions to four large MG subdivisions, supporting the concept of parallel pathways and the conclusion that the MG subdivisions serve distinct functions.</p></sec><sec id="s4-4"><title>Crosstalk between parallel pathways</title><p>The presence of non-dominant projections, i.e., small projections that connect IC and MG subdivisions less heavily than the dominant projections, have been noted since the earliest descriptions of parallel pathways (e.g., Calford and Aitkin, <xref rid="B8" ref-type="bibr">1983</xref>). Such cross-talk could allow for integration of information carried in different pathways, or allow activity in one pathway to influence processing in another pathway. Our results from GAD staining show that both GABAergic cells and glutamatergic cells contribute to the non-dominant connections described above. Thus, crosstalk between the pathways could include both excitatory and inhibitory components. A common role of inhibitory projections in many brain areas is lateral inhibition, and one might predict that the GABAergic projections serve to heighten the contrast between various channels and promote transmission through a particular channel. Both GABAergic and glutamatergic projections could allow for integration of information carried in the different channels. Such speculations await further insights into the role of the GABAergic and glutamatergic projections within channels as well as through crosstalk projections.</p><p>As mentioned above, the near absence of GABAergic tectothalamic projections distinguishes the MGsg from the other MG subdivisions. Previous studies have distinguished the MGsg based on connections with other regions of the brainstem, especially regarding strong projections to the MGsg from the superior colliculus (Tanaka et al., <xref rid="B36" ref-type="bibr">1985</xref>; Hicks et al., <xref rid="B15" ref-type="bibr">1986</xref>; Hoshino et al., <xref rid="B16" ref-type="bibr">2010</xref>) and the sagulum (Morest, <xref rid="B26" ref-type="bibr">1965</xref>). Examination of these regions in our experiments suggests that the MGsg in guinea pigs receives similar inputs. These connections suggest that the MGsg may play a role in integrating auditory and visual information and contribute to orientation or attention. The present results suggest that auditory tectothalamic contributions to these functions are carried out mainly by excitatory projections.</p><p>In summary, the present data suggest that tectothalamic projections in guinea pigs can be conceptualized by the same parallel pathways described in other species. Both excitatory and inhibitory projections contribute to these pathways and may provide a basis for more refined definitions and more complete understanding of the interactions of these pathways in the thalamus. The concept of parallel pathways has proven valuable for understanding many aspects of sensory processing. An interesting question for future work will be to determine the extent to which the current concept of parallel auditory pathways can accommodate new data on both subcortical and cortical connections of the MG.</p></sec></sec><sec id="s5"><title>Author contributions</title><p>Designed research, wrote the paper: Jeffrey G. Mellott, Brett R. Schofield; performed research, analyzed data: all authors.</p></sec><sec id="s6"><title>Conflict of interest statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec>
The crimson conundrum: heme toxicity and tolerance in GAS	<p>The massive erythrocyte lysis caused by the Group A Streptococcus (GAS) suggests that the β-hemolytic pathogen is likely to encounter free heme during the course of infection. In this study, we investigated GAS mechanisms for heme sensing and tolerance. We compared the minimal inhibitory concentration of heme among several isolates and established that excess heme is bacteriostatic and exposure to sub-lethal concentrations of heme resulted in noticeable damage to membrane lipids and proteins. Pre-exposure of the bacteria to 0.1 μM heme shortened the extended lag period that is otherwise observed when naive cells are inoculated into heme-containing medium, implying that GAS is able to adapt. The global response to heme exposure was determined using microarray analysis revealing a significant transcriptome shift that included 79 up regulated and 84 down regulated genes. Among other changes, the induction of stress-related chaperones and proteases, including <italic>groEL/ES</italic> (8x), the stress regulators <italic>spxA2</italic> (5x) and <italic>ctsR</italic> (3x), as well as redox active enzymes were prominent. The heme stimulon also encompassed a number of regulatory proteins and two-component systems that are important for virulence. A three-gene cluster that is homologous to the <italic>pefRCD</italic> system of the Group B Streptococcus was also induced by heme. PefR, a MarR-like regulator, specifically binds heme with stoichiometry of 1:2 and protoporphyrin IX (PPIX) with stoichiometry of 1:1, implicating it is one of the GAS mediators to heme response. In summary, here we provide evidence that heme induces a broad stress response in GAS, and that its success as a pathogen relies on mechanisms for heme sensing, detoxification, and repair.</p>	<contrib contrib-type="author"><name><surname>Sachla</surname><given-names>Ankita J.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/102929"/></contrib><contrib contrib-type="author"><name><surname>Le Breton</surname><given-names>Yoann</given-names></name><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/191085"/></contrib><contrib contrib-type="author"><name><surname>Akhter</surname><given-names>Fahmina</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/184055"/></contrib><contrib contrib-type="author"><name><surname>McIver</surname><given-names>Kevin S.</given-names></name><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/17779"/></contrib><contrib contrib-type="author"><name><surname>Eichenbaum</surname><given-names>Zehava</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="author-notes" rid="fn001"><sup>*</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/102907"/></contrib>	Frontiers in Cellular and Infection Microbiology	<sec sec-type="introduction" id="s1"><title>Introduction</title><p>Heme is vital to many biological systems, ranging from planktonic marine microorganisms to the highly complex and evolved humans. The heterocyclic ring in heme, protoporphyrin IX (PPIX), co-ordinates metal iron (Fe<sup>+2</sup>); this architectural arrangement supports a wide array of roles upon presentation of an appropriate microenvironment. Hemoglobin and myoglobin, for example, use heme as a prosthetic moiety for the transport and storage of oxygen and other diatomic gases. A multitude of redox enzymes engage heme as a catalyst of electron transfer. Heme is also associated with energy conservation and biotransformation in processes such as photosynthesis and respiration. In addition, signal transduction pathways are integrated via heme due to its ability to bind gaseous ligands such as oxygen, carbon dioxide, carbon monoxide, and nitric oxide (Heinemann et al., <xref rid="B30" ref-type="bibr">2008</xref>; Chiabrando et al., <xref rid="B16" ref-type="bibr">2014</xref>; Hogle et al., <xref rid="B32" ref-type="bibr">2014</xref>). Notably, heme in hemoproteins constitutes a major iron reserve used by invading microorganisms during infection (Brown and Holden, <xref rid="B13" ref-type="bibr">2002</xref>; Nobles and Maresso, <xref rid="B59" ref-type="bibr">2011</xref>). Blood in particular is an immediate source of iron and heme for the majority of pathogens. Invading bacteria are often able to obtain heme directly from the circulating pool of hemoproteins. Some pathogens also deploy hemolysins to trigger erythrocyte destruction and hemoglobin release (Nobles and Maresso, <xref rid="B59" ref-type="bibr">2011</xref>; Kozarov, <xref rid="B39" ref-type="bibr">2012</xref>; Caza and Kronstad, <xref rid="B14" ref-type="bibr">2013</xref>).</p><p>Ironically, while heme is essential for many functions, the transitional nature of the coordinated iron renders it a significant pro-oxidant that can harm many cellular entities including DNA, proteins, cytoskeleton, and membranes (Maines and Kappas, <xref rid="B50" ref-type="bibr">1975</xref>; Solar et al., <xref rid="B68" ref-type="bibr">1990</xref>; Chiabrando et al., <xref rid="B16" ref-type="bibr">2014</xref>). To prevent cellular damage and to control infections, mammals restrict the pool of free circulating hemoproteins through scavenger proteins such as serum albumin, hemopexin (for heme), and heptoglobin (for hemoglobin) (Solar et al., <xref rid="B69" ref-type="bibr">1989</xref>; Krishnamurthy et al., <xref rid="B40" ref-type="bibr">2007</xref>). To minimize oxidative damage, the cellular levels of heme are managed by heme oxygenases. These enzymes degrade heme to biliverdin, which is then reduced to the cytoprotectant molecule bilirubin (Khan and Quigley, <xref rid="B37" ref-type="bibr">2011</xref>). Since heme could act as a menace, it is advantageous only in amounts that meet the bacterial requirements and, as long as it remains sequestered, preventing access to susceptible macromolecules (Anzaldi and Skaar, <xref rid="B4" ref-type="bibr">2010</xref>). Bacteria that are exposed to heme and take advantage of it during infection had to develop strategies to manage its toxic effects. While we do not fully understand how bacteria avoid and manage the negative ramifications of heme exposure, recent studies have begun to shed light on some of the molecular mechanisms that are involved in this process (Anzaldi and Skaar, <xref rid="B4" ref-type="bibr">2010</xref>; Fernandez et al., <xref rid="B25" ref-type="bibr">2010</xref>; Wakeman et al., <xref rid="B84" ref-type="bibr">2014</xref>).</p><p>Heme tolerance in bacteria is typically based on tight regulation of heme uptake and degradation. Often bacteria employ Fur- or DtxR-like metallorepressors to orchestrate heme metabolism in response to iron availability (Nobles and Maresso, <xref rid="B59" ref-type="bibr">2011</xref>; Wakeman and Skaar, <xref rid="B83" ref-type="bibr">2012</xref>). In some cases, auxiliary defense systems that facilitate repair, detoxification, and expelling of heme surplus are activated when the cellular mechanisms for heme homeostasis are overwhelmed. A number of two-component systems (TCS) that coordinate the response to heme stress have been described in bacteria. The Heme Sensor System (HssRS) is a TCS that activates the expression of a heme efflux transporter (<italic>hrtAB</italic>) in <italic>Staphylococcus aureus, Bacillus anthracis</italic>, and possibly the Group B Streptococcus (GBS) (Torres et al., <xref rid="B77" ref-type="bibr">2007</xref>; Stauff and Skaar, <xref rid="B71" ref-type="bibr">2009</xref>; Fernandez et al., <xref rid="B25" ref-type="bibr">2010</xref>). In response to heme pressure, the ChrAS TCS from <italic>Corynebacterium diphtheria</italic> inhibits the heme biosynthetic gene, <italic>hemA</italic>, and activates the transcription of the heme oxygenase gene, <italic>hmuO</italic>, required for heme catabolism (Bibb et al., <xref rid="B10" ref-type="bibr">2005</xref>, <xref rid="B11" ref-type="bibr">2007</xref>; Ito et al., <xref rid="B34" ref-type="bibr">2009</xref>; Heyer et al., <xref rid="B31" ref-type="bibr">2012</xref>). Interestingly, in addition to putative HssRS TCS, GBS also codes for <italic>pefR</italic>, a heme-responding repressor (Fernandez et al., <xref rid="B25" ref-type="bibr">2010</xref>). PefR binding to heme leads to derepression of two heme export systems, namely <italic>pefAB</italic> and <italic>pefCD</italic> (Fernandez et al., <xref rid="B25" ref-type="bibr">2010</xref>). It is suggested that GBS uses PefR to fine-tune heme homeostasis, while it deploys the HssRS system in response to high heme exposure.</p><p>As mentioned above, bacteria often detoxify, using active export systems that directly eliminate accumulated heme or its toxic byproducts. The efflux system, MtrCDE, in <italic>Neisseria gonorrhea</italic>, is crucial for bacterial resistance to hydrophobic agents and heme; overexpression of this machinery leads to increased bacterial burden in vaginal fluids (Hagman et al., <xref rid="B28" ref-type="bibr">1995</xref>; Bozja et al., <xref rid="B12" ref-type="bibr">2004</xref>). GBS mutants in <italic>pefAB</italic> or <italic>pefCD</italic> genes are over-sensitive to heme and demonstrate increased accumulation of intracellular heme and PPIX (Fernandez et al., <xref rid="B25" ref-type="bibr">2010</xref>). Respiration-linked studies in <italic>Lactococcus lactis</italic> incriminated the <italic>hrtAB</italic> homologous genes <italic>ygfAB</italic> in heme efflux (Pedersen et al., <xref rid="B62" ref-type="bibr">2008</xref>). A family of cytoplasmic proteins that bind and sequester heme forms another facet of heme tolerance. The proteins HemS of <italic>Yersinia enterocolitica</italic>, ShuS of <italic>Shigella dysenteriae</italic>, PhuS of <italic>Pseudomonas aeruginosa</italic>, and HmuS of <italic>Y. pestis</italic> are established members of this family (Stojiljkovic and Hantke, <xref rid="B72" ref-type="bibr">1994</xref>; Thompson et al., <xref rid="B76" ref-type="bibr">1999</xref>; Wilks, <xref rid="B87" ref-type="bibr">2001</xref>; Wyckoff et al., <xref rid="B89" ref-type="bibr">2005</xref>; Lansky et al., <xref rid="B42" ref-type="bibr">2006</xref>; Lechardeur et al., <xref rid="B45" ref-type="bibr">2010</xref>). Interestingly, the enzymes SodC of <italic>Haemophilus ducreyi</italic> and AhpC of the GBS have both heme sequestration roles in addition to enzymatic activity (Negari et al., <xref rid="B57" ref-type="bibr">2008</xref>; Lechardeur et al., <xref rid="B45" ref-type="bibr">2010</xref>). Bacteria also degrade heme to alleviate toxicity. Heme oxygenases were described in several pathogenic bacteria (Zhang et al., <xref rid="B91" ref-type="bibr">2011</xref>; Uchida et al., <xref rid="B79" ref-type="bibr">2012</xref>; Nambu et al., <xref rid="B55" ref-type="bibr">2013</xref>; Wilks and Heinzl, <xref rid="B88" ref-type="bibr">2014</xref>). Despite the clear role heme metabolism has in pathogenesis, our understanding of heme tolerance is lacking in numerous pathogens.</p><p>The Group A Streptococcus (GAS) is a Gram-positive obligate human pathogen. GAS infections range from mild diseases such as pharyngitis and impetigo to invasive and systemic manifestations, including necrotizing fasciitis and streptococcal toxic shock syndrome. GAS can also produce post-infection complications such as glomerulonephritis and rheumatic fever (Cunningham, <xref rid="B18" ref-type="bibr">2008</xref>). In the absence of a vaccine, GAS infections are commonly treated with β-lactam antibiotics. However, due to the swift progression rate and extent of tissue damage, surgical intervention is often required to manage invasive GAS infections (Cole et al., <xref rid="B17" ref-type="bibr">2011</xref>). The grave nature of invasive GAS infection is attributed to many virulence factors, including its β –hemolytic property that causes massive erythrocyte and tissue lysis (Nizet, <xref rid="B58" ref-type="bibr">2002</xref>). GAS requires iron for growth and the pathogen can fulfill its needs for the metal by acquiring heme from lysing erythrocytes (Eichenbaum et al., <xref rid="B22" ref-type="bibr">1996</xref>; Montanez et al., <xref rid="B53" ref-type="bibr">2005</xref>). Heme acquisition is mediated by the <italic>sia</italic> (<italic>streptococcal iron acquisition</italic>) operon that facilitates heme relay and transport across the bacterial membrane (Bates et al., <xref rid="B9" ref-type="bibr">2003</xref>; Liu and Lei, <xref rid="B47" ref-type="bibr">2005</xref>; Nygaard et al., <xref rid="B60" ref-type="bibr">2006</xref>; Sook et al., <xref rid="B70" ref-type="bibr">2008</xref>; Ouattara et al., <xref rid="B61" ref-type="bibr">2010</xref>). <italic>In vivo</italic> studies on a second operon named <italic>siu</italic> (<italic>streptococcal iron uptake</italic>) linked it to the uptake of heme and ferric ion (Montanez et al., <xref rid="B53" ref-type="bibr">2005</xref>). Although heme receptors and transport proteins have been identified, our understanding of heme metabolism in GAS is still incomplete. The focus of this study was to fill in the knowledge gaps in GAS heme tolerance. Here, we show that excess heme is inhibitory for the growth of GAS <italic>in vitro</italic>; heme exposure caused a global transcriptome shift wherein it significantly up regulated genes that are important for redox stress, which includes sensing and management along with protein damage and rescue.</p></sec><sec sec-type="materials\|methods" id="s2"><title>Materials and methods</title><sec><title>Bacterial strains and growth conditions</title><p>Strains and plasmids used in this study are listed in Table <xref ref-type="table" rid="T1">1</xref>. GAS was grown under static conditions at 37°C in Todd–Hewitt broth with 0.2% (wt/vol) yeast extract (THY broth, Difco Laboratories) or in C-medium consisting of 0.5% Proteose Peptone #3 (BD), 1.5% yeast extract (BD), 10 mM K<sub>2</sub>HPO<sub>4</sub>, 0.4 mM MgSO<sub>4</sub>, 17 mM NaCl, adjusted to pH 7.5 (Lyon et al., <xref rid="B49" ref-type="bibr">1998</xref>). In some experiments, hemin chloride (Sigma) from stock solutions prepared in DMSO was added to the growth media at different concentrations. <italic>E. coli</italic> cells were used for cloning and protein expression purposes. <italic>E. coli</italic> was grown aerobically in a Luria–Bertani medium (pH 7.0) supplemented with kanamycin (30 μg/ml) at 37°C.</p><table-wrap id="T1" position="float"><label>Table 1</label><caption><p><bold>Strains used in this study</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th valign="top" align="left" rowspan="1" colspan="1"><bold>Name</bold></th><th valign="top" align="left" rowspan="1" colspan="1"><bold>Description</bold></th><th valign="top" align="left" rowspan="1" colspan="1"><bold>References</bold></th></tr></thead><tbody><tr><td valign="top" align="left" colspan="3" rowspan="1"><bold>STRAINS</bold></td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">MGAS5005</td><td valign="top" align="left" rowspan="1" colspan="1">GAS (M1), isolated from cerebrospinal fluid</td><td valign="top" align="left" rowspan="1" colspan="1">Sumby et al., <xref rid="B73" ref-type="bibr">2005</xref></td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">NZ131</td><td valign="top" align="left" rowspan="1" colspan="1">GAS (M49) isolated from acute post-glomerulonephritis infection</td><td valign="top" align="left" rowspan="1" colspan="1">Mcshan et al., <xref rid="B51" ref-type="bibr">2008</xref></td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">GA01398</td><td valign="top" align="left" rowspan="1" colspan="1">GAS (M11).</td><td valign="top" align="left" rowspan="1" colspan="1">GA-EIP<xref ref-type="table-fn" rid="TN1"><sup></sup></xref></td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">GA06439</td><td valign="top" align="left" rowspan="1" colspan="1">GAS (M114)</td><td valign="top" align="left" rowspan="1" colspan="1">GA-EIP<xref ref-type="table-fn" rid="TN1"><sup></sup></xref></td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">GA02581</td><td valign="top" align="left" rowspan="1" colspan="1">GAS (M1)</td><td valign="top" align="left" rowspan="1" colspan="1">GA-EIP<xref ref-type="table-fn" rid="TN1"><sup></sup></xref></td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">GA02582</td><td valign="top" align="left" rowspan="1" colspan="1">GAS (M1)</td><td valign="top" align="left" rowspan="1" colspan="1">GA-EIP<xref ref-type="table-fn" rid="TN1"><sup></sup></xref></td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">GA10156</td><td valign="top" align="left" rowspan="1" colspan="1">GAS (M75)</td><td valign="top" align="left" rowspan="1" colspan="1">GA-EIP<xref ref-type="table-fn" rid="TN1"><sup></sup></xref></td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">COL (MRSA)</td><td valign="top" align="left" rowspan="1" colspan="1"><italic>S. aureus</italic>, clinical specimen isolated from operating theater in England</td><td valign="top" align="left" rowspan="1" colspan="1">Gill et al., <xref rid="B26" ref-type="bibr">2005</xref></td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"><italic>E. cloni</italic><sup>®</sup> 10G</td><td valign="top" align="left" rowspan="1" colspan="1">Host for pAJS11 propagation and expression</td><td valign="top" align="left" rowspan="1" colspan="1">Lucigen</td></tr><tr><td valign="top" align="left" colspan="3" rowspan="1"><bold>PLASMIDS</bold></td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">pRham™ N-His</td><td valign="top" align="left" rowspan="1" colspan="1">Protein expression vector, Kan<sup>R</sup>, expressed from rhaP<sub>BAD</sub></td><td valign="top" align="left" rowspan="1" colspan="1">Lucigen</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">pAJS11</td><td valign="top" align="left" rowspan="1" colspan="1">Expresses N-Terminal His-tag PefR, KanR</td><td valign="top" align="left" rowspan="1" colspan="1">This study</td></tr></tbody></table><table-wrap-foot><fn id="TN1"><label></label><p><italic>Georgia Emerging Infections Program (GA-EIP)</italic>.</p></fn></table-wrap-foot></table-wrap></sec><sec><title>Determination of the minimal inhibitory concentration (MIC)</title><sec><title>Agar dilution method</title><p>This procedure was performed as described in Wiegand et al. (<xref rid="B85" ref-type="bibr">2008</xref>) with minor modifications. Briefly, the turbidity of overnight grown cultures in C-media was adjusted to OD<sub>600 nm</sub> = 0.08–0.1, representing the turbidity of 0.5 MacFarland's acidified barium chloride standard as described (McFarland, <xref rid="B93" ref-type="bibr">1907</xref>) and spotted (1 μL) onto C-media agar containing varying concentrations of heme. The plates were incubated overnight at 37°C and the minimal heme concentration that did not allow for significant bacterial growth was determined.</p></sec><sec><title>Broth macrodilution method</title><p>This method was performed as per the instruction of Wiegand et al. (<xref rid="B85" ref-type="bibr">2008</xref>). In brief, THYB containing heme at a range of 0–100 μM in 5 μM intervals was inoculated with GAS cells (OD<sub>600 nm</sub> = 0.05) and incubated at 37°C for 17 h. The minimal heme concentration that did not support growth (OD<sub>600 nm</sub> ≤ 0.2) was determined.</p></sec><sec><title>Disc diffusion method</title><p>This procedure was adapted from Drew et al. (<xref rid="B20" ref-type="bibr">1972</xref>). Briefly, sterile Whatman filter paper discs (8.0 mm diameter and 1.2 mm width) were submerged in 10 mM heme (in DMSO) and impregnated onto C-media agar that was plated with 0.1 ml of GAS culture (final OD<sub>600 nm</sub> = 0.1). The plates were incubated at 37°C for 17 h and the zone of clearance around the discs were measured.</p></sec></sec><sec><title>Thiobarbituric acid-reactive-substances (TBARS) assay for lipid damage</title><p>GAS culture samples with equal cell numbers were harvested 30, 60, and 90 min following treatment with 4 μM heme (in 0.035% DMSO) or with mock (0.035% DMSO, control) and washed twice with phosphate buffer saline, pH 7.4 (PBS). The resulting cell pellets were resuspended in 5 ml of PBS with lysozyme (1 mg/ml, Sigma) and 400 U of mutanolysin (sigma) and incubated at 37°C for 30 min. The cells were then subjected to sonication (5 s, 10% amplitude). TBARS formation in the membrane samples was quantified using the TBARS assay kit (ZeptoMetrix Corporation) according to the manufacture's recommendations. Briefly, 100 μl cell lysate samples were treated with 100 μl of SDS and 2.5 ml of the TBA reagent and incubated at 95°C for 60 min. The sample supernatant was collected following 10 min incubation on ice by centrifugation (15 min at 835 × g). The absorbance at 532 nm (indicative of TBARS) was measured using the DU 730 Life Science UV/Vis spectrophotometer. The amount of TBARS in the experimental samples was derived from a standard curve generated using the malondialdehyde (MDA) reagent supplied with the kit.</p></sec><sec><title>Detection of protein damage</title><p>The TBARS assay was followed as above except the cell pellets were washed twice with TSM buffer (100 mM Tris, 500 mM sucrose, 10 mM MgCl<sub>2</sub>, pH 7.0), resuspended in 0.5 ml of TSM with 400 U of Mutanolysin (Sigma) and incubated at 37°C for 30 min. The protoplasts were centrifuged at 20,000 × g for 5 min at 4°C, suspended in 0.2 ml of lysis buffer (50 mM Tris-HCl, 60 mM KCl, 10 mM MgCl<sub>2</sub>, pH 7.0, 2% β-mercaptoethanol) and subjected to sonication (10 s, 10% amplitude). The membrane components were collected by centrifugation at 100,000 × g for 30 min at 4°C. The supernatant was retained as a cytoplasmic fraction and pellets were resuspended in 0.2 ml of lysis buffer and were considered as the membrane fractions. Protein damage was detected using the OxyBlot™ protein oxidation detection kit (Millipore) according to the manufacturer's instructions. In this assay, oxidized proteins are derivatized with 2,4-dinitrophenylhydrazine (DNP), which is then detected with primary antibodies specific for DNP and secondary antibody conjugated to horseradish peroxidase (HRP) using a chemiluminescence protocol. The intensity of the signal in individual lanes was quantified using ImageQuant LAS 4000 and the ImageQuant TL software (GE).</p></sec><sec><title>RNA extraction</title><p>Cell samples were harvested 30, 60, and 90 min following heme or mock treatment by centrifugation (5000 × g for 20 min) at 4°C. RNA was extracted from the cell pellet using the RiboPure™ RNA purification kit (Ambion) followed by DNaseI digestion (Ambion) performed according to manufacturer's instructions. Microarray analysis was performed with total RNA extracted from 90 min post treatment cell samples. Real-time RT-PCR transcript analysis for selected genes was performed using total RNA extracted from all three-time points post treatments.</p></sec><sec><title>Microarray analysis and real-time RT-PCR validation</title><p>The GAS microarray used in this study (Ribardo and McIver, <xref rid="B64" ref-type="bibr">2006</xref>) consists of 2328 70-mer oligonucleotide probes targeting unique non-repetitive ORFs from the sequenced genomes of serotypes M1 (SF370), M3 (MGAS315), and M18 (MGAS8232). Probes were synthesized by Qiagen Operon with a melting temperature of 76 ± 5°C, ≤70% cross-hybridization identity to another gene within the same strain, ≤20 contiguous bases in common with another gene, and probe location within 3′ end of ORF. Microarrays were printed at Microarrays Inc. (<ext-link ext-link-type="uri" xlink:href="http://www.microarrays.com">http://www.microarrays.com</ext-link>), with 10 pl of each oligonucleotide probe spotted onto slides (UltraGAPS2; Corning) using a 12-pin contact printer. The microarray hybridization was performed as previously described (Jiang et al., <xref rid="B35" ref-type="bibr">2008</xref>). Briefly, 10 μg of DNase I-treated total RNAs to be compared were used for reverse transcription into single-stranded cDNA using 200 U Superscript II reverse transcriptase (Life Technologies), 6 μg random hexamers, 1X first strand buffer, 10 mM dithiothreitol (DTT), 0.5 mM dATP, 0.5 mM dCTP, 0.5 mM dGTP, 0.3 mM dTTP, and 0.2 mM of amino-allyl dUTP. The mixture was incubated at 42°C for 2 h and the reaction stopped by addition of 10 μl 0.5 M EDTA and 1 M NaOH. Amine-modified cDNA was purified by ethanol precipitation followed by chemical labeling with Cy3- or Cy5-NHS-ester fluorescent dyes (GE Healthcare) in a final step. Yield and incorporation of dye was determined using a Nanodrop ND-1000 (Thermo Scientific). Slides were pre-hybridized in a 50 ml solution of 5X SSC, 0.1% SDS and 1% BSA for 30 min at 42°C, washed 4X in water and once in isopropanol, then dried by brief centrifugation. Labeled probes were resuspended in hybridization buffer (30% formamide, 5X SSC, 0.1% SDS, 0.6 μg/μL salmon sperm DNA) and hybridized to the microarray slides in a 42°C water bath for 16–20 h. Slides were washed twice in a low stringency buffer (2X SSC, 0.1% SDS) at 55°C for 5 min, twice in a medium stringency buffer (0.1X SSC, 0.1% SDS) at room temperature for 5 min and finally twice in a high stringency buffer (0.1X SSC) at room temperature for 5 min, and then dried by brief centrifugation. Synthesized cDNA from each RNA sample from three independent cell cultures was hybridized on three separate microarray slides (biological replicates), and independently synthesized cDNA from each of these same RNA samples was hybridized in a repeat dye-swap experiment (technical replicates) to test technical reproducibility. Slides were scanned using an Axon 4100A personal array scanner and GenePixPro 6.0 software (Molecular Devices). Data obtained from MGAS5005 cells incubated in the presence of DMSO or 4 μM of heme were compared for twofold changes in expression, ≥2.0 or ≤0.50 with Acuity 4.0 software (Molecular Devices). Using a ratio-based normalization, data were normalized by the ratio of the means (635/532) and samples were removed when four out of the six experiments did not show significance. Data was validated on 11 independent genes by real-time RT-PCR as described below. Correlation coefficients for the arrays were determined by plotting the log value of the array on the x-axis to the log value of the real-time RT-PCR on the y-axis. An equation determining the line of best fit was determined, and the resulting R<sup>2</sup> value was calculated to be 0.889. Array data was submitted to the Gene Expression Omnibus (GEO) at the National Center for Biotechnology Information under the accession number GSE61415.</p></sec><sec><title>Quantification of expression by real-time RT-PCR</title><p>For microarray data validation, real-time RT-PCR analysis was carried out using primers in Table <xref ref-type="table" rid="T2">2</xref> as follows: 25 ng of DNaseI-treated total RNAs were added to SYBR green master mix (AB) containing 200 nM of each specific real-time primer for the one-step protocol. The real-time RT-PCR experiments were completed using a LightCycler 480 instrument (Roche), and levels represent the ratio of non-treated to treated experimental values relative to the level of expression of <italic>gyrA</italic> transcript as an endogenous control. For expression of the <italic>pefRC</italic> operon, a SYBR Green based quantitative PCR reactions were performed using the Power SYBR® Green RNA-to-Ct™ 1-Step Kit (AB) on 7500 Fast Real-Time PCR machine (AB) according to the manufacturer's specifications. Briefly, the reaction mixture (20 μl) contained, 10 μl of RT-PCR Mix (final 1X), 200 nM of each forward and reverse primers, 25 ng of total RNA, and RT enzyme Mix in 1X final concentrations. The relative quantification with comparative ΔΔCT method was employed to calculate differences in <italic>pefRC</italic> expression levels at different times after heme exposure. The relative expression levels of <italic>pefR</italic> and <italic>pefC</italic> genes were normalized to the level of <italic>rpsL</italic> transcript as an endogenous control. The levels of transcripts in heme treated samples were compared to the control samples.</p><table-wrap id="T2" position="float"><label>Table 2</label><caption><p><bold>Primers used in this study</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th valign="top" align="left" rowspan="1" colspan="1"><bold>Target</bold></th><th valign="top" align="left" rowspan="1" colspan="1"><bold>PCR primers</bold></th><th valign="top" align="left" rowspan="1" colspan="1"><bold>Sequence (5′-3′)</bold></th></tr></thead><tbody><tr><td valign="top" align="left" rowspan="1" colspan="1"><italic>pefR</italic> ORF</td><td valign="top" align="left" rowspan="1" colspan="1">ZE515</td><td valign="top" align="left" rowspan="1" colspan="1">CATCATCACCACCATCACTCACAAGTGATAGGTGATTTACG</td></tr><tr><td rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">ZE516</td><td valign="top" align="left" rowspan="1" colspan="1">GTGGCGGCCGCTCTATTAAGCATCGTTGTCTCCTTTATAA</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">P<sub><italic>pef</italic></sub></td><td valign="top" align="left" rowspan="1" colspan="1">ZE561</td><td valign="top" align="left" rowspan="1" colspan="1">AAGGCGTTCCCAAGAGAGCTAG</td></tr><tr><td rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">ZE562</td><td valign="top" align="left" rowspan="1" colspan="1">CCTTGAGGACCTGCTAGATGCTCTAC</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"><italic>pefR</italic></td><td valign="top" align="left" rowspan="1" colspan="1">ZE569</td><td valign="top" align="left" rowspan="1" colspan="1">CAATGTGATGCCTTCCCAAG</td></tr><tr><td rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">ZE570</td><td valign="top" align="left" rowspan="1" colspan="1">CGCTGTCAGCAACTCTTC</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"><italic>pefC</italic></td><td valign="top" align="left" rowspan="1" colspan="1">ZE571</td><td valign="top" align="left" rowspan="1" colspan="1">CCTCATCATGGGTTTGGTG</td></tr><tr><td rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">ZE572</td><td valign="top" align="left" rowspan="1" colspan="1">ACGCCACGGAGATTTTCC</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"><italic>rpsL</italic></td><td valign="top" align="left" rowspan="1" colspan="1">ZE581</td><td valign="top" align="left" rowspan="1" colspan="1">CAGATTCACCAGCTTTGAAC</td></tr><tr><td rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">ZE582</td><td valign="top" align="left" rowspan="1" colspan="1">CAACACGAGTAGCAACG</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">prsA</td><td valign="top" align="left" rowspan="1" colspan="1">prsA-M1-RT-L</td><td valign="top" align="left" rowspan="1" colspan="1">GGGCAGACTTTGCAGCTATTG</td></tr><tr><td rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">prsA-M1-RT-R</td><td valign="top" align="left" rowspan="1" colspan="1">TCGCCTGAGTCAAACGTATAGG</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"><italic>clpL</italic></td><td valign="top" align="left" rowspan="1" colspan="1">clpL-M1-RT-L</td><td valign="top" align="left" rowspan="1" colspan="1">TGGCTTGAGCTAAACCTTCA</td></tr><tr><td rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">clpL-M1-RT-R</td><td valign="top" align="left" rowspan="1" colspan="1">CTTGGCACGACGAACTAAAA</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"><italic>opuAA</italic></td><td valign="top" align="left" rowspan="1" colspan="1">opuAA-M1-RT-L</td><td valign="top" align="left" rowspan="1" colspan="1">TGATTTGCAAGACAGCATGA</td></tr><tr><td rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">opuAA-M1-RT-L</td><td valign="top" align="left" rowspan="1" colspan="1">CATCAAAGCAATCCGATCAC</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"><italic>endoS</italic></td><td valign="top" align="left" rowspan="1" colspan="1">endoS-M1-RT-L</td><td valign="top" align="left" rowspan="1" colspan="1">CTCGGTCAATAGCGTAGGAGAAG</td></tr><tr><td rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">endoS-M1-RT-L</td><td valign="top" align="left" rowspan="1" colspan="1">GCGTGCCGAACGGTATG</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"><italic>ptsA</italic></td><td valign="top" align="left" rowspan="1" colspan="1">ptsA-M1-RT-L</td><td valign="top" align="left" rowspan="1" colspan="1">TTTTTTAAAACCAGGCGAAGC</td></tr><tr><td rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">ptsA-M1-RT-L</td><td valign="top" align="left" rowspan="1" colspan="1">TTGTCTCAGGGACCAAATCC</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"><italic>sagA</italic></td><td valign="top" align="left" rowspan="1" colspan="1">sagA-M1-RT-L</td><td valign="top" align="left" rowspan="1" colspan="1">GCTACTAGTGTAGCTGAAACAACTCAA</td></tr><tr><td rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">sagA-M1-RT-L</td><td valign="top" align="left" rowspan="1" colspan="1">AGCAACAAGTAGTACAGCAGCAA</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"><italic>spt7R</italic></td><td valign="top" align="left" rowspan="1" colspan="1">spt7R-M1-RT-L</td><td valign="top" align="left" rowspan="1" colspan="1">TCATTTGCGGCTGAAATAATAATG</td></tr><tr><td rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">spt7R-M1-RT-L</td><td valign="top" align="left" rowspan="1" colspan="1">GCAATGGGATTCAATTTTTGGA</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"><italic>slo</italic></td><td valign="top" align="left" rowspan="1" colspan="1">slo-M1-RT-L</td><td valign="top" align="left" rowspan="1" colspan="1">TTGTTGAGGATAATGTAAGAATGTTTAG</td></tr><tr><td rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">slo-M1-RT-R</td><td valign="top" align="left" rowspan="1" colspan="1">TCCTGGCTTGCAACTGATTG</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"><italic>fasC</italic></td><td valign="top" align="left" rowspan="1" colspan="1">fasC-M1-RT-L</td><td valign="top" align="left" rowspan="1" colspan="1">TGCGCACAAATTATGAAATATCTTC</td></tr><tr><td rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">fasC-M1-RT-R</td><td valign="top" align="left" rowspan="1" colspan="1">GAGCTTCAAGCAATTTGGAATTC</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"><italic>mtsA</italic></td><td valign="top" align="left" rowspan="1" colspan="1">mtsA-M1-RT-L</td><td valign="top" align="left" rowspan="1" colspan="1">TGAGGGTCTTGACCGATTG</td></tr><tr><td rowspan="1" colspan="1"/><td valign="top" align="left" rowspan="1" colspan="1">mtsA-M1-RT-R</td><td valign="top" align="left" rowspan="1" colspan="1">AAGTCGTGGCAACCAATTC</td></tr></tbody></table></table-wrap></sec><sec><title><italic>In silico</italic> analysis</title><p>The PefR binding motif was identified within the putative promoter region of MGAS5005 <italic>spy_0195</italic> (<italic>pefR</italic>) using Multiple Em for Motif Elicitation (MEME) suit hosted by the National Biomedical Computation Resource (Bailey et al., <xref rid="B7" ref-type="bibr">2006</xref>). The outcome was further analyzed by the MAST application of MEME for its genome wide occurrence within the upstream sequences of GAS under both the stringent (<italic>E</italic>-value ≤ 0.01) and relaxed (<italic>E</italic>-value ≤ 10) parameters. All of the sequences were acquired from Kyoto Encyclopedia of Genes and Genomes (KEGG) database (Kanehisa, <xref rid="B36" ref-type="bibr">2002</xref>) for comparative sequence analysis; sequences were aligned using ClustalW (Thompson et al., <xref rid="B75" ref-type="bibr">2002</xref>).</p></sec><sec><title>Cloning, overexpression, and purification of PefR</title><p>The <italic>spy_0195</italic> ORF was amplified from GAS MGAS5005 genomic DNA by PCR using primers ZE515 and ZE516 (Table <xref ref-type="table" rid="T2">2</xref>). To generate pAJS11 plasmid (Table <xref ref-type="table" rid="T1">1</xref>), the PCR fragment with <italic>pefR</italic> ORF was cloned into the pRham™ expression vector (Lucigen) by Expressioneering™ technology and introduced into <italic>E. cloni®</italic> 10G competent cells. The cloning was confirmed by sequence analysis. For PefR expression, cells harboring pAJS11 were induced at OD<sub>600 nm</sub> = 0.8 with 0.2% rhamnose. The cells were harvested (8000 × g for 5 min at 4°C) following 16 h incubation at 28°C. The resulting pellets were resuspended in extraction buffer (20 mM Tris pH 8, 100 mM NaCl, 0.1% Triton X-100) containing 0.5 mg/mL Complete, mini-EDTA-free protease inhibitor cocktail (Roche), sonicated and the cell debris were removed by centrifugation. The resulting lysate was purified over 5 ml HisTrap™ HP (GE) nickel affinity column. Protein fractions were dialyzed overnight in sodium phosphate buffer (SPB: 20 mM sodium phosphate, 500 mM NaCl pH 7.4). Expression of the recombinant PefR was evaluated by SDS-PAGE and western blot analysis with mouse anti-His antibodies (Sigma).</p></sec><sec><title>Heme and PPIX binding assay</title><p>Heme binding by PefR was assessed spectroscopically as described in Puri and O'brian (<xref rid="B63" ref-type="bibr">2006</xref>) and Ouattara et al. (<xref rid="B61" ref-type="bibr">2010</xref>) with minor modifications. Briefly, an increasing concentration of hemin chloride (2–30 μM) was added to both the test cuvette containing 10 μM of PefR protein in SPB and the reference cuvette (containing SPB alone) and the changes in absorbance across the wavelength of 250–700 nm region were recorded every 6 min. The PefR to heme stoichiometry and dissociation constant (<italic>K<sub>d</sub></italic>) were determined by plotting the absorbance at 435 nm as a function of the heme concentrations. The extinction coefficient (ε<sub>max</sub>) for PefR was calculated from the hemocromogen method described in Asakura et al. (<xref rid="B6" ref-type="bibr">1964</xref>). All of the spectroscopic measurements were made using the DU730 Life Science UV/Vis spectrophotometer (Beckman Coulter). PPIX binding was tested by titrating PefR (5 μM) with 0–6 μM PPIX (in 0.5 μM increments) prepared in acidified methanol:dimethylformamide (1:1). The changes in absorbance were recorded.</p></sec></sec><sec sec-type="results" id="s3"><title>Results</title><sec><title>Excess of heme is inhibitory to the growth of GAS</title><p>Host heme is the immediate source of iron during infection for invading pathogens such as GAS. Recent studies demonstrated that bacteria must maintain the intracellular levels of heme at equilibrium in order to benefit from its nutritional value, while eluding the toxicity that is associated with heme overload (Torres et al., <xref rid="B77" ref-type="bibr">2007</xref>; Fernandez et al., <xref rid="B25" ref-type="bibr">2010</xref>; Mike et al., <xref rid="B52" ref-type="bibr">2014</xref>). In this work, we began evaluating the impact of heme on GAS physiology. We found that the addition of a disc saturated with 10 mM heme onto agar plates seeded with a confluent lawn of GAS resulted in a zone of clearance similar to those observed with antibiotic-impregnate discs. This observation indicates that heme can have a bacteriostatic effect on GAS (Table <xref ref-type="table" rid="T3">3</xref>). To compare the sensitivity to heme among GAS isolates we determined the MIC values of heme using two common methods (Table <xref ref-type="table" rid="T3">3</xref>). The agar dilution assay employs solid C-medium containing varying concentrations of heme. We determined that the heme MIC in this method is between 10 and 20 μM with the nephritogenic M49 GAS skin isolate, NZ131 (Simon and Ferretti, <xref rid="B66" ref-type="bibr">1991</xref>; Mcshan et al., <xref rid="B51" ref-type="bibr">2008</xref>), exhibiting the highest sensitivity of all GAS strains tested. The highly pathogenic M1T1 MGAS5005 strain (Sumby et al., <xref rid="B73" ref-type="bibr">2005</xref>) as well as clinical isolates from patients with invasive diseases (Table <xref ref-type="table" rid="T1">1</xref>) demonstrated higher heme tolerance. In the broth dilution assay, we measured bacterial growth in THYB supplemented with varying heme concentrations. This method resulted in higher values of heme MIC (30–50 μM) for MGAS5005 and the clinical isolates. However, both methods indicated that the invasive strains were more resistant to heme than NZ131.</p><table-wrap id="T3" position="float"><label>Table 3</label><caption><p><bold>Bacteriostatic effect of heme</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th valign="top" align="left" rowspan="1" colspan="1"><bold>Strain<xref ref-type="table-fn" rid="TN2"><sup>a</sup></xref></bold></th><th valign="top" align="center" rowspan="1" colspan="1"><bold>Agar dilution (μM)</bold></th><th valign="top" align="center" rowspan="1" colspan="1"><bold>Broth dilution (μM)</bold></th><th valign="top" align="center" rowspan="1" colspan="1"><bold>Disc diffusion (mm)</bold></th></tr></thead><tbody><tr><td valign="top" align="left" rowspan="1" colspan="1">NZ131 (M49)</td><td valign="top" align="center" rowspan="1" colspan="1">10</td><td valign="top" align="center" rowspan="1" colspan="1">10</td><td valign="top" align="left" rowspan="1" colspan="1">9</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">GA01398 (M11)</td><td valign="top" align="center" rowspan="1" colspan="1">20</td><td valign="top" align="center" rowspan="1" colspan="1">30</td><td valign="top" align="char" char="." rowspan="1" colspan="1">1.25</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">GA06439 (M114)</td><td valign="top" align="center" rowspan="1" colspan="1">20</td><td valign="top" align="center" rowspan="1" colspan="1">30</td><td valign="top" align="char" char="." rowspan="1" colspan="1">6.75</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">GA02581 (M1)</td><td valign="top" align="center" rowspan="1" colspan="1">20</td><td valign="top" align="center" rowspan="1" colspan="1">50</td><td valign="top" align="left" rowspan="1" colspan="1">6</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">GA02582 (M1)</td><td valign="top" align="center" rowspan="1" colspan="1">15</td><td valign="top" align="center" rowspan="1" colspan="1">50</td><td valign="top" align="char" char="." rowspan="1" colspan="1">4.75</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">GA10156 (M75)</td><td valign="top" align="center" rowspan="1" colspan="1">20</td><td valign="top" align="center" rowspan="1" colspan="1">30</td><td valign="top" align="char" char="." rowspan="1" colspan="1">4.5</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1">MGAS5005 (M1)</td><td valign="top" align="center" rowspan="1" colspan="1">20</td><td valign="top" align="center" rowspan="1" colspan="1">50</td><td valign="top" align="char" char="." rowspan="1" colspan="1">5.5</td></tr><tr><td valign="top" align="left" rowspan="1" colspan="1"><italic>S. aureus (control)</italic></td><td valign="top" align="center" rowspan="1" colspan="1">260</td><td valign="top" align="center" rowspan="1" colspan="1">80</td><td valign="top" align="char" char="." rowspan="1" colspan="1">4.25</td></tr></tbody></table><table-wrap-foot><p><italic>The minimal heme concentration that inhibits bacterial growth (MIC) was determined by the methods of agar dilution and broth macrodilution. The growth inhibitory effect of heme is also expressed as the clearance zone around heme impregnate discs (10 mM). See material and methods for details. Data are representative of two independent replicates</italic>.</p><fn id="TN2"><label>a</label><p><italic>GAS M type (if available) is indicated in parentheses</italic>.</p></fn></table-wrap-foot></table-wrap></sec><sec><title>Exposure to sub-lethal heme levels cause lipid peroxidation in GAS</title><p>To learn why heme is detrimental for GAS growth, we investigated its impact on GAS macromolecules. Heme is reported to damage directly or through the generation of reactive oxidative species the integrity of membrane lipids (Chang et al., <xref rid="B15" ref-type="bibr">2003</xref>; Chiabrando et al., <xref rid="B16" ref-type="bibr">2014</xref>). Therefore, we evaluated the extent of lipid peroxidation in GAS following heme exposure. Using low heme concentration that could still support cell growth when applied at the early logarithmic phase allowed us to examine the kinetics of damage production and later the bacterial transcriptome response to heme. As seen in Figure <xref ref-type="fig" rid="F1">1A</xref>, the addition of 4 μM heme to MGAS5005 at the early logarithmic phase of growth did not result in a significant change in the bacterial growth profile. To detect lipid peroxidation, we used an established method that relies on the tendency of oxidized lipids to react with a thiobarbituric acid (TBA) reagent to form adducts (named TBARS) that absorb at 532 nm (Hong et al., <xref rid="B33" ref-type="bibr">2012</xref>). The time course of TBARS production in GAS cells exposed to heme (or in control samples) was quantified using a standard curve (Figure <xref ref-type="fig" rid="F1">1B</xref>). Significantly higher occurrence of lipid peroxidation was observed in all of the heme-treated samples compared with control samples (Figure <xref ref-type="fig" rid="F1">1C</xref>). Lipid peroxidation was the highest (22.5 fold over background) 30 min after heme treatment; while lower levels (~5 fold over background) were found in the samples collected 60 and 90 min post-heme treatment (Figure <xref ref-type="fig" rid="F1">1C</xref>), indicating possible repair mechanism. In eukaryotes, membrane repair involves enzymes such as phospholipase A2 that hydrolyse peroxidized fatty acid esters and glutathione peroxidase (van Kuijk et al., <xref rid="B80" ref-type="bibr">1987</xref>; Thomas et al., <xref rid="B74" ref-type="bibr">1990</xref>).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Heme exposure leads to lipid peroxidation in GAS</bold>. <bold>(A)</bold> Cell growth. MGAS5005 cells growing in C-media were treated with 4 μM heme in 0.035% DMSO (Heme) or with mock treatment (0.035% DMSO, Control) at the early logarithmic growth phase. Culture samples were harvested 30, 60 and 90 min post exposure and processed. <bold>(B)</bold> Malondialdehyde (MDA) standard curve. 100 μl samples of MDA at a concentration of 0, 12.5, 25, 50, and 100 nmol/ml were allowed to react with thiobarbituric acid (TBA, see Materials and Methods). The absorption at 532 nm of the supernatants from processed samples was determined and plotted as a function of MDA concentration exposure. The linear equation obtained for the standard curve was <italic>y</italic> = 0.0059 × with R<sup>2</sup> value of 0.9997. <bold>(C)</bold> Lipid peroxidation following heme exposure. Cell lysates were prepared and allowed to react with TBA. The sample absorption at 532 nm was determined and the formation of TBA-reactive-substances (TBARS) was calculated using the standard curve shown in <bold>(B)</bold>. All samples were standardized with respect to the cell number in the corresponding culture. Data are representative of biological triplicates.</p></caption><graphic xlink:href="fcimb-04-00159-g0001"/></fig></sec><sec><title>Exposure to heme damaged membrane proteins</title><p>We next evaluated the effect of heme on GAS proteins. Oxidation introduces carbonyl groups into protein side chains that can then react with 2,4-dinitrophenylhydrazine (DNPH) to generate 2,4-dinitrophenylhydrazone (DNP) derivatives (Dalle-Donne et al., <xref rid="B19" ref-type="bibr">2003</xref>). To monitor protein oxidation in GAS, membrane and cytoplasmic proteins were extracted from culture samples collected at 30, 60, and 90 min post treatment and allowed to react with DNPH. Oxidation was then visualized by immunoblot with antibodies specific for DNP and the damage was quantified by densitometry. Analysis of the membrane fractions revealed that the anti-DNP antibody strongly reacted with all of the samples from heme exposed cells (Figure <xref ref-type="fig" rid="F2">2A</xref>, upper panel). While general staining confirmed the presence of proteins in all samples (Figure <xref ref-type="fig" rid="F2">2A</xref>, bottom panel), protein oxidation was only observed in the heme-treated cells in the early sample. At the later time points (60 and 90 min) some background oxidation also occurred in the control samples. In all cases though, a significantly higher amount of protein damage was observed in the heme-exposed cells compared to the controls (Figure <xref ref-type="fig" rid="F2">2C</xref>). In addition, the degree of protein oxidation was the highest in the 90 min samples. Similar analysis of cytoplasmic proteins did not reveal detectable damage (Figure <xref ref-type="fig" rid="F2">2B</xref>); therefore, protein oxidation following heme exposure is primarily linked to the membrane fraction under our experimental conditions.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Protein oxidation following heme exposure</bold>. The membrane <bold>(A)</bold> and cytoplasmic proteins <bold>(B)</bold> extracted from cells harvested at different time points post treatment were allowed to react with DNPH (see Materials and Methods) before fractionation by SDS-PAGE. Sample identity (H for heme treated cells and C for controls) and collection times (30, 60, and 90 min) are indicated (top panel). Western blot with anti-DNP antibodies (OxyBlot) developed by chemiluminescence (bottom panel). Coomassie blue-stained gel. All samples were standardized with respect to the cell number in the corresponding culture. <bold>(C)</bold> Quantification of protein oxidation. The reaction with anti-DNP antibody in each membrane sample was quantified by densitometry. Arbitrary values for the heme treated samples (Heme) and mock treated (Control) after background subtraction are plotted. Error bars are shown. Data are representative of biological triplicates. Note that the western blot showing larger spots in an area outside of the lane is a technical artifact associated with the film development and it's not a true reaction (prominent in <bold>B</bold>, top panel).</p></caption><graphic xlink:href="fcimb-04-00159-g0002"/></fig></sec><sec><title>GAS demonstrates adaptation response to heme</title><p>When overnight cultures of NZ131 grown in heme-free medium were inoculated into medium containing 1 μM heme, which is below the growth inhibiting concentrations at the early logarithmic phase, an extended lag phase that lasted for more than 8 h was observed (Figure <xref ref-type="fig" rid="F3">3</xref>). Although exposure to heme inhibited growth initially, the cultures reached after overnight incubation the cell density observed in cultures that grew in heme-free medium (data not shown). To test for possible adaptation to environmental heme, GAS cells were grown overnight in the presence of 0.1 μM heme before they were inoculated into medium with 1 μM heme. Indeed, pre-exposure to low heme level eliminated the extended lag period (Figure <xref ref-type="fig" rid="F3">3</xref>). These observations suggest that the treatment with low level of heme-triggered adaptation to environmental heme.</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>GAS employs an adaptation strategy to avert growth phenotypes associated with heme stress</bold>. NZ131 cells were grown overnight in C-media supplemented with 0.1 μM (adapted culture) and without heme (non-adapted culture). The overnight cultures were used as an inoculum to monitor growth in presence and absence of heme. Briefly, non-adapted culture was grown in C-media supplemented with 0 and 1 μM heme, whereas adapted culture was sub-cultured into C-media containing 0.1 and 1 μM heme. The growth was monitored colorimetrically for 8 h at 37°C.</p></caption><graphic xlink:href="fcimb-04-00159-g0003"/></fig><p>In light of recent evidence that bacteria can sense and respond to heme (Fernandez et al., <xref rid="B25" ref-type="bibr">2010</xref>; Lechardeur et al., <xref rid="B44" ref-type="bibr">2012</xref>; Mike et al., <xref rid="B52" ref-type="bibr">2014</xref>; Wakeman et al., <xref rid="B84" ref-type="bibr">2014</xref>) and with evidence for a similar response in GAS, we wanted to gain insight into the pathogen's mechanisms for heme sensing and tolerance. We examined the differential global gene expression pattern between GAS cells subjected to heme stress (4 μM) and control treatment. Total RNA was isolated from cells 90 min post treatment and analyzed using a 70-mer oligonucleotide microarray (representing GAS genomes M1, M3, and M18) (Ribardo and McIver, <xref rid="B64" ref-type="bibr">2006</xref>). Our data showed a significant shift in the transcript profile of different genes and operons in response to heme stimulus; referred to as the “GAS heme stimulon.” This stimulon involved changes in 163 total genes, including 79 genes that were up regulated and 84 genes that were down regulated (Figure <xref ref-type="fig" rid="F4">4</xref> and Table <xref ref-type="supplementary-material" rid="SM1">S1</xref>). The array analysis was validated by quantitative RT-PCR (qPCR) on 11 differentially regulated genes (Table <xref ref-type="supplementary-material" rid="SM1">S1</xref>), showing a strong correlation, with an R<sup>2</sup> value of 0.889 (Figure <xref ref-type="supplementary-material" rid="SM1">S1</xref>).</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>Genes responding to heme stress in M1T1 GAS strain MGAS5005</bold>. Total RNA was extracted 90 min post treatment and analyzed by microarray. <bold>(A)</bold> Genes repressed or activated following exposure to 4 μM heme (in 0.035% DMSO) compared with mock treatment (0.035% DMSO). Data for genes whose expression was significantly different between cells exposed to heme and those subjected to mock treatment were combined and assigned to categories. <bold>(B)</bold> Selected heme-activated genes revealing protein and redox stress response. Values represent relative expression levels (fold-change) in heme treated bacteria compare with mock treatment.</p></caption><graphic xlink:href="fcimb-04-00159-g0004"/></fig><p>Genes that were down regulated significantly in response to heme are found in the categories of metabolic pathways, sugar and metabolite transport, and two-component system (TCS, Figure <xref ref-type="fig" rid="F4">4A</xref>). Of particular interest is the down regulation of the TCS, <italic>trxSR</italic>. The response regulator, TrxR activates the Mga virulence regulon in GAS, and a <italic>trxR</italic> mutant is attenuated for virulence in a murine infection model (Leday et al., <xref rid="B46" ref-type="bibr">2008</xref>; Baruch et al., <xref rid="B8" ref-type="bibr">2014</xref>). Therefore, environmental heme may impact the expression of key virulence factors in GAS via the TrxSR/Mga pathway. Since the TrxSR system is implicated in asparagine sensing, our observations also imply that the damage to the cell envelope introduced by heme is associated with increased levels of asparagine (or other amino acids) (Baruch et al., <xref rid="B8" ref-type="bibr">2014</xref>). In contrast, heme-activated genes and highly expressed transcripts encode for protein folding and degradation (such as <italic>groEL/ES</italic> ~8 fold, <italic>DnaJ/K</italic> 3-4 fold, and <italic>clpE/L</italic> ~4 fold) and components of the Clp protease machinery (<italic>clpCP</italic>- 2.5 fold) (Lund, <xref rid="B48" ref-type="bibr">2001</xref>; Alexopoulos et al., <xref rid="B2" ref-type="bibr">2012</xref>) (Figure <xref ref-type="fig" rid="F4">4B</xref>). In addition, the GAS heme-activated stimulon includes genes whose functions are linked to regulation of redox stress management, including <italic>spxA2</italic> (5 fold) and <italic>ctsR</italic> (2.5 fold) (Elsholz et al., <xref rid="B23" ref-type="bibr">2010</xref>; Antelmann and Helmann, <xref rid="B3" ref-type="bibr">2011</xref>). Accordingly, members of the downstream genes of these regulators were also activated by heme such as thioredoxin (<italic>trx</italic> ~2.5 fold) and oxidoreductases (3 fold). Finally, the expression of two TCSs was also up regulated in response to heme exposure, including the pneumococcal-like TCS homologous system, <italic>ciaHR</italic> (~2.3 fold) (Ahn et al., <xref rid="B1" ref-type="bibr">2006</xref>) and the <italic>ihk/irr</italic> TCS (2.5 fold) involved in GAS response to oxidative stress and survival within phagocytes (Han et al., <xref rid="B29" ref-type="bibr">2012</xref>).</p></sec><sec><title>Identification of the <italic>pefRCD</italic> locus in GAS</title><p>The transcriptomic analysis also revealed that several putative efflux proteins also showed an increase in transcriptional activity in response to heme stress. These included a 3-gene cluster (MGAS5005 <italic>spy_0195, 0196</italic>, and <italic>0197)</italic> with homology to the <italic>pefRCD</italic> genes from GBS (Fernandez et al., <xref rid="B25" ref-type="bibr">2010</xref>). Since the <italic>pefRCD</italic> genes are important for the management of heme stress in GBS, we focused our attention on their heme-induced homologs in GAS. Comparative sequence analysis demonstrated that the putative GAS genes, <italic>spy_0195</italic> (84% similarity; 56% identity), <italic>spy_0196</italic> (76% similarity; 48% identity), and <italic>spy_0197</italic> (76% similarity; 51.3% identity) show significant sequence homology to the GBS <italic>pefRCD genes</italic>, respectively (Figure <xref ref-type="fig" rid="F5">5</xref>). In addition to the high sequence homology demonstrated by each of these GAS genes, the genomic arrangement of the GBS <italic>pefRCD</italic> locus and its immediate chromosomal location are also conserved in all of the published GAS genomes (Figure <xref ref-type="fig" rid="F6">6A</xref>). Moreover, using the MEME algorithm, we identified a 17-bp inverted repeat region within the putative promoter of <italic>spy_0195</italic> that shared 76% identity with the PefR binding site from GBS (Figure <xref ref-type="fig" rid="F6">6A</xref>). Based on this sequence conservation and the observed induction by heme, we hypothesized that like GBS, the above-mentioned GAS genes encode a heme dependent repressor (<italic>spy_0195</italic>) and an ABC heme exporter (<italic>spy_0196</italic> and <italic>spy_0197</italic>) and named them <italic>pefRCD</italic>, respectively.</p><fig id="F5" position="float"><label>Figure 5</label><caption><p><bold>Multiple sequence alignment of PefR amino acid sequence from GBS and GAS</bold>. The degree of homology between GAS PefR (encoded by MGAS5005 <italic>spy_0195</italic>) and GBS PefR (encoded by <italic>gbs1402</italic>) is depicted in pair-wise sequence alignment performed using ClustalW. The amino acids are represented by single letter code. The symbols: (<sup>*</sup>) indicate identical residues; (:) indicate strongly similar residues; (.) designate weakly similar residues. An accuracy of alignment was also confirmed using MUSCLE tool (Edgar, <xref rid="B21" ref-type="bibr">2004</xref>) that showed identical alignment output as well.</p></caption><graphic xlink:href="fcimb-04-00159-g0005"/></fig><fig id="F6" position="float"><label>Figure 6</label><caption><p><bold>The streptococcal <italic>pefRCD</italic> locus</bold>. <bold>(A)</bold> Genetic organization of the <italic>pefRCD</italic> gene cluster in GAS. The putative PefR binding identified <italic>in silico</italic> and its homology to the PefR binding motif from GBS (Fernandez et al., <xref rid="B25" ref-type="bibr">2010</xref>) are shown. <bold>(B)</bold> Time course of <italic>pefRC</italic> expression in response to heme stress. Total RNA was extracted at different time points following heme or mock treatment and the relative expression of the <italic>pefR</italic> and <italic>pefC</italic> genes was evaluated by Real-time quantitative RT-PCR.</p></caption><graphic xlink:href="fcimb-04-00159-g0006"/></fig><p>We examined the time course of <italic>pefRC</italic> expression in response to heme by qPCR (Figure <xref ref-type="fig" rid="F6">6B</xref>). As seen in the microarray analysis, the expression of <italic>pefR</italic> and <italic>C</italic> was comparable to one another in all of the time points, supporting an operon structure. Heme exposure resulted in an increase of <italic>pefR</italic> and <italic>pefC</italic> expression over time, reaching 2.5 fold over background at 30 min, 4.5 fold at 60 min and remained high at 90 min post treatment. This temporal increase in <italic>pef</italic> expression in response to heme insinuates an active participation of these genes in heme tolerance.</p></sec><sec><title>GAS PefR binds heme and PPIX with high affinity</title><p>In GBS, PefR is a MarR-like repressor that binds to DNA and blocks the transcription of both <italic>pefRCD</italic> and <italic>pefAB</italic> operons. In this system, heme acts as an inducer that binds to PefR protein to alleviate repression. The GAS protein shares 43% identity and 74% similarity with PefR from GBS (Figure <xref ref-type="fig" rid="F5">5</xref>), implying the two proteins may function in a similar manner. We cloned the <italic>pefR</italic> gene from GAS and expressed it as an N-terminal hexahistidine (6x-His) fusion protein. SDS-PAGE and western blot analysis of the recombinant protein following purification from <italic>E. coli</italic> confirmed the presence of a single protein at the expected size that reacts with anti-his tag antibodies (Figure <xref ref-type="fig" rid="F7">7A</xref> and data not shown). Interestingly, the purified PefR had a light brown color (Figure <xref ref-type="fig" rid="F7">7B</xref>, left inset) and exhibited an absorption in the Soret region (435 nm). Upon titration with free heme, PefR displayed a growing Soret band and concomitant increase in absorption at 530 and 560 nm (β and α bands of heme). These spectral features are characteristics of heme bound protein and are easily separated from the absorption displayed by heme that is free in solution (Zhu et al., <xref rid="B92" ref-type="bibr">2000</xref>). The holo-PefR maintained a bright red color after the removal of heme excess by gel filtration and dialysis (Figure <xref ref-type="fig" rid="F7">7B</xref>, right inset). Therefore, PefR was purified from <italic>E. coli</italic> with some heme and readily bound heme <italic>in vitro</italic>. The plot of changes in absorbance at 435 nm as a function of heme concentrations indicated a binding stoichiometry of 1:2 (PefR:heme, Figure <xref ref-type="fig" rid="F7">7C</xref>). We determined the dissociation constant (<italic>Kd</italic>) for PefR by fitting the plot of changes at 435 nm in the modified Hill's equation (Goutelle et al., <xref rid="B27" ref-type="bibr">2008</xref>). We found the <italic>Kd</italic> to be 10 μM and Hill's coefficient (<italic>nH)</italic> to be 1.3, greater than 1 (Yifrach, <xref rid="B90" ref-type="bibr">2004</xref>). The molar extinction co-efficient at 435 nm of holo-PefR was determined using the pyridine hemocromogen method (Asakura et al., <xref rid="B6" ref-type="bibr">1964</xref>) and resulted in ε<sub>435</sub> = 30,407 M<sup>−1</sup>cm<sup>−1</sup>. Similarly, titration of PefR with PPIX indicated that it could also bind PPIX in 1:1 stoichiometry with <italic>Kd</italic> of 9 μM (Figure <xref ref-type="fig" rid="F8">8</xref>).</p><fig id="F7" position="float"><label>Figure 7</label><caption><p><bold>The PefR from GAS is a heme binding protein. (A)</bold> A Coomassie blue-stained gel showing purified recombinant PefR from GAS next to a molecular marker. <bold>(B)</bold> Heme titration of PefR. UV-visible spectrum of 10 μM PefR (in SPB) following incubation with an increasing concentration of hemin chloride (2–30 μM) in 2 μM increments. SBP with the corresponding heme concentration was used as blank and subtracted from the sample spectrum. Heme binding by PefR is indicated by the increased absorption at 435, 530, and 560 nm. The arrow indicates the direction of the absorption changes. The images are of PefR as-purified from <italic>E. coli</italic> and of holo-PefR (after heme reconstitution and removal of heme excess by gel filtration and dialysis). This data represents three independent heme-titration experiments. <bold>(C)</bold> Stoichiometry of heme binding to PefR. The changes in absorbance at 435 nm were plotted against heme concentration. The <italic>Kd</italic> was determined by fitting the data into modified hills equation for multiple binding sites (Goutelle et al., <xref rid="B27" ref-type="bibr">2008</xref>). The <italic>Kd</italic> equals 10 μM and <italic>n</italic>H (hill's coefficient) is 1.37 μM.</p></caption><graphic xlink:href="fcimb-04-00159-g0007"/></fig><fig id="F8" position="float"><label>Figure 8</label><caption><p><bold>The PefR is capable of PPIX binding. (A)</bold> PPIX titration of PefR. Change in the UV-visible spectral profile of 5 μM of PefR (in SPB) when titrated with an increasing concentration of PPIX (0.25–6 μM) in 0.5 μM increments was monitored across wavelength (300–700 nm). Free PPIX spectrum (shown in black) showed absorbance peaks in 432, 521, 556, 596, 655, and 685 nm regions. The PefR bound PPIX demonstrated maximum absorbance at 437 and 686 nm; absorbance at these wavelengths showed increase with PPIX concentrations (0.25–6 μM; shown in different colors). The arrow indicates the direction of the absorption changes. <bold>(B)</bold> Stoichiometry of PPIX binding to PefR. The spectral changes at 437 nm were plotted against PPIX concentration. The <italic>Kd</italic> was determined by fitting the data into linear equation.</p></caption><graphic xlink:href="fcimb-04-00159-g0008"/></fig></sec></sec><sec sec-type="discussion" id="s4"><title>Discussion</title><p>It is the biochemical properties of heme that contribute to its characteristics as a double-edged sword in a variety of biological systems. Heme toxicity in humans stems from the lysis of erythrocytes due to disease, inflammation, or physical damage, which could raise the heme levels in the bloodstream up to 20 μM (Arruda et al., <xref rid="B5" ref-type="bibr">2004</xref>; Kumar and Bandyopadhyay, <xref rid="B41" ref-type="bibr">2005</xref>). The generation of reactive oxygen species, alteration in membrane permeability, damage to macromolecules, and decrease in the pool of reductants are some of the cellular mayhem that follows heme overload in eukaryotes (Kumar and Bandyopadhyay, <xref rid="B41" ref-type="bibr">2005</xref>). In bacteria, despite the importance of heme to physiology and pathogenesis, the molecular mechanisms for heme toxicity and tolerance are poorly understood (Anzaldi and Skaar, <xref rid="B4" ref-type="bibr">2010</xref>). In this study, we probed these processes in GAS for the first time and established that heme stress is very significant to the physiology of the β-hemolytic pathogen; heme in excess inhibits GAS growth and exposure to low heme levels damages membrane lipids and proteins. We demonstrated the presence of an adaptation process to environmental heme in GAS and revealed a comprehensive transcriptome response to excess heme exposure. Finally, our findings also implicate a new gene cluster, <italic>pefRCD</italic>, in GAS heme sensing and tolerance.</p><p>We found that heme tolerance in GAS depends on medium type, growth conditions and genetic background (Table <xref ref-type="table" rid="T3">3</xref>). Differences in bacterial metabolism, aeration, and cell-to-cell contact may all affect the bacterial sensitivity to heme. In addition, the THY medium is prepared from brain and heart infusion and is likely to contain some heme. Growth in THYB therefore, may allow for bacterial adaptation, possibly contributing to the higher heme MIC values observed in this medium. A significant variation in heme tolerance was exhibited by different strains of GAS; with NZ131, a serotype M49 skin isolate (Simon and Ferretti, <xref rid="B66" ref-type="bibr">1991</xref>; Mcshan et al., <xref rid="B51" ref-type="bibr">2008</xref>), demonstrating the lowest heme tolerance, whereas the more invasive strains such as M1T1 MGAS5005 (Sumby et al., <xref rid="B73" ref-type="bibr">2005</xref>) and other clinical isolates showing higher heme resistance. It is tempting to speculate that these variations in heme sensitivity might add to the inclination of different GAS strains to colonize certain sites and to their invasive potential.</p><p>The absence of significant damage to cytosolic proteins in GAS suggests that the membrane is the primary target of heme in this bacterium (Figures <xref ref-type="fig" rid="F1">1</xref>, <xref ref-type="fig" rid="F2">2</xref>). This observation is in accordance with findings made in <italic>S. aureus</italic> (Wakeman et al., <xref rid="B82" ref-type="bibr">2012</xref>) and is likely to result from heme accumulation in the cell envelope. The negative regulation of the heme uptake machinery works to limit the amount of heme that can reach the GAS cytoplasm (Bates et al., <xref rid="B9" ref-type="bibr">2003</xref>; Montanez et al., <xref rid="B53" ref-type="bibr">2005</xref>; Toukoki et al., <xref rid="B78" ref-type="bibr">2010</xref>). In addition to active import, heme can diffuse through biological membranes, albeit with slow diffusion rates that can lead to accumulation in this compartment. Indeed it is well established that heme surplus tends to accumulate in the membranes of the eukaryotic cells, where it can be transformed into highly reactive species (Krishnamurthy et al., <xref rid="B40" ref-type="bibr">2007</xref>; Chiabrando et al., <xref rid="B16" ref-type="bibr">2014</xref>). There is less information regarding heme localization in bacteria; however, studies in <italic>S. aureus</italic> suggest a similar buildup in the cell envelope (Skaar et al., <xref rid="B67" ref-type="bibr">2004</xref>). In addition to the preferential localization of heme outside of the cytoplasm, the reducing environment and the detox mechanism induced by heme (e.g., heat shock and redox factors) could mitigate some of the oxidative effects of the heme in this compartment. In this study we only probed the impact of heme at 4 μM, a concentration that is below the MIC level in early exponential phase, and for 90 min post treatment. Further investigations are needed to determine if heme at higher concentrations or after longer exposure time is affecting other cellular compartments.</p><p>The extended lag period that is observed when naïve GAS cells are introduced into medium containing heme and the absence of the growth delay in cells that were pre-exposed to low heme levels (Figure <xref ref-type="fig" rid="F3">3</xref>) provide strong evidence for the presence of heme sensing and coping mechanisms in GAS. Indeed, microarray analysis indicated that heme triggers comprehensive changes in GAS transcript levels (Table <xref ref-type="supplementary-material" rid="SM1">S1</xref> and Figure <xref ref-type="fig" rid="F4">4A</xref>). The shift in the expression of regulatory proteins such as <italic>spxA2, ihk/irr, ctsR</italic>, and <italic>hrcA</italic> and their downstream-regulated genes, suggests that heme leads to significant redox, oxidative, and protein damage (Figure <xref ref-type="fig" rid="F4">4B</xref>). Spx is an RNA polymerase binding protein with CXXC redox-sensing center that acts to restore thiol balance by controlling redox enzymes and antioxidants formation (Nakano et al., <xref rid="B54" ref-type="bibr">2003</xref>). The <italic>ihk/irr</italic> TCS regulates oxidative stress genes that protect GAS from killing by phagocytic cells and hydrogen peroxide (Voyich et al., <xref rid="B81" ref-type="bibr">2004</xref>; Han et al., <xref rid="B29" ref-type="bibr">2012</xref>). The induction of this regulatory system is consistent with ROS formation and cell envelope damage following heme exposure. The master regulators CtsR and HrcA, originally classified as class III and class I heat shock gene regulators, respectively, have been associated with a wide variety of stresses such as temperature shifts, antibiotics, carbonyl, electrophiles, etc. (Narberhaus, <xref rid="B56" ref-type="bibr">1999</xref>; Elsholz et al., <xref rid="B24" ref-type="bibr">2011</xref>). Here we show their association with heme stress and the elevated expression of chaperones and <italic>clp</italic> proteases further corroborates our hypothesis of heme-induced protein unfolding and findings of protein oxidation.</p><p>Heme treatment also resulted in significant inhibition of the genes encoding lactate oxidase (<italic>lctO</italic>) and of a V-type ATP synthases system (<italic>ntpKCFABD</italic>). LctO reaction leads to hydrogen peroxide production that can exacerbate the damage induced by heme (Seki et al., <xref rid="B65" ref-type="bibr">2004</xref>; Kietzman and Caparon, <xref rid="B38" ref-type="bibr">2010</xref>). The down regulation of the proton pump may serve to increase the reducing power at the cell surface. Further selective down regulation of genes such as <italic>citDEFX</italic> and up-regulation of <italic>fruRBA</italic>, which are involved in import and utilization of citrate and fructose, respectively, could indicate a metabolic state that is more conducive to detoxification.</p><p>Le Breton et al. recently used a <italic>mariner</italic>-based transposon library to screen for genes that are important for GAS survival in human blood (Le Breton et al., <xref rid="B43" ref-type="bibr">2013</xref>). Comparing the heme stimulon identified in this study to the findings of Le Breton et al. revealed significant overlap, including <italic>clpE</italic> and <italic>L, fhs.2, fruA, citF, trxS, lacZ</italic>, and the <italic>has</italic> operon. This comparison suggests that heme, directly or indirectly, act as a signal that mediates adaptation <italic>in vivo</italic> and that the management of heme toxicity is relevant to GAS pathogenesis, in particular during spread through the blood.</p><p>The heme stimulon in GAS also includes three genes (<italic>spy_0195-7)</italic> that share high homology to the GBS <italic>pefRCD</italic> genes. In GBS these genes code for a heme responsive repressor (PefR) from the MarR-like family (Wilkinson and Grove, <xref rid="B86" ref-type="bibr">2006</xref>) and to an ABC heme exporter (PefCD) (Fernandez et al., <xref rid="B25" ref-type="bibr">2010</xref>). While the presence and genomic arrangements of <italic>pefRCD</italic> are highly conserved among different GAS strains (Figures <xref ref-type="fig" rid="F5">5</xref>, <xref ref-type="fig" rid="F6">6</xref>), we failed to identify homologs to <italic>pefAB</italic> (and <italic>hrtAB</italic>) that also participate in heme tolerance in GBS and other Gram-positive bacteria. The high degree of sequence similarity and the temporal increase in expression in response to heme stress (Figure <xref ref-type="fig" rid="F6">6B</xref>) suggest that the 3-gene locus encodes for a true homolog of the <italic>pef</italic> operon. This hypothesis is further supported by the identification of a putative PefR heme-binding box that is highly comparable to that of PefR from GBS (Figure <xref ref-type="fig" rid="F5">5A</xref>) (Fernandez et al., <xref rid="B25" ref-type="bibr">2010</xref>) and by our heme and PPIX reconstitution experiments (Figures <xref ref-type="fig" rid="F7">7</xref>, <xref ref-type="fig" rid="F8">8</xref>). The ability of PefR to coordinate two heme molecules and curve fitting data demonstrating positive Hill's coefficient of cooperativity most likely suggest that the binding of heme at one binding site increases the affinity of heme binding at another site (Yifrach, <xref rid="B90" ref-type="bibr">2004</xref>; Goutelle et al., <xref rid="B27" ref-type="bibr">2008</xref>). Further experiments are underway to determine if and how heme and PPIX are modulating PefR binding to DNA and to establish the role of this 3-gene cluster in heme and protoporphyrin tolerance in GAS.</p><p>To the best of our knowledge our work is: (1) the first effort to characterize GAS mechanisms for heme toxicity and tolerance; (2) shows that environmental heme constitutes a significant redox stress in GAS with damaging effects on membrane lipids and proteins; (3) provides evidence for heme sensing and adaptation that involves global transcriptome shifts; and (4) links heme stress to several key regulators and TCS systems of this important human pathogen.</p><sec><title>Conflict of interest statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec></sec>
Hydrothermal synthesis and structural characterization of ammonium ion-templated lanthanide(III) carboxylate-phosphonates	<p>Using <italic>N</italic>-(phosphonomethyl) iminodiacetic acid (H<sub>4</sub>PMIDA), as a complexing agent, two new complexes, (NH<sub>4</sub>)La(PMIDA)(H<sub>2</sub>O)•H<sub>2</sub>O, <bold>1</bold> and (NH<sub>4</sub>)Yb(PMIDA), <bold>2</bold> have been synthesized hydrothermally. In both compounds, the metal ions are trapped by a three five-membered chelate rings by the chelating PMIDA anions giving a tricapped trigonal prismatic LaO<sub>8</sub>N and monocapped trigonal prismatic YbO<sub>6</sub>N geometries for <bold>1</bold> and <bold>2</bold>, respectively. The structure of <bold>1</bold> consists of La(PMIDA)(H<sub>2</sub>O) chelating units, linked together by the phosphonate oxygen atoms O1 and O3 to form a chain along the <italic>c</italic>-axis. The chains are then connected together by the bridging phosphonate oxygen O2 to form a 2D layered structure with alternating 4- and 8-membered apertures. The structure of <bold>2</bold> consists Yb(PMIDA) chelating units, which are connected by alternating bridging carboxylate and phosphonate groups along the [010] direction forming chains with a corrugated pattern. The third phosphonate oxygen bridges the chains together along the [001] direction to build the two-dimensional layer with 4- and 6-membered apertures in the <italic>bc</italic>-plane. Under excitation of 330 nm, compound <bold>2</bold> shows a broad emission band at λ<sub>max</sub> = 460 nm. This emission is essentially in the blue luminescent region, which corresponds to ligand centered fluorescence.</p>	<contrib contrib-type="author"><name><surname>Ayi</surname><given-names>Ayi A.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><xref ref-type="author-notes" rid="fn001"><sup></sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/175764"/></contrib><contrib contrib-type="author"><name><surname>Kinnibrugh</surname><given-names>Tiffany L.</given-names></name><xref ref-type="aff" rid="aff2"><sup>2</sup></xref></contrib><contrib contrib-type="author"><name><surname>Clearfield</surname><given-names>Abraham</given-names></name><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><xref ref-type="author-notes" rid="fn002"><sup></sup></xref></contrib>	Frontiers in Chemistry	<sec sec-type="introduction" id="s1"><title>Introduction</title><p>The complexing ability of phosphonic acids have been extensively exploited in the design and synthesis of metal-organophosphonate-type metal organic frameworks (MOFs) with the formation of new and interesting compounds (Clearfield, <xref rid="B12" ref-type="bibr">1988</xref>; Gagnon et al., <xref rid="B19" ref-type="bibr">2012</xref>). The great interest in metal phosphonates is not only for their many unusual structural features, but also for their potential applications in different fields including adsorption, separation, gas storage, catalysis, photoluminescence, and drug delivery (Zhang and Clearfield, <xref rid="B59" ref-type="bibr">1997</xref>; Sharma and Clearfield, <xref rid="B45" ref-type="bibr">2000</xref>; Lukes et al., <xref rid="B30" ref-type="bibr">2001</xref>; Bazaga-García et al., <xref rid="B7" ref-type="bibr">2014</xref>; Liu et al., <xref rid="B29" ref-type="bibr">2014</xref>). Recently, we have prepared and structurally characterized a molybdenum–diphosphonate coordination network, and shown that it can undergo reversible dehydration, which occurs with a structural change. The dehydrated material shows size selective adsorption of alcohols, adsorbing methanol but not ethanol (Ayi et al., <xref rid="B3" ref-type="bibr">2013</xref>). Furthermore, we have reported crystal structures of two anionic 3D frameworks using 1,3,5-benzenetriphosphonic acid (BTP) with small amines as counter cations, Zn<sub>2.5</sub>(H)<sub>0.4–0.5</sub>(C<sub>6</sub>H<sub>3</sub>O<sub>9</sub>P<sub>3</sub>)(H<sub>2</sub>O)<sub>1.9–2</sub>(NH<sub>4</sub>)<sub>0.5–0.6</sub>(ZBP-NH<sub>4)</sub>, and Zn<sub>2.5</sub>(H)<sub>0.75</sub>(C<sub>6</sub>H<sub>3</sub>O<sub>9</sub>P<sub>3</sub>)(H<sub>2</sub>O)<sub>2</sub>(CH<sub>3</sub>NH<sub>3</sub>)<sub>0.25</sub> (ZBP-CH<sub>3</sub>NH<sub>3</sub>), resulting from the hydrothermal synthesis using Zn(II) cations and BTP (Kinnibrugh et al., <xref rid="B26" ref-type="bibr">2013</xref>). The compounds were found to exhibit a reversible dehydration process with two phase transitions resulting in a loss of volume. Research has shown that poorly crystalline metal phosphonates are common and a combination of phosphonic acids with additional carboxylic functional groups gives variety of interesting crystalline compounds (Poojary et al., <xref rid="B39" ref-type="bibr">1994</xref>). The construction of multi-dimensional coordination polymers with carboxylate- and/ or phosphonate-based organic linkers has been the focus of many research groups (Galdecka et al., <xref rid="B20" ref-type="bibr">2000</xref>; Paz et al., <xref rid="B38" ref-type="bibr">2005</xref>; Ananias et al., <xref rid="B2" ref-type="bibr">2006</xref>; Soares-Santos et al., <xref rid="B51" ref-type="bibr">2006</xref>; Bao et al., <xref rid="B5" ref-type="bibr">2007</xref>; Cunha-Silva et al., <xref rid="B14" ref-type="bibr">2007</xref>; Girginova et al., <xref rid="B21" ref-type="bibr">2007</xref>; Paz and Klinowski, <xref rid="B37" ref-type="bibr">2007</xref>, <xref rid="B36" ref-type="bibr">2008</xref>; Tang et al., <xref rid="B54" ref-type="bibr">2007</xref>; Chelebaeva et al., <xref rid="B11" ref-type="bibr">2008</xref>; Ferreira et al., <xref rid="B18" ref-type="bibr">2008</xref>; Rodrigues et al., <xref rid="B40" ref-type="bibr">2008</xref>; Shi et al., <xref rid="B50" ref-type="bibr">2008</xref>). The use of highly flexible and/or aromatic-based organic moieties in conjunction with two or more chelating phosphonate groups has enabled the isolation of a number of structures in recent times (Serpaggi and Ferey, <xref rid="B42" ref-type="bibr">1998</xref>; Evans et al., <xref rid="B17" ref-type="bibr">2001</xref>; Groves et al., <xref rid="B23" ref-type="bibr">2005</xref>, <xref rid="B22" ref-type="bibr">2006</xref>; Ying and Mao, <xref rid="B57" ref-type="bibr">2006</xref>). In the last two decades, we and others have investigated the use of a versatile chelating organic ligand: <italic>N</italic>-(phosphonomethyl) iminodiacetic acid (H<sub>4</sub>PMIDA) in the synthesis of metal carboxylate-phosphonate hybrid compounds (Zhang et al., <xref rid="B61" ref-type="bibr">1996</xref>, <xref rid="B60" ref-type="bibr">1998</xref>; Gutschke et al., <xref rid="B24" ref-type="bibr">1999</xref>; Mao and Clearfield, <xref rid="B33" ref-type="bibr">2002</xref>; Almeida Paz et al., <xref rid="B1" ref-type="bibr">2004</xref>; Shi et al., <xref rid="B49" ref-type="bibr">2005</xref>, <xref rid="B48" ref-type="bibr">2006</xref>; Tang et al., <xref rid="B55" ref-type="bibr">2006a</xref>). Using H<sub>4</sub>PMIDA, as a complexing agent in the presence of phosphoric acid, a mixed phosphate phosphonate layered zirconium compound was obtained by our group (Zhang et al., <xref rid="B61" ref-type="bibr">1996</xref>). A linear chain compound was isolated when the reaction was carried out in the absence of phosphoric acid (Zhang et al., <xref rid="B60" ref-type="bibr">1998</xref>). In both cases, the iminodiacetic moieties are only involved in hydrogen bonding, and are available for further metal complexing. An antiferromagnet K<sub>2</sub>Co(PMIDA)}·xH<sub>2</sub>O, whose crystal structure features a hexameric ring in the chair conformation was isolated by Wood and co-workers through the interaction of the salt of cobalt with H<sub>4</sub>PMIDA as ligand (Gutschke et al., <xref rid="B24" ref-type="bibr">1999</xref>). Also with H<sub>4</sub>PMIDA ligand, we reported two divalent metal carboxylate-phosphonate hybrid compounds of composition [Co<sub>2</sub>(PMIDA)(H<sub>2</sub>O)<sub>5</sub>]·H<sub>2</sub>O and [Zn<sub>2</sub>(PMIDA)(CH<sub>3</sub>CO<sub>2</sub>H)]·2H<sub>2</sub>O. The structure of cobalt compound contains double layers of Co(II)carboxylate interconnected by layers of Co(II)phosphonate, while the crystal structure of zinc compound features a zinc carboxylate-phosphonate hybrid layer along the [202] plane (Mao and Clearfield, <xref rid="B33" ref-type="bibr">2002</xref>). Mao and co-workers isolated isostructural lanthanide carboxyphosphonates Ln(HPMIDA)(H<sub>2</sub>O)<sub>2</sub>•3H<sub>2</sub>O (Ln = Gd, Tb, Dy, Y, Er, Yb, Lu), based on H<sub>4</sub>PMIDA anion, which exhibit a three-dimensional (3D) open-framework structures with helical tunnels (Tang et al., <xref rid="B55" ref-type="bibr">2006a</xref>). Several research groups have constructed multi-dimensional frameworks by using [V<sub>2</sub>O<sub>2</sub>(PMIDA)<sub>2</sub>]<sup>4−</sup> anionic unit, as well as a one-dimensional coordination polymer containing H<sub>4</sub>PMIDA residues and Fe<sup>2+</sup> centers (Almeida Paz et al., <xref rid="B1" ref-type="bibr">2004</xref>; Shi et al., <xref rid="B49" ref-type="bibr">2005</xref>, <xref rid="B48" ref-type="bibr">2006</xref>). In continuation of our investigation of the flexible coordinating properties of H<sub>4</sub>PMIDA, as a ligand, we extended our research to the lanthanides system. The design and syntheses of porous lanthanide phosphonates are attractive in developing new materials with multifunctions (Song and Mao, <xref rid="B53" ref-type="bibr">2005</xref>; Ying and Mao, <xref rid="B57" ref-type="bibr">2006</xref>; Liu et al., <xref rid="B28" ref-type="bibr">2007</xref>; Mao, <xref rid="B32" ref-type="bibr">2007</xref>). The successful synthesis of the first open-framework lanthanide carboxyphosphonate Pr<sub>4</sub>(H<sub>2</sub>O)<sub>7</sub>(O<sub>3</sub>PCH<sub>2</sub>NC<sub>5</sub>H<sub>9</sub>COO)<sub>4</sub>(H<sub>2</sub>O), opened the way for a number of other lanthanide hybrid solids to be isolated (Massiot et al., <xref rid="B34" ref-type="bibr">1997</xref>; Legendziewicz et al., <xref rid="B27" ref-type="bibr">1998</xref>; Serre et al., <xref rid="B43" ref-type="bibr">2004</xref>; Bauer et al., <xref rid="B6" ref-type="bibr">2006</xref>; Tang et al., <xref rid="B55" ref-type="bibr">2006a</xref>,<xref rid="B56" ref-type="bibr">b</xref>; Huang et al., <xref rid="B25" ref-type="bibr">2007</xref>; Zhou et al., <xref rid="B62" ref-type="bibr">2010</xref>). For example, using lanthanide chlorides and <italic>N</italic>-(carboxymethyl)iminodi(methylphosphonic acid)(H<sub>5</sub>cmp) a series of layered [Ln(H<sub>2</sub>cmp)(H<sub>2</sub>O)] materials [where Ln<sup>3+</sup> = Y<sup>3+</sup>, La<sup>3+</sup>, Pr<sup>3+</sup>, Nd<sup>3+</sup>, Sm<sup>3+</sup>, Eu<sup>3+</sup>, Gd<sup>3+</sup>, Tb<sup>3+</sup>, Dy<sup>3+</sup>, Ho<sup>3+</sup>, and Er<sup>3+</sup>], and the mixed-lanthanide [(Gd<sub>0.95</sub>Eu<sub>0.05</sub>)(H<sub>2</sub>cmp)(H<sub>2</sub>O)] material, have been successfully isolated from hydrothermal synthesis as phase-pure micro-crystalline compounds, (Cunha-Silva et al., <xref rid="B13" ref-type="bibr">2009</xref>) and found to be supramolecular polymorphs of the compound that was reported by Mao and co-workers (Massiot et al., <xref rid="B34" ref-type="bibr">1997</xref>; Bauer et al., <xref rid="B6" ref-type="bibr">2006</xref>; Tang et al., <xref rid="B56" ref-type="bibr">2006b</xref>; Zhou et al., <xref rid="B62" ref-type="bibr">2010</xref>). Using phoshonoacetic acid as a complexing agent, a linear chain aluminum(III)carboxyphosphonate with ammonium ion as counter cation has been reported (Ayi et al., <xref rid="B4" ref-type="bibr">2011</xref>). The ammonium cations were generated <italic>in-situ</italic> from the partial decomposition of urea. By employing a similar technique with the rare earth elements, we have been able to isolate two new materials (NH<sub>4</sub>)[La(PMIDA)(H<sub>2</sub>O)]•H<sub>2</sub>O, <bold>1</bold>, and (NH<sub>4</sub>)[Yb(PMIDA)], <bold>2</bold>, exhibiting 2D structures. This paper reports the synthesis and characterization of these two lanthanide(III)carboxyphosphonates.</p></sec><sec><title>Experimental</title><sec><title>Synthesis and chemical analysis</title><p>The two compounds (NH<sub>4</sub>La(PMIDA)(H<sub>2</sub>O)•H<sub>2</sub>O, <bold>1</bold> and (NH<sub>4</sub>)Yb(PMIDA), <bold>2</bold> were hydrothermally synthesized (autogenous pressure for 6 days) at 160°C from a mixture of lanthanum chloride heptahydrate LaCl<sub>3</sub>.7H<sub>2</sub>O (Aldrich, 98%) for <bold>1</bold>, ytterbium oxide Yb<sub>2</sub>O<sub>3</sub> for <bold>2</bold>, HCl (Fisher Scientific), <italic>N</italic>-(phoshonomethyl)iminodiacetic acid, H<sub>4</sub>PMIDA(Aldrich, 95%), potassium acetate (Fisher Scientific), urea (EM Science), and H<sub>2</sub>O/dioxane in the molar ratio 1:2:1: 2: 2.2: 80/10. In a typical synthesis of <bold>1</bold>, LaCl<sub>3</sub>·7H<sub>2</sub>O (0.557 g, 1.5 mmol) was dispersed in 2 ml of water followed by the addition of 0.26 ml HCl and 0.84 ml 1,4-dioxane. To this mixture was added H<sub>4</sub>PMIDA (0.34g, 1.5 mmol), urea (0.20 g, 3.3 mmol) and potassium acetate (0.10 g, 1.0 mmol). The resulting suspension with a pH of 1 was sealed in a Teflon-lined steel autoclave and heated at 160°C for 6 days. The product, a crop of colorless plate-like crystals was filtered and washed with distilled water and dried at ambient temperature. Compound <bold>2</bold> was obtained similarly from the composition Yb<sub>2</sub>O<sub>3</sub>(0.28 g, 1.5 mmol); 0.26 ml HCl; 2 ml H<sub>2</sub>O; 0.84 ml Dioxane; H<sub>4</sub>PMIDA(0.34 g, 1.5 mmol); and Urea (0.20 g, 3.3 mmol). The final pH was 3.5 and 2 for compounds <bold>1</bold> and <bold>2</bold>, respectively. Initial characterization was carried out by powder X-ray diffraction (PXRD), inductively coupled plasma-atomic emission spectroscopy (ICP-AES), thermogravimetric analysis (TGA), elemental CHN analysis, and IR spectroscopy. The PXRD patterns of <bold>1</bold> and <bold>2</bold> compared with their simulated patterns from single crystal analyses are presented in Figure <xref ref-type="supplementary-material" rid="SM1">S1</xref>. ICP-AES gave a Ln: P ratio of 1: 1 in agreement with the formula. C<sub>5</sub>H<sub>14</sub>N<sub>2</sub>O<sub>9</sub>PLa (416.06) <bold>1</bold>, (based on single-crystal data): calcd. La 33.55, P 7.48, C 14.43, H 3.39, N 6.73; found La 34.54, P 7.32, C 14.21, H 3.35, N 6.49; C<sub>5</sub>H<sub>10</sub>N<sub>2</sub>O<sub>7</sub>PYb (414.16) <bold>2</bold>, (based on single-crystal data): calcd. Yb 41.77, P 7.48, C 14.51, H 2.43, N 6.76, found Yb 40.80, P 7.18, C 14.52, H: 2.37, N 6.63. TGA data (mass losses), <bold>1</bold>: 43–197°C 4.72% (DTG peak at 62°C); 231–393°C 6.89% (DTG peaks at 267, 294, and 326°C); 436–893°C 41.2% (DTG peaks at 638, 755, 794, and 845°C. <bold>2</bold>: 260–460°C 28.82% (DTG peak at 362°C); 860–958°C 35.9% (DTG peak at 958°C); Selected ATR-FTIR data (cm<sup>−1</sup>), <bold>1</bold>: υ (O-H and N-H involved in hydrogen bonding interactions) = 3440–3000 s, (very broad), υ (C-H in –CH<sub>2</sub>-) = 2981 w, υ<sub>asym</sub>(CO<sup>−</sup><sub>2</sub>) = 1558 vs. υ<sub>sym</sub>(CO<sup>−</sup><sub>2</sub>) = 1399 s, δ (O-H… O) = 1338 m, υ (C-O) = 1247 m, υ (C-N) 1117, υ<sub>asym</sub>(P-O) = 1018 s, υ<sub>sym</sub>(P-O) = 981 s, υ (P-C) = 787 m. <bold>2:</bold> υ (C-H in –CH<sub>2</sub>-) = 2948 w, υ<sub>asym</sub>(CO<sup>−</sup><sub>2</sub>) = 1593 vs. υ<sub>sym</sub>(CO<sup>−</sup><sub>2</sub>) = 1405 s, 1384 s, δ (-CH<sub>2</sub>-) = 1448 m, υ (C-O) = 1341 m, 1333 sh, 1257 w, 1232 m, δ (C–C–N, amines) = 1154 m, υ (C-N) = 1122 w, υ<sub>asym</sub>(P-O) = 1084 vs. υ<sub>sym</sub>(P-O) = 1020 s, 1006 m, 974 w, υ (P-C) = 786 s, 713 s.</p></sec><sec><title>Instrumentation</title><p>A PXRD pattern was recorded at ambient temperature on a Bruker D8 Advance diffractometer (CuK<sub>α</sub> radiation λ = 1.54056 Å) fitted with Lynx EYE detector. Data were collected using a flat plate sample holder. Intensity data were collected by the continuous counting method (step 0.03° and time 3 s) in the range 5–50° 2θ. Excel and Origin 7 were used to analyse the data. Elemental analysis (C, H, and N) was performed by Atlantic Microlab, Inc. For La/P and Yb/P ratios, samples were digested in conc HNO<sub>3</sub> and Anderson Analytical determined the relative amounts by ICP-AES. Thermal analysis was carried out with a Rigaku Thermoflex 8110 unit at a heating rate of 5°C/min under nitrogen atmosphere from room temperature to 1000°C. Attenuated total reflection Fourier transform infrared (ATR-FTIR) spectra (4000–400 cm<sup>−1</sup>) were recorded with a Perkin-Elmer 883 spectrometer.</p></sec><sec><title>Crystal structure determination</title><p>Single crystal data were collected on a Bruker-AXS Apex II CCD X-ray diffractometer (MoKα radiation, λ = 0.71073 Å) operating at 110 K. The data were reduced using SAINTPLUS, (Bruker, <xref rid="B8" ref-type="bibr">2005</xref>) and an empirical absorption correction was applied using the SADABS program (Sheldrick, <xref rid="B47" ref-type="bibr">2008b</xref>). The structures were solved by direct methods and refined by the full-matrix least-squares technique against F<sup>2</sup> with the anisotropic displacement parameters for all non-hydrogen atoms using SHELXL-2008 (Sheldrick, <xref rid="B46" ref-type="bibr">2008a</xref>). All hydrogen atoms except those for the water molecules and ammonium, were added in idealized positions and refined using a riding model with <italic>U<sub>iso</sub></italic> = <italic>nU<sub>eq</sub></italic> for carbon atoms connected to the relevant H-atom where <italic>n</italic> = 1.5 for methyl and <italic>n</italic> = 1.2 for other H-atoms. The hydrogen atoms for the water molecules and ammonium ions in both compounds were located from difference Fourier maps and were refined using a riding mode. Anisotropic displacement parameters were established for all non-hydrogen atoms. Selected data collection and refinement parameters are summarized in Table <xref ref-type="table" rid="T1">1</xref>. More details on crystallographic studies as well as atom displacement parameters are given in the Supporting Information (CCDC 847459) and (CCDC 847460) for compounds <bold>1</bold> and <bold>2</bold>, respectively).</p><table-wrap id="T1" position="float"><label>Table 1</label><caption><p><bold>Crystal data and structure refinement for 1 and 2</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th rowspan="1" colspan="1"/><th align="left" rowspan="1" colspan="1"><bold>1</bold></th><th align="left" rowspan="1" colspan="1"><bold>2</bold></th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">Formula</td><td align="left" rowspan="1" colspan="1">C<sub>5</sub> H<sub>14</sub> N<sub>2</sub> O<sub>9</sub> P La</td><td align="left" rowspan="1" colspan="1">C<sub>5</sub> H<sub>10</sub> N<sub>2</sub> O<sub>7</sub> P Yb</td></tr><tr><td align="left" rowspan="1" colspan="1">Formula mass</td><td align="left" rowspan="1" colspan="1">416.06</td><td align="left" rowspan="1" colspan="1">414.16</td></tr><tr><td align="left" rowspan="1" colspan="1">Crystal system</td><td align="left" rowspan="1" colspan="1">Monoclinic</td><td align="left" rowspan="1" colspan="1">Monoclinic</td></tr><tr><td align="left" rowspan="1" colspan="1">Space group</td><td align="left" rowspan="1" colspan="1">P 2<sub>1</sub>/c</td><td align="left" rowspan="1" colspan="1">P 2<sub>1</sub>/c</td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>a</italic> (Å)</td><td align="left" rowspan="1" colspan="1">7.059(4)</td><td align="left" rowspan="1" colspan="1">9.181(3)</td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>b</italic> (Å)</td><td align="left" rowspan="1" colspan="1">23.577(12)</td><td align="left" rowspan="1" colspan="1">8.889(3)</td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>c</italic> (Å)</td><td align="left" rowspan="1" colspan="1">6.871(3) Å</td><td align="left" rowspan="1" colspan="1">12.827(4)</td></tr><tr><td align="left" rowspan="1" colspan="1">β (°)</td><td align="left" rowspan="1" colspan="1">94.292(6)°</td><td align="left" rowspan="1" colspan="1">101.414(3)</td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>V</italic> (Å<sup>3</sup>)</td><td align="left" rowspan="1" colspan="1">1140.4(10)</td><td align="left" rowspan="1" colspan="1">1026.0(5)</td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>Z</italic></td><td align="left" rowspan="1" colspan="1">4</td><td align="left" rowspan="1" colspan="1">4</td></tr><tr><td align="left" rowspan="1" colspan="1"><italic>ρ<sub>c</sub></italic> (Mg m<sup>−3</sup>)</td><td align="left" rowspan="1" colspan="1">2.423</td><td align="left" rowspan="1" colspan="1">2.681</td></tr><tr><td align="left" rowspan="1" colspan="1">μ (Mo-K<sub>α</sub>)(mm<sup>−1</sup>)</td><td align="left" rowspan="1" colspan="1">3.931</td><td align="left" rowspan="1" colspan="1">9.294</td></tr><tr><td align="left" rowspan="1" colspan="1">F(000)</td><td align="left" rowspan="1" colspan="1">808</td><td align="left" rowspan="1" colspan="1">780</td></tr><tr><td align="left" rowspan="1" colspan="1">Reflections collected</td><td align="left" rowspan="1" colspan="1">7029</td><td align="left" rowspan="1" colspan="1">8719</td></tr><tr><td align="left" rowspan="1" colspan="1">Independent reflections</td><td align="left" rowspan="1" colspan="1">2460 [R(int) = 0.1268]</td><td align="left" rowspan="1" colspan="1">2454 [R(int) = 0.0718]</td></tr><tr><td align="left" rowspan="1" colspan="1">GOF on F<sup>2</sup></td><td align="left" rowspan="1" colspan="1">1.015</td><td align="left" rowspan="1" colspan="1">1.028</td></tr><tr><td align="left" rowspan="1" colspan="1">Final R indices [I > 2σ (I)]<xref ref-type="table-fn" rid="TN1"><sup>a</sup></xref></td><td align="left" rowspan="1" colspan="1">R<sub>1</sub> = 0.0688, wR<sub>2</sub> = 0.1361 R1 = 0.1483, wR2 = 0.1680</td><td align="left" rowspan="1" colspan="1">R<sub>1</sub> = 0.0321, wR<sub>2</sub> = 0.0880 R<sub>1</sub> = 0.0363, wR<sub>2</sub> = 0.0928</td></tr><tr><td align="left" rowspan="1" colspan="1">Final R indices (All data)</td><td align="left" rowspan="1" colspan="1">1.849,</td><td align="left" rowspan="1" colspan="1">1.698,</td></tr><tr><td align="left" rowspan="1" colspan="1">(Δρ)<sub>max</sub>, (Δρ)<sub>min</sub>(e.Å<sup>−3</sup>)</td><td align="left" rowspan="1" colspan="1">−2.057</td><td align="left" rowspan="1" colspan="1">−2.685</td></tr></tbody></table><table-wrap-foot><fn id="TN1"><label>a</label><p>R<sub><italic>1</italic></sub> = Σ \|\|F<sub>o</sub>\| − \|F<sub>c</sub>\|\|/Σ \|F<sub>o</sub>\|. wR<sub><italic>2</italic></sub> = [Σ w(F<sup><italic>2</italic></sup><sub>o</sub> - F<sup><italic>2</italic></sup><sub>c</sub>)<sup><italic>2</italic></sup>/Σ w(F<sup><italic>2</italic></sup><sub>o</sub>)2]<sup><italic>1/2</italic></sup>.</p></fn></table-wrap-foot></table-wrap></sec></sec><sec><title>Results and discussion</title><p>The hydrothermal treatment of H<sub>4</sub>PMIDA with the lanthanides (Ln = La, <bold>1</bold>; Yb, <bold>2</bold>) in the presence of HCl-urea afforded two new lanthanide(III)carboxylate-phosphonates, namely (NH<sub>4</sub>)[La(PMIDA)(H<sub>2</sub>O)]•H<sub>2</sub>O, <bold>1</bold>, and (NH<sub>4</sub>)[Yb(PMIDA)], <bold>2</bold> incorporating ammonium ions to balance the anionic framework [Ln(PMIDA)]<sup>−</sup> charge of −1. The ammonium cations are generated <italic>in-situ</italic> from the partial decomposition of urea and are crucial in the reaction as structure directors. The pH of the reaction mixtures significantly influences the formation of the product. The initial pH for both compounds was 1.0. Whereas a final pH in the range of 3.5–4.0 favors the formation of compound <bold>1</bold>, a pH of 2.0 was found to favor compound <bold>2</bold>. The addition of urea and potassium acetate in the reaction vessel were needed to control the pH of the reaction media and to yield crystalline samples as direct addition of ammonia solution to the synthetic mixture could not lead to the formation of the products. The two compounds exhibit two-dimensional layered structures with distinct features.</p><p>The asymmetric unit of <bold>1</bold> consists of one crystallographically independent Lanthanum(III) ion, a PMIDA<sup>4−</sup> anion, aqua ligand, a lattice water and an ammonium ion for charge balancing. The PMIDA<sup>4−</sup> anion coordinates to the central La ion in a tetradentate fashion via one oxygen atom from each of the carboxylates (O4 and O6), one oxygen atom from the phosphonate (O1), and a nitrogen atom from the amino group (N1). This forms three five-membered chelation rings. The one independent La(III) ion is nine-coordinate and to fulfill the coordination, three other PMIDA anions coordinates to the metal center through the phosphonate oxygen atoms (O3A, O2B, O3C, and 02C) as shown in Figure <xref ref-type="fig" rid="F1">1</xref>. The geometry about the central atom is trigonal prismatic tricapped by N(1) and O(8). The interatomic distances are well defined. La-O distances are within the 2.458(9)–2.732(9) Å range [La-O<sub>av</sub> = 2.568 Å] and La-N distance is 2.821(12) Å, while the C-O and P-O distances are within the 1.248(16)–1.271(16), and 1.508(9)–1.543(10) Å ranges, respectively and are indicative of complete deprotonation of both the carboxylate and phosphonate groups. The longest P-O distance belongs to the μ<sup>2</sup>- bridging oxygen atoms. The P-C distance is 1.791(14) Å. These distances are in good agreement with similar compounds in the literature (Legendziewicz et al., <xref rid="B27" ref-type="bibr">1998</xref>; Serre et al., <xref rid="B43" ref-type="bibr">2004</xref>; Song and Mao, <xref rid="B53" ref-type="bibr">2005</xref>; Tang et al., <xref rid="B55" ref-type="bibr">2006a</xref>; Ying and Mao, <xref rid="B57" ref-type="bibr">2006</xref>; Huang et al., <xref rid="B25" ref-type="bibr">2007</xref>; Liu et al., <xref rid="B28" ref-type="bibr">2007</xref>; Mao, <xref rid="B32" ref-type="bibr">2007</xref>).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>ORTEP plot of 1 showing the labeling scheme for non-hydrogen atoms</bold>. The thermal ellipsoids are drawn at 50% probability level. Symmetry codes for the generated atoms: (a) x, y, z−1; (b) 1−x, 2−y, 1−z; (c) x, y, 1+z.</p></caption><graphic xlink:href="fchem-02-00094-g0001"/></fig><p>The PMIDA<sup>4−</sup> anion is a polydentate ligand, which chelates with a lanthanum(III) ion in a tetradentate fashion and bridges three other lanthanum ions using the phosphonate oxygen atoms. The phosphorus atom P(1) on the tetradentate ligand coordinating to La(1) via O1, forms multiple bonds to three equivalent lanthanum atoms [La(1A), La(1B), and La(1C)] through the remaining two oxygen atoms O2 and O3 giving rise to a [122] connectivity mode (Massiot et al., <xref rid="B34" ref-type="bibr">1997</xref>; Bauer et al., <xref rid="B6" ref-type="bibr">2006</xref>; Tang et al., <xref rid="B55" ref-type="bibr">2006a</xref>,<xref rid="B56" ref-type="bibr">b</xref>; Zhou et al., <xref rid="B62" ref-type="bibr">2010</xref>). While the phosphonate oxygen O2 and O3 both chelate with the La(1A) center, O3 also bridges with La(1B) to form a chain propagating along the [001] direction as shown in Figure <xref ref-type="fig" rid="F2">2</xref>. Within the chain, the La… La distances over μ<sub>3</sub>-O3 and O1–P1–O3 bridges are 4.227(2) and 6.871(3) Å, respectively. The chains are then connected together by the bridging phosphonate oxygen O2 to form a 2D layered structure with alternating 4- and 8-membered rings as shown in Figure <xref ref-type="fig" rid="F3">3</xref>. Each of the two carboxylate oxygen coordinates to the La(III) center in a monodentate fashion. The two non-coordinating carboxylate oxygen atoms [O5, O7] point into the interlamellar space and interact with the lattice water and ammonium ions in the interlayers via hydrogen bonds (Figure <xref ref-type="fig" rid="F2">2</xref>, Table <xref ref-type="table" rid="T2">2</xref>). The adjacent layers separated by 12.467(6) Å, are connected together via these multiple hydrogen bonds.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>View of compound 1 along [001] direction showing the chain structure</bold>.</p></caption><graphic xlink:href="fchem-02-00094-g0002"/></fig><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Structure of 1 viewed along <italic>b</italic>-axis showing the connectivity that gives rise to 2D layer</bold>. The lattic water and ammonium ions are ommitted for clarity.</p></caption><graphic xlink:href="fchem-02-00094-g0003"/></fig><table-wrap id="T2" position="float"><label>Table 2</label><caption><p><bold>Hydrogen bonds for compounds 1 and 2</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="left" rowspan="1" colspan="1"><bold>D-H… A</bold></th><th align="center" rowspan="1" colspan="1"><bold>(D-H-A)/<sup><bold>O</bold></sup></bold></th><th align="center" rowspan="1" colspan="1"><bold>d(D… A)/Å</bold></th><th align="left" rowspan="1" colspan="1"><bold>Symmetry operation</bold></th></tr></thead><tbody><tr><td align="left" colspan="4" rowspan="1"><bold>COMPOUND 1</bold></td></tr><tr><td align="left" rowspan="1" colspan="1">N2-H2-O7</td><td align="center" rowspan="1" colspan="1">152.54</td><td align="center" rowspan="1" colspan="1">2.820(16)</td><td align="left" rowspan="1" colspan="1">1 + x, y, z</td></tr><tr><td align="left" rowspan="1" colspan="1">N2-H3-O5</td><td align="center" rowspan="1" colspan="1">163.65</td><td align="center" rowspan="1" colspan="1">2.825(15)</td><td align="left" rowspan="1" colspan="1">x, y, z</td></tr><tr><td align="left" rowspan="1" colspan="1">N2-H4-O10</td><td align="center" rowspan="1" colspan="1">168.50</td><td align="center" rowspan="1" colspan="1">2.962(14)</td><td align="left" rowspan="1" colspan="1">x, y, z</td></tr><tr><td align="left" rowspan="1" colspan="1">O10-H10A-O4</td><td align="center" rowspan="1" colspan="1">114.58</td><td align="center" rowspan="1" colspan="1">2.919(16)</td><td align="left" rowspan="1" colspan="1">x, y, 1 + z</td></tr><tr><td align="left" rowspan="1" colspan="1">O8-H8A-O6</td><td align="center" rowspan="1" colspan="1">122.54</td><td align="center" rowspan="1" colspan="1">2.656(14)</td><td align="left" rowspan="1" colspan="1">1 − x, 2 − y, 2 − z</td></tr><tr><td align="left" colspan="4" rowspan="1"><bold>COMPOUND 2</bold></td></tr><tr><td align="left" rowspan="1" colspan="1">N2-H2C-O7</td><td align="center" rowspan="1" colspan="1">167.14</td><td align="center" rowspan="1" colspan="1">2.794(5)</td><td align="left" rowspan="1" colspan="1">−1 + x, y, z</td></tr><tr><td align="left" rowspan="1" colspan="1">N2-H2D-O6</td><td align="center" rowspan="1" colspan="1">150.82</td><td align="center" rowspan="1" colspan="1">3.032(4)</td><td align="left" rowspan="1" colspan="1">1 − x, ½ + y, ½ − z</td></tr><tr><td align="left" rowspan="1" colspan="1">N2-H2E-O5</td><td align="center" rowspan="1" colspan="1">120.46</td><td align="center" rowspan="1" colspan="1">3.069(5)</td><td align="left" rowspan="1" colspan="1">1 − x, −½ + y, ½ − z</td></tr><tr><td align="left" rowspan="1" colspan="1">N2-H2F-O3</td><td align="center" rowspan="1" colspan="1">170.86</td><td align="center" rowspan="1" colspan="1">2.855(6)</td><td align="left" rowspan="1" colspan="1">1 − x, 1 − y, 1 − z</td></tr></tbody></table></table-wrap><p>The asymmetric unit of <bold>2</bold> consists of one crystallographically independent ytterbium(III) ion, a PMIDA<sup>4−</sup> anion, and an ammonium ion for charge balance (Figure <xref ref-type="fig" rid="F4">4</xref>). The Yb(III) ion is coordinated by a polydentate PMIDA anion in a tetradentate fashion via one oxygen atom from each of the carboxylates (O5 and O6), one oxygen atom from the phosphonate (O1), and a nitrogen atom from the amino group (N1). This forms three five-membered chelation rings similar to the lanthanum compound, <bold>1</bold>. Three other PMIDA anions coordinating through two phosphonate oxygen atoms (02A and 03A) and one carboxylate oxygen atom (04B) completes the coordination number of seven around the central Yb(III) ion. Interestingly, the two carboxylate groups in PMIDA<sup>4−</sup> adopt different coordination modes. One carboxylate group is only coordinated to Yb atom by O6 in a monodentate fashion, while the other is bidentate bridging through O4 and O5. The geometry about the central metal ion is a capped trigonal prism YbO<sub>6</sub>N. The Yb-O distances are in the range 2.174(3)–2.318(3) Å, while the Yb-N distance is 2.576(5) Å and are comparable with those reported for similar compounds in the literature (Zabicky, <xref rid="B58" ref-type="bibr">1970</xref>; Burwell and Thompson, <xref rid="B9" ref-type="bibr">1991a</xref>,<xref rid="B10" ref-type="bibr">b</xref>; Tang et al., <xref rid="B55" ref-type="bibr">2006a</xref>). The Yb(PMIDA) chelating units are connected by alternating bridging carboxylate and phosphonate groups along the [010] direction, forming chains with a corrugated pattern. The phosphonate oxygen O2, bridges the chains together along the [001] direction to build the layered material as shown in Figure <xref ref-type="fig" rid="F5">5</xref>.</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>ORTEP plot of 2 showing the labeling scheme</bold>. The thermal ellipsoids are drawn at the 50% probability level. Symmetry codes for the generated atoms: (a) x, 0.5−y, −0.5+z; (b) 1−x, −y, 1−z; (c) 1−x, 1−y, 1−z; (d) x, 0.5−y, 0.5−z.</p></caption><graphic xlink:href="fchem-02-00094-g0004"/></fig><fig id="F5" position="float"><label>Figure 5</label><caption><p><bold>Structure of 2 viewed along [100] direction showing the connectivity that gives the 2D layer</bold>. Note the phosphonate group [111] connectivity mode. Ammonium ions are ommitted for clarity.</p></caption><graphic xlink:href="fchem-02-00094-g0005"/></fig><p>The PMIDA<sup>4−</sup> anion is a heptadentate ligand, coordinating through three phosphonate oxygen atoms (O1, O2, O3), three carboxylate oxygen atoms (O4, O5, and O6) and the amino nitrogen (N1). The phosphonate group is tridentate and bridges with three equivalent Yb(PMIDA) chelating units, that is the three oxygen atoms of the PO<sub>3</sub> group are bonded to different ytterbium atoms, such that each is directly connected to only one ytterbium atom in a [111] connectivity (Massiot et al., <xref rid="B34" ref-type="bibr">1997</xref>; Bauer et al., <xref rid="B6" ref-type="bibr">2006</xref>; Tang et al., <xref rid="B55" ref-type="bibr">2006a</xref>,<xref rid="B56" ref-type="bibr">b</xref>; Zhou et al., <xref rid="B62" ref-type="bibr">2010</xref>). Both the Yb(III) and phosphonate group are 3-connectors in terms of topology giving the 2D layer a vertex symbol of 6 (Ayi et al., <xref rid="B3" ref-type="bibr">2013</xref>), with two types of four-coordinate nodes. This kind of topology has also been observed in similar lanthanide(III) complexes, but with two types of three-coordinate nodes (Massiot et al., <xref rid="B34" ref-type="bibr">1997</xref>; Bauer et al., <xref rid="B6" ref-type="bibr">2006</xref>; Tang et al., <xref rid="B55" ref-type="bibr">2006a</xref>,<xref rid="B56" ref-type="bibr">b</xref>; Zhou et al., <xref rid="B62" ref-type="bibr">2010</xref>). The non-coordinating carboxylate oxygen O(7) points into the interlamellar space and interacts with the ammonium ions through hydrogen bonding (Table <xref ref-type="table" rid="T2">2</xref>). The interlayer distance is 9.181(3) Å. Hydrogen bonding between the ammonium ions and the layers hold the layers together.</p><p>It is interesting to note that whereas the two reactions took place in aqueous solution, compound <bold>1</bold> has coordinated and lattice water, while compound <bold>2</bold> is an anhydrous complex. There is a change in coordination number from 9 in compound <bold>1</bold>–7 in compound <bold>2</bold> attributed to the decrease in the size in going from La<sup>3+</sup> (103pm) to Yb<sup>3+</sup> (87pm) ions (Shannon, <xref rid="B44" ref-type="bibr">1976</xref>; Ma et al., <xref rid="B31" ref-type="bibr">2007</xref>) Similar hydrothermal reactions of H<sub>4</sub>PMIDA with lanthanide(III) salts reported by Mao and coworkers gave lanthanide(III) carboxylate-phosphonates, which was formulated as Ln(HPMIDA)(H<sub>2</sub>O)<sub>2</sub>·H<sub>2</sub>O (Ln) Gd, <bold>1</bold>; Tb, <bold>2</bold>; Dy, <bold>3</bold>; Y, <bold>4</bold>; Er, <bold>5</bold>; Yb, <bold>6</bold>; Lu, <bold>7</bold>) (Tang et al., <xref rid="B55" ref-type="bibr">2006a</xref>). Their structures feature a three-dimensional network with helical tunnels. In this present investigation, compound <bold>1</bold> is formulated as (NH<sub>4</sub>)[La(PMIDA)(H<sub>2</sub>O)]•H<sub>2</sub>O and compound <bold>2</bold> has the formula (NH<sub>4</sub>)[Yb(PMIDA], both featuring a two-dimensional structure with extensive hydrogen bonding involving the ammonium ion. In the previously reported compounds, the phosphonate group is singly protonated, whereas the compounds under present study shows complete deprotonation of the acidic oxygen atoms. The complete deprotonation is made possible by the addition of urea into the synthetic media, which resulted in the incorporation of ammonium ion in the present structure.</p><p>The infrared spectra of the compounds are particularly informative, providing supporting evidence for the structural differences existing between compounds <bold>1</bold> and <bold>2</bold>. The free ligand shows absorption peaks in the spectral range between 1731 and 1216 cm<sup>−1</sup> arising from stretching and bending vibrational modes associated with C = O, C-O, and C-H bonds. While the band at 1731 cm<sup>−1</sup> is assigned to υ<sub>asym</sub>(C = O), the one at 1473 cm<sup>−1</sup> is due to υ<sub>sym</sub>(C = O). The bands at 1336, 1265, 1244, and 1216 cm<sup>−1</sup> are attributed to υ<sub>s</sub>(C-O) whereas those at 1442 and 1423 cm<sup>−1</sup> are due to δ (C-H in –CH<sub>2</sub>-). In the infrared spectrum of <bold>1</bold>, the broad band in the spectral region 3500–3360 cm<sup>−1</sup> (peaking at ca. 3436 cm<sup>−1</sup>) is attributed to the υ (O-H) of water molecules involved in hydrogen bonds. This feature is absent in compound <bold>2</bold>. The broad absorption bands observed at ca. 3207 cm<sup>−1</sup> (<bold>1</bold>) and at ca. 3180 cm<sup>−1</sup> (<bold>2</bold>) are attributed to the υ<sub>s</sub>(N-H) vibrations involved in hydrogen bonding. The stretching mode of–CH<sub>2</sub> groups is markedly visible in the spectra, giving rise to peak around 2981 cm<sup>−1</sup> in <bold>1</bold> and 2948 cm<sup>−1</sup> in <bold>2</bold>. The strong peaks observed at 1558 and 1399 cm<sup>−1</sup> in <bold>1</bold> (1593, and 1384 cm<sup>−1</sup> in <bold>2</bold>) are due to the asymmetric and symmetric stretching mode of the CO<sup>−</sup><sub>2</sub> bonds of the carboxyl group. This is lower than the 1731 cm<sup>−1</sup> absorption peak seen in that of the free ligand. This down field shift in the absorption frequency is a clear indication that the carboxylate groups are involved in coordinating to the metal center (Zabicky, <xref rid="B58" ref-type="bibr">1970</xref>; Nakamato, <xref rid="B35" ref-type="bibr">1978</xref>; Deacon and Phillips, <xref rid="B15" ref-type="bibr">1980</xref>; Burwell and Thompson, <xref rid="B9" ref-type="bibr">1991a</xref>,<xref rid="B10" ref-type="bibr">b</xref>; Drumel et al., <xref rid="B16" ref-type="bibr">1995</xref>; Ayi et al., <xref rid="B4" ref-type="bibr">2011</xref>) in both <bold>1</bold> and <bold>2</bold>. In the infrared spectrum of <bold>2</bold>, the characteristic antisymmetric and symmetric stretching bands for the carboxylate ions are present with the corresponding Δ [ν<sub>assym</sub>(CO<sup>−</sup><sub>2</sub>) − ν<sub>sym</sub>(CO<sup>−</sup><sub>2</sub>)] values being 188 and 209 cm<sup>−1</sup> indicating the presence of carboxylate groups in the anti-unidentate and bridging-η<sup>2</sup>-<italic>anti</italic>,<italic>anti</italic>-chelate coordination modes, respectively (Nakamato, <xref rid="B35" ref-type="bibr">1978</xref>; Deacon and Phillips, <xref rid="B15" ref-type="bibr">1980</xref>; Drumel et al., <xref rid="B16" ref-type="bibr">1995</xref>). The vibrational modes of the phosphonate (PO<sub>3</sub>) units are also noticeable in the infrared spectra of compounds <bold>1</bold> and <bold>2</bold>. In <bold>1</bold>, the assymmetric stretching vibrational band of P-O group is observed at 1018 cm<sup>−1</sup> and at 1084 cm<sup>−1</sup> for <bold>2</bold>, while the symmetric stretching mode is at 981 cm<sup>−1</sup> in <bold>1</bold> and at 1020 cm<sup>−1</sup> in <bold>2</bold>. The P-C stretching modes are obsereved around 787 cm<sup>−1</sup> in both compounds.</p><p>In order to investigate the thermal stability of these materials, the TGA curves of compounds <bold>1</bold> and <bold>2</bold> were measured (Figure <xref ref-type="fig" rid="F6">6</xref>). Compound <bold>1</bold>, releases the interlayer water molecule of crystallization in the temperature range 62–197°C. The observed weight loss of 4.62% is close to the calculated value (4.35%). In the region 202–376°C, there is a weight loss of 8.92% (calc. 8.84%) attributed to the loss of coordinated water molecule and NH<sub>3</sub>, as well as a loss of about 11.50% due to the presence of little impurities in the bulk sample. However, the final weight loss of 42.29% (calc. 43.78%) corresponds to the decomposition of the organic part of the material to give LaPO<sub>4</sub> (JCPDF card no. 01-084-0600) as the final product. Compound <bold>2</bold> is thermally stable up to 300°C. The first weight loss of 28.82% (calc. 28.25%) is attributed to the loss of NH<sub>3</sub>, 2CO, and CO<sub>2</sub>. A plateau appears in the range 422–849°C, above which a final weight loss of 35.9% (calc. 35.28%) occurs, corresponding to the decomposition of the organic part to give YbPO<sub>4</sub> (JCPDF card no. 01-076-1643).</p><fig id="F6" position="float"><label>Figure 6</label><caption><p><bold>TGA curves of compounds 1(solid) and 2 (dashed)</bold>.</p></caption><graphic xlink:href="fchem-02-00094-g0006"/></fig><p>The photoluminecence properties of compound <bold>2</bold> was investigated in the solid state at room temperature (Figure <xref ref-type="fig" rid="F7">7</xref>). Under excitation of 330 nm, the compound shows a broad emission band at λ<sub>max</sub> = 460 nm, This emission is essentially in the blue luminescent region, which corresponds to ligand centered fluorescence (Tang et al., <xref rid="B55" ref-type="bibr">2006a</xref>; Zhou et al., <xref rid="B62" ref-type="bibr">2010</xref>). Owing to the quenching effect of the luminiscent state reported for complexes with coordinated water molecules, (Song et al., <xref rid="B52" ref-type="bibr">2004</xref>; Sarkar et al., <xref rid="B41" ref-type="bibr">2006</xref>; Deng et al., <xref rid="B63" ref-type="bibr">2011</xref>) the solid state luminescence of <bold>1</bold> was not studied.</p><fig id="F7" position="float"><label>Figure 7</label><caption><p><bold>Solid state photoluminescent studies of compound 2 at room temperature</bold>.</p></caption><graphic xlink:href="fchem-02-00094-g0007"/></fig></sec><sec sec-type="conclusion" id="s2"><title>Conclusion</title><p>We have successfully synthesized hydrothermally, two new compounds based on <italic>N</italic>-(phosphonomethyl)iminodiacetic acid (H<sub>4</sub>PMIDA), namely (NH<sub>4</sub>La(PMIDA)(H<sub>2</sub>O)•H<sub>2</sub>O, <bold>1</bold>, and (NH<sub>4</sub>)Yb(PMIDA), <bold>2</bold>. The presence of the ammonium ions serves to compensate the framework negative charge in both compounds and is crucial in the syntheses of the compounds under investigation as it stabilizes the structures through hydrogen bonding interactions. The change in the coordination number from 9 for compound <bold>1</bold>–7 for compound <bold>2</bold> clearly shows that the size of the cation plays an important role in determining the coordination number. Thus, ion with larger radius favors a higher coordination number and a larger cavity to accommodate more water molecules (Shannon, <xref rid="B44" ref-type="bibr">1976</xref>; Ma et al., <xref rid="B31" ref-type="bibr">2007</xref>). Solid state photoluminescent studies of compound <bold>2</bold> at room temperature shows broad emission in the blue luminiscent region, which is essentially attributed to ligand centered fluorescence. Efforts are under way to synthesize the complete series of the ammonium ion-templated lanthanide(III) complexes with this particular ligand with a view to elucidate their crystal structures, magnetic and luminescent properties.</p><sec><title>Conflict of interest statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec></sec>
Local anaphor licensing in an SOV language: implications for retrieval strategies	<p>Because morphological and syntactic constraints govern the distribution of potential antecedents for local anaphors, local antecedent retrieval might be expected to make equal use of both syntactic and morphological cues. However, previous research (e.g., Dillon et al., <xref rid="B14" ref-type="bibr">2013</xref>) has shown that local antecedent retrieval is not susceptible to the same morphological interference effects observed during the resolution of morphologically-driven grammatical dependencies, such as subject-verb agreement checking (e.g., Pearlmutter et al., <xref rid="B39" ref-type="bibr">1999</xref>). Although this lack of interference has been taken as evidence that syntactic cues are given priority over morphological cues in local antecedent retrieval, the absence of interference could also be the result of a confound in the materials used: the post-verbal position of local anaphors in prior studies may obscure morphological interference that would otherwise be visible if the critical anaphor were in a different position. We investigated the licensing of local anaphors (reciprocals) in Hindi, an SOV language, in order to determine whether pre-verbal anaphors are subject to morphological interference from feature-matching distractors in a way that post-verbal anaphors are not. Computational simulations using a version of the ACT-R parser (Lewis and Vasishth, <xref rid="B23" ref-type="bibr">2005</xref>) predicted that a feature-matching distractor should facilitate the processing of an unlicensed reciprocal if morphological cues are used in antecedent retrieval. In a self-paced reading study we found no evidence that distractors eased processing of an unlicensed reciprocal. However, the presence of a distractor increased difficulty of processing following the reciprocal. We discuss the significance of these results for theories of cue selection in retrieval.</p>	<contrib contrib-type="author"><name><surname>Kush</surname><given-names>Dave</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="author-notes" rid="fn001"><sup>*</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/107899"/></contrib><contrib contrib-type="author"><name><surname>Phillips</surname><given-names>Colin</given-names></name><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><xref ref-type="aff" rid="aff3"><sup>3</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/80498"/></contrib>	Frontiers in Psychology	<sec sec-type="introduction" id="s1"><title>Introduction</title><p>In order to establish grammatical dependencies between words across a distance during routine sentence processing comprehenders rely heavily on their ability to encode and retrieve items from memory. For example, processing of a local anaphor such as the reflexive <italic>themselves</italic> or the reciprocal <italic>each other</italic> in (1) requires recalling the previously seen noun phrase (NP) <italic>the people</italic> from memory so that it may be interpreted as the antecedent.</p><list list-type="simple"><list-item><p>(1) The people talked to <italic>themselves/each other</italic>.</p></list-item></list><p>The mechanism by which previously encountered items are retrieved for subsequent processing has been the subject of recent research. A number of recent studies have motivated a processing model that exploits a cue-based access mechanism to retrieve items from content-addressable memory (e.g., McElree, <xref rid="B28" ref-type="bibr">2000</xref>; McElree et al., <xref rid="B31" ref-type="bibr">2003</xref>; Lewis et al., <xref rid="B24" ref-type="bibr">2006</xref>; Van Dyke, <xref rid="B42" ref-type="bibr">2007</xref>; Martin and McElree, <xref rid="B25" ref-type="bibr">2008</xref>, <xref rid="B26" ref-type="bibr">2009</xref>; Van Dyke and McElree, <xref rid="B44" ref-type="bibr">2011</xref>).</p><p>A hallmark property of cue-based retrieval is that it is susceptible to interference (Nairne, <xref rid="B34" ref-type="bibr">1990</xref>). Task-irrelevant items in memory whose features overlap with a probe's retrieval cues (distractors) can exert influence on the retrieval of a target item. In the context of sentence processing retrieval interference is said to occur when grammatically inappropriate distractors influence the processing of a phrase that must enter into a dependency with a previously encountered head. The influence of distractors can be <italic>inhibitory</italic>: a distractor may increase the difficulty of retrieving an appropriate item. Van Dyke (<xref rid="B42" ref-type="bibr">2007</xref>) found that distractor NPs increased the difficulty of retrieving a grammatically appropriate subject for the purposes of thematic integration with a verb (see also Van Dyke and McElree, <xref rid="B43" ref-type="bibr">2006</xref>, <xref rid="B44" ref-type="bibr">2011</xref>). A distractor's influence may also be <italic>facilitatory</italic> if its presence decreases the difficulty of processing an otherwise ungrammatical or unlicensed element. Comprehenders have repeatedly showed signs of facilitatory interference during the processing of subject-verb agreement (e.g., Pearlmutter et al., <xref rid="B39" ref-type="bibr">1999</xref>; Wagers et al., <xref rid="B46" ref-type="bibr">2009</xref>). Wagers and colleagues found that reading times immediately following the plural verb <italic>were</italic>, which mismatched the features of the singular subject <italic>key</italic>, were decreased when an intervening distractor [<italic>cabinet(s)</italic>] was plural, compared to when the distractor was singular.</p><list list-type="simple"><list-item><p>(2) The key to the <bold>cabinet(s)</bold> unsurprisingly <italic>were</italic> rusty from years of disuse.</p></list-item></list><p>The authors argued that facilitation arose because comprehenders erroneously retrieved the plural distractor on some portion of trials when attempting to find a licensor for the plural marking on the verb. These kinds of facilitatory interference effects have also been observed in the processing of other grammatical dependencies such as negative polarity item (NPI) licensing (e.g., Drenhaus et al., <xref rid="B15" ref-type="bibr">2005</xref>; Vasishth et al., <xref rid="B45" ref-type="bibr">2008</xref>; Xiang et al., <xref rid="B47" ref-type="bibr">2009</xref>; Parker and Phillips, submitted), and the retrieval of antecedents for null pronominal subjects (PRO) in adjunct clauses (Parker et al., <xref rid="B37" ref-type="bibr">2012</xref>) and many authors have attributed these effects to misretrieval of a distractor under (partial) match with a set of retrieval cues.</p><p>Although facilitatory interference has been repeatedly observed in the processing of some dependencies, other dependencies that recruit retrieval have displayed virtual immunity to facilitation from distractors. Recent work has found that the processing of a local anaphor that lacks a grammatical antecedent is unaffected by the morphological feature-content of intervening distractors (e.g., Sturt, <xref rid="B41" ref-type="bibr">2003</xref>; Dillon et al., <xref rid="B14" ref-type="bibr">2013</xref>). For instance, Dillon et al. (<xref rid="B14" ref-type="bibr">2013</xref>) demonstrated that the processing of the unlicensed plural reflexive <italic>themselves</italic> in (3) is not influenced by plural-marking on the distractor <italic>manager(s)</italic>.</p><list list-type="simple"><list-item><p>(3) The new executive who oversaw the manager(s) apparently doubted <italic>themselves</italic>…</p></list-item></list><p>The lack of facilitatory interference effects is unexpected on the assumption that the same cues as those used to find licensors for agreement dependencies (e.g., morphological features such as number) are used to identify potential antecedents of reflexives. As with agreement, reflexives must match their licensors in number and gender, so the use of morphological features as cue for retrieval of appropriate antecedents would appear to be motivated. On analogy to agreement licensing, the use of these morphological cues should in turn render antecedent retrieval subject to interference.</p><p>The results suggest that morphological features may play a different role in antecedent retrieval for local anaphors than they do in agreement licensing. One option, advocated by Dillon et al. (<xref rid="B14" ref-type="bibr">2013</xref>), is that antecedent retrieval forgoes the use of interference-prone morphological features, opting instead to exclusively use <italic>positional</italic> syntactic features to access the local subject. Another option is that antecedent retrieval preferentially weights syntactic cues over morphological cues instead of avoiding them altogether. This second account predicts a small but non-negligible interference effect that the first does not, but previous experiments may not have had sufficient power to find this effect, so they cannot distinguish between the two competing explanations.</p><p>Although the two accounts differ, they both assign priority to positional cues. This goes against the general assumption that retrieval identifies targets through the use of a maximal cue set that uniformly weights lexical, morphological, syntactic, and semantic features (see Van Dyke and McElree, <xref rid="B44" ref-type="bibr">2011</xref> for discussion).</p><p>As it stands the previous studies may not be sufficient to establish a preference for positional features. It is possible that the absence of facilitatory interference could be attributed to a confound that masks the contribution of morphological features that are weighted equally to syntactic cues. In almost all previous studies the critical anaphor immediately followed its verb, which could potentially play a role in reducing the incidence of facilitatory interference (see King et al., <xref rid="B20" ref-type="bibr">2012</xref> for a similar suggestion).</p><p>As Dillon et al. (<xref rid="B14" ref-type="bibr">2013</xref>) note, the post-verbal position can provide an anaphor with privileged access to the local subject by means of recent activation alone. If subjects are retrieved by their verbs for thematic integration, the local subject <italic>the executive</italic> in (3) should be recalled by the verb <italic>doubted</italic>. Retrieval of the local subject entails that it should have the highest baseline activation out of all other items in memory immediately following the verb. At the time that a verb-adjacent reflexive is encountered, this high degree of activation may be strong enough to guarantee retrieval of the local subject instead of the feature-matching distractor even if morphological cues were used.</p><p>Alternatively, it may be that previous studies on reflexives do not provide a measure of susceptibility to facilitatory interference because establishing a dependency between the local subject and a post-verbal anaphor might not require retrieval at all. Some theories assume that the most recently retrieved item is maintained in a state that the parser can access without retrieval. In some theories this state is referred to as the <italic>focus of attention</italic> (e.g., McElree, <xref rid="B28" ref-type="bibr">2000</xref>), in others such as Lewis and Vasishth's (<xref rid="B23" ref-type="bibr">2005</xref>) parsing model it is the <italic>problem buffer</italic>. When an anaphor is encountered immediately following the verb, it is possible that it consults the contents of this buffer to find its antecedent rather than initiating a retrieval from memory.</p><p>In this study we address the extent to which the lack of facilitatory interference in anaphoric licensing depends on an anaphor's post-verbal position. If the absence of interference is a consequence of the target anaphor occupying an immediately post-verbal position, then in languages where anaphors uniformly precede their verbs, local anaphor licensing should display facilitatory effects that have not been seen in English. We tested this prediction by investigating the processing of Hindi reciprocals. Hindi is a language in which all arguments and adjuncts precede the verb in unmarked word order. In (4), for example, the subject <italic>LaRkoN</italic> (“boys”), the reciprocal object <italic>ek-dusre</italic> (“each other”), and the adjunct <italic>kal</italic> (“tomorrow”) precede the verb <italic>dekhaa</italic> (“saw”).</p><list list-type="simple"><list-item><p>(4) LaRkoN-ne ek-dusre-ko kal dekhaa.</p></list-item><list-item><p>Boys-Erg each.other-Acc yesterday saw.</p></list-item><list-item><p>‘(The) boys saw each other yesterday.’</p></list-item></list><p>Hindi reciprocals provide a minimal contrast to English reflexives because they are subject to nearly identical licensing conditions as English local anaphors. Their antecedent must have matching morphological features: in order to license the reciprocal in (5), the local subject must bear plural features. The reciprocal's antecedent must be contained in the same local clause as the reciprocal: the main clause subject in (6) cannot antecede the reciprocal in the embedded clause, despite bearing correct number marking, because it is not local to the reciprocal. Finally, the reciprocal's antecedent must also c-command the reciprocal (cf. Dayal, <xref rid="B13" ref-type="bibr">1994</xref>). In (7), the plural NP <italic>boys</italic> does not c-command the reciprocal because it is embedded inside the adjunct phrase <italic>at the boys' party</italic>. It is therefore ineligible to license the anaphor.</p><list list-type="simple"><list-item><p>(5) <italic>LaRk</italic>-{<sup></sup>-e/oN}-ne <italic>ek-dusre</italic>-ko kal dekhaa.</p></list-item><list-item><p>Boy-{Sing./Pl.}-Erg each.other-Acc yesterday saw.</p></list-item><list-item><p>‘(The) boy(s) saw each other yesterday.’</p></list-item><list-item><p>(6) <sup></sup><italic>LaRkoN</italic>-ne kahaa ki Mary-ne <italic>ek-dusre</italic>-ko dekhaa.</p></list-item><list-item><p>Boys-Erg said that Mary-Erg each.other-Acc saw.</p></list-item><list-item><p><sup></sup>‘(The) boys said that Mary saw each other.’</p></list-item><list-item><p>(7) <sup></sup>Mary-ne [<italic>larkoN</italic>-ki parTi me] <italic>ek-dusre</italic>-ko dekhaa.</p></list-item><list-item><p>Mary-Erg boys' party in one-another-Acc saw</p></list-item><list-item><p><sup></sup>‘Mary saw each other at the boys' party.’</p></list-item></list><p>We test whether morphological number features engender facilitatory interference effects during the processing of Hindi reciprocals.</p></sec><sec><title>Simulations</title><p>We ran a series of computational simulations that modeled local anaphor resolution in Hindi using equally-weighted morphological and positional features as cues for retrieval. Modeling was carried out to obtain qualitative predictions about the character and direction of interference from the distractor's morphological features that could then be compared with empirical reading times in the self-paced reading experiment.</p><sec><title>Procedure</title><p>We implemented a modified version of Lewis and Vasishth's (<xref rid="B23" ref-type="bibr">2005</xref>) ACT-R model of sentence processing [using code originally developed by Badecker and Lewis (<xref rid="B6" ref-type="bibr">2007</xref>)]. ACT-R is a general cognitive architecture that has been used to model a wide range of phenomena in cognitive psychology (Anderson, <xref rid="B2" ref-type="bibr">1990</xref>). In the model, items are stored as “chunks” in a content-addressable memory and are retrieved with a success proportional to their overall activation at the time of retrieval, which is in turn determined by the overlap of their features with those of a retrieval probe. Memory access is modeled as a rational procedure that employs a general retrieval mechanism that minimizes retrieval error in the limit (Anderson, <xref rid="B1" ref-type="bibr">1989</xref>; Anderson and Milson, <xref rid="B3" ref-type="bibr">1989</xref>; Anderson and Schooler, <xref rid="B4" ref-type="bibr">1991</xref>). Although fully implemented ACT-R parsing models exist (e.g., Lewis and Vasishth's, <xref rid="B23" ref-type="bibr">2005</xref> ACT-R parser), the simulations here focus solely on modeling retrieval latencies, abstracting away from the contributions of other modules. Retrieval latencies do not exhaust the processes that must be carried out in order to advance to the next word in a parsing task (other operations include structural attachment and integration), but for current purposes we adopt the standard assumption that longer retrieval latencies entail longer RTs (Anderson and Milson, <xref rid="B3" ref-type="bibr">1989</xref>).</p><p>In the model the probability of retrieving an item i is governed by its activation A<sub>i</sub>, computed as in (8). B<sub>i</sub> is chunk i's baseline activation. The weight assigned to the individual cue j is represented w<sub>j</sub>. For the purposes of our simulations cues were assigned uniform weights, so this term can be effectively dropped. S<sub>ji</sub> is the strength of association between cue j and chunk i. PM in the equation below is a term that penalizes partial matches. The term ε introduces stochastic noise.</p><list list-type="simple"><list-item><p>(8) A<sub>i</sub> = B<sub>i</sub> Σ w<sub>j</sub>S<sub>ji</sub> + PM + ε<sub>i</sub></p></list-item></list><p>S<sub>ji</sub> is calculated according to the Equation in (9), where S is a parameter that specifies the maximum strength of association allowed. The fan<sub>j</sub> term reflects the number of items that bear cue j. The term provides a way of quantifying the distinctiveness of a particular cue. The fan serves to decrease the associative strength between item i and cue j as a function of the number of total cues in memory that bear j.</p><list list-type="simple"><list-item><p>(9) S<sub>ji</sub> = S − ln(fan<sub>j</sub>)</p></list-item></list><p>Baseline activation is calculated according to (10), where d is the decay rate of a chunk's activation in memory at a given point since retrieval time t<sub>m</sub>.</p><list list-type="simple"><list-item><p>(10) B<sub>i</sub> = ln[Σ<sub>m</sub> t<sup>−d</sup><sub>m</sub>]</p></list-item></list><p>The chunk with the highest activation has the shortest retrieval latency (T<sub>i</sub>) as calculated according to the equation below, where <italic>F</italic> is a scaling parameter. The chunk with the shortest retrieval latency is the chunk that is retrieved in simulations.</p><list list-type="simple"><list-item><p>(11) T<sub>i</sub> = <italic>F</italic>e<sup>−A<sub>i</sub></sup></p></list-item></list><p>The model equations above contain a number of free parameters whose settings could impact the results of the simulation. We ran a series of simulations that systematically combined parameter values from across the range of those reported in previous work. Values of the <italic>total source activation</italic>, <italic>activation noise</italic>, <italic>fan</italic>, <italic>decay rate</italic>, and <italic>match-penalty</italic> parameters were manipulated<xref ref-type="fn" rid="fn0001"><sup>1</sup></xref>. The scaling factor (F) was held constant at 0.75 across all simulations. This resulted in the construction of 324 different models with unique parameter value combinations. As noted by Dillon et al. (<xref rid="B14" ref-type="bibr">2013</xref>), conducting such a sweep through the space of possible parameter values and combinations enables the identification of model predictions that are independent of idiosyncratic parameter combinations. 10,000 Monte Carlo simulations were run for each model, providing for each simulation a prediction of the most probable retrieval target and its retrieval latency.</p></sec><sec><title>Materials</title><p>We simulated antecedent retrieval time-locked to a position corresponding to the critical reciprocal in a sentence that contained three preceding NPs. The first NP, the <italic>subject</italic>, corresponded to a structurally appropriate antecedent for the reciprocal. The second NP, introduced at a lag after the subject NP, corresponded to a structurally inappropriate distractor. A third NP (NP3) was also introduced to more directly model the materials in our self-paced reading (SPR) experiment, the design of which is discussed below. The three NPs were introduced at 300 ms, 900 ms, and 1500 ms after simulation onset. Retrieval of the critical reciprocal was scheduled at 2400 ms after simulation onset.</p><p>Each NP in the simulation was marked with three features relevant for retrieval: its <italic>category</italic>, <italic>number</italic>, and <italic>clause index</italic>. All NPs bore the NP category feature. Number features could be either <italic>singular</italic> or <italic>plural</italic>. The <italic>clause index</italic> feature was used as a proxy feature for encoding an NP's structural appropriateness for the purposes of binding the reciprocal: the local licensing requirement is assumed to be satisfied if the antecedent bears the same clause index as the reciprocal. This indexing scheme can be viewed as a feature-based implementation of the clause-mate constraint on local anaphor licensing (see Lasnik, <xref rid="B22" ref-type="bibr">2002</xref> for a review of such constraints, which can differ in formulation from the c-command constraints of Chomsky, <xref rid="B11" ref-type="bibr">1981</xref>; Reinhart, <xref rid="B40" ref-type="bibr">1983</xref>).</p><p>Models were run to simulate four distinct conditions, corresponding to different feature combinations on the subject and distractor. The number features on the subject and the distractor were manipulated, resulting in the 2 × 2 factorial design schematized in (12). In grammatical conditions the subject was plural-marked, in ungrammatical conditions the subject was singular. In <italic>NoInterference</italic> conditions the distractor was singular, while in <italic>Interference</italic> conditions it was plural-marked. The structurally appropriate subject NP was marked with the <italic>main clause</italic> feature, while both the distractor and NP3 were marked as <italic>embedded</italic> and were therefore ineligible to antecede the reciprocal.</p><list list-type="simple"><list-item><p>(12)</p><list list-type="alpha-lower"><list-item><p>Grammatical-NoInterference</p><p>[Subject]+PL… [Distractor]+SG… [NP3]+SG… [RECIPROCAL]+PL</p></list-item><list-item><p>Grammatical-Interference</p><p>[Subject]+PL… [Distractor]+PL… [NP3]+SG… [RECIPROCAL]+PL</p></list-item><list-item><p>Ungrammatical-NoInterference</p><p>[Subject]+SG… [Distractor]+SG… [NP3]+SG… [RECIPROCAL]+PL</p></list-item><list-item><p>Ungrammatical-Interference</p><p>[Subject]+SG… [Distractor]+PL… [NP3]+SG… [RECIPROCAL]+PL</p></list-item></list></list-item></list><p>Antecedent retrieval at the reciprocal was modeled as specifying <italic>NP</italic> as a category cue and <italic>main clause</italic> as the clause cue. The number feature <italic>plural</italic> was also used in the retrieval cue set, to measure the interference effect associated with morphological features.</p></sec><sec><title>Results</title><p>We report three measures of interest from the simulations run for each condition: (i) predicted error rate, (ii) average predicted latency by condition, and (iii) predicted interference effect.</p><p>Predicted error rate corresponds to the percentage of the runs when the distractor, rather than the appropriate subject, was retrieved as an antecedent for the reciprocal. This measure is a relevant index of facilitatory interference in the ungrammatical conditions if facilitation stems from erroneous retrieval of the distractor instead of an appropriate target NP.</p><p>Predicted latency provides a measure of how long on average the winning retrieval should take in each condition. In simulations, the chunk with the shortest retrieval latency is the chunk that is retrieved from memory. According to the fully implemented ACT-R model, reading times on a particular word or phrase are the sum of the latency of retrieval triggered at that phrase and the amount of time associated with subsequent processing required by that word or phrase. Retrieval latencies should therefore map monotonically to reading times, with longer retrieval latencies corresponding to longer overall reading times, although the mental processes that intervene between retrieval and button-press may interact or contribute additional difficulty in such a way as to distort the underlying pattern of retrieval. Despite the possibility of later processing concealing underlying retrieval patterns, previous work has found a degree of relative transparency between the qualitative pattern of retrieval latencies furnished by the model and observed effects of facilitatory interference in self-paced reading or eye-tracking measures (see e.g., Wagers et al., <xref rid="B46" ref-type="bibr">2009</xref>; Dillon et al., <xref rid="B14" ref-type="bibr">2013</xref>).</p><p>The interference effect is a difference measure that compares average retrieval latencies between two conditions that differ on a single feature, as a way of estimating the magnitude and direction of interference contributed by the retrieval probe matching that one feature. We report two interference effects: the difference between the two grammatical conditions, as well as the difference between the two ungrammatical conditions. These comparisons provide a quantitative prediction of the effect of distractor plural marking when the features of the appropriate subject are held constant.</p></sec><sec><title>Predicted error rates</title><p>Error rates are reported in Table <xref ref-type="table" rid="T1">1</xref>. The error rates are consistent with a profile of facilitatory interference. Between the <italic>Ungrammatical</italic> conditions, plural marking on the distractor is predicted to increase rates of erroneous retrieval compared to when there is no NP in the sentence that matches the reciprocal in features (26.1 vs. 6.5%). On some proportion of trials, the recency of the distractor is predicted to increase the NP's baseline level of activation enough to result in it being the most highly-activated NP at retrieval. In the <italic>Ungrammatical-NoInterference</italic> condition, the distractor does not share any features with the reciprocal's cue set, so the main subject is still more likely to be retrieved, as it matches the retrieval probe's clause index cue. Error rate is expected to differ slightly between the two grammatical conditions: misretrieval of the distractor is 5.4% more common when it bears plural marking and the main subject matches the retrieval cues completely.</p><table-wrap id="T1" position="float"><label>Table 1</label><caption><p><bold>Retrieval error rates by condition for retrieval using morphological and syntactic cues calculated as the percentage of trials on which the distractor was retrieved across 10,000 runs each of 324 different models with unique parameter combinations</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th rowspan="1" colspan="1"/><th align="center" rowspan="1" colspan="1"><bold>NoInterference (%)</bold></th><th align="center" rowspan="1" colspan="1"><bold>Interference (%)</bold></th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">Grammatical</td><td align="center" rowspan="1" colspan="1">2.0</td><td align="center" rowspan="1" colspan="1">7.4</td></tr><tr><td align="left" rowspan="1" colspan="1">Ungrammatical</td><td align="center" rowspan="1" colspan="1">6.5</td><td align="center" rowspan="1" colspan="1">26.1</td></tr></tbody></table></table-wrap></sec><sec><title>Average predicted retrieval latencies</title><p>In the simulations the presence of a feature-matching subject has a facilitative effect on retrieval latencies (see Figure <xref ref-type="fig" rid="F1">1</xref>). Overall, retrieval times should be faster in the grammatical conditions because the grammatical subject, which matches the reciprocal's morphological and syntactic retrieval cues completely, is highly activated. Increased activation due to greater feature-match with the probe results in faster retrieval latencies in accordance with Equation (11). In the <italic>Ungrammatical</italic> conditions, where the main subject matches only on syntactic cues, retrieval latencies should be longer because the retrieved chunk should never match the probe completely. The appropriate subject only matches the probe's category and positional cues. The distractor matches the category cue and, in the <italic>Ungrammatical-Interference</italic> condition, the reciprocal's number feature. A pairwise difference is also predicted between the average retrieval latencies in the <italic>Ungrammatical-NoInterference</italic> and <italic>Ungrammatical-Interference</italic> conditions, which can be linked to the presence of morphological plural marking on the distractor. On the proportion of trials where the distractor is retrieved in the <italic>Ungrammatical-Interference</italic> condition, latencies are reduced relative to when the main clause subject is retrieved. This results in a reduction of average latency across retrievals.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Retrieval latencies by condition as predicted by the model in Experiment 1</bold>. Reported retrieval latencies represent the mean latency by condition across all simulations.</p></caption><graphic xlink:href="fpsyg-05-01252-g0001"/></fig></sec><sec><title>Interference effects</title><p>Predicted interference effects are shown in Table <xref ref-type="table" rid="T2">2</xref>. The grammatical interference effect was calculated by subtracting the average predicted latency in the <italic>Grammatical-Interference</italic> condition from the predicted latency in the <italic>Grammatical-NoInterference</italic> condition. The same difference was calculated for the two ungrammatical conditions. 95% confidence intervals represent the range of predicted interference effects across simulations.</p><table-wrap id="T2" position="float"><label>Table 2</label><caption><p><bold>Average interference effects across 10,000 runs each of 324 different models</bold>.</p></caption><table frame="hsides" rules="groups"><thead><tr><th rowspan="1" colspan="1"/><th align="center" rowspan="1" colspan="1"><bold>Interference effect</bold></th><th align="center" rowspan="1" colspan="1"><bold>Middle 95% of simulated distributions</bold></th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">Grammatical</td><td align="center" rowspan="1" colspan="1">+36 ms</td><td align="center" rowspan="1" colspan="1">+11, +82 ms</td></tr><tr><td align="left" rowspan="1" colspan="1">Ungrammatical</td><td align="center" rowspan="1" colspan="1">−63 ms</td><td align="center" rowspan="1" colspan="1">−18, −142 ms</td></tr></tbody></table></table-wrap><p>The simulation results predict that a plural-marked distractor should cause facilitatory interference in the ungrammatical conditions. The <italic>Ungrammatical-Interference</italic> condition exhibits faster average retrieval latencies than the <italic>Ungrammatical-NoInterference</italic> condition. Though the size of the effect varies, a facilitatory effect was consistently observed across all parameter combinations.</p><p>A small effect of inhibitory interference is also predicted in the grammatical conditions. This inhibition can be attributed to the fan effect (see Equation 8). In the <italic>Grammatical-Interference</italic> condition, the strength of association between the appropriate subject and the plural retrieval cue is decreased relative to the <italic>Grammatical-NoInterference</italic> condition, due to the presence of another plural-marked NP (the distractor).</p></sec><sec><title>Discussion</title><p>The goal of the simulations was to obtain predictions about the effect that a feature-matching but syntactically inappropriate distractor would have on the retrieval of an antecedent for a local reflexive if that retrieval used morphological features as cues that were weighted equally to syntactic cues.</p><p>The simulations show that when morphological cues are assigned the same weight as syntactic cues, the presence of a feature-matching distractor should decrease the parser's ability to retrieve a syntactically appropriate but feature-mismatching subject as an antecedent for a local anaphor. Some proportion of the time, the distractor is expected to be erroneously retrieved as a result of partial overlap with the retrieval cues. This misretrieval is predicted to have a facilitating effect on reading times in comparison to a case of retrieval when neither the distractor nor the local subject match the reflexive.</p></sec></sec><sec><title>Self-paced reading experiment</title><p>The modeling results predict that retrieval of a pre-verbal reciprocal's antecedent should display facilitatory interference effects from structurally inappropriate distractors, if morphological cues such as number are assigned the same weight as syntactic cues in retrieval. The experiment below used the self-paced reading method to investigate whether evidence of the predicted facilitatory interference would be found.</p><sec><title>Materials</title><p>The experiment had a 2 × 2 factorial design that matched the simulated conditions. The design manipulated the factors G<sc>rammaticality</sc> and I<sc>nterference</sc>. The structure of the test items is schematized in (13) and an example item is given in (14). All conditions contained a critical reciprocal (<italic>ek-dusre</italic>) that required a plural-marked antecedent in the main clause. The reciprocal was contained in a post-positional phrase that preceded a manner adverbial (<italic>gupt-rup-se</italic>, “<italic>secretly”</italic>) and the main clause verb (<italic>baat kii</italic>, “<italic>chatted” lit.</italic> “<italic>chat did”</italic>).</p><p>G<sc>rammaticality</sc> was manipulated by changing whether the main clause subject was plural-marked [<italic>Doctor(-oN)</italic>, “<italic>doctor(s)”</italic>]. Plural marking was unambiguously marked by the inflectional suffix <italic>–oN</italic>. In the grammatical conditions, the main clause subject was plural and could therefore act as a grammatical antecedent for the reciprocal. In the ungrammatical conditions, the local subject was singular and the reciprocal therefore lacked a clause-mate antecedent. The factor I<sc>nterference</sc> manipulated whether the distractor [<italic>mariiz-(oN)</italic>, “<italic>patient(s)”</italic>] was plural-marked.</p><p>In previous studies on local anaphor licensing (e.g., Sturt, <xref rid="B41" ref-type="bibr">2003</xref>; Dillon et al., <xref rid="B14" ref-type="bibr">2013</xref>) distractors have been positioned within relative clauses (RCs) attached to the main clause subject. RC-modification of subjects is a marked construction in Hindi, so the present study embedded the distractor inside a locative phrase that preceded the critical reciprocal.</p><p>The locative phrase contained an NP denoting a location modified by an animate possessor (<italic>nurse-ke steSan</italic>, “the nurse's station”). The distractor was embedded as the object of a verb within a prenominal RC that was attached to this possessor. In this position the distractor was not a clause-mate of the reciprocal and was therefore ineligible to act as a potential antecedent.</p><p>Critical reciprocals were always followed by a case marking post-position, either the genitive <italic>ke</italic>, the objective <italic>ko</italic>, or the dative <italic>se</italic>. When followed by the genitive, reciprocals were embedded in a complex post-position that was an argument to the main verb (e.g., <italic>ke bare-me</italic> “about” in 14). In sentences with <italic>ko</italic> or <italic>se</italic>, adverbial material was introduced after the post-position to maintain consistent length across sentences.</p><list list-type="simple"><list-item><p>(13) Subject-{sg/pl} [<sub>PP</sub>[<sub>RC</sub> Distractor-{sg/pl} V] NP's Location] Reciprocal P Adv V</p></list-item><list-item><p>(14)</p><list list-type="alpha-lower"><list-item><p>Grammatical-NoInterference</p><p>DoctoroN-ne mariiz-ki dekhbaal karne-wali nars-ke sTeSan-me ek-dusre ke-bare-me gupt-rup-se baat kii.</p><p>Doctors-Erg patient-Gen care doing-RP nurse's station-in each-other aboutsecretly chat did.</p><p>‘The doctors secretly spoke about each other in the station of the nurse taking care of (a/the) patient.’</p></list-item><list-item><p>Grammatical-Interference</p><p>DoctoroN-ne mariizoN-ki dekhbaal karne-wali nars-ke sTeSan-me ek-dusre ke-bare-me gupt-rup-se baat kii.</p><p>Doctors-Erg patients-Gen care doing-RP nurse's station-in each-other about secretly chat did.</p><p>‘The doctors secretly spoke about each other in the station of the nurse taking care of (the) patients.’</p></list-item><list-item><p>Ungrammatical-NoInterference</p><p>Doctor-ne mariiz-Gen dekhbaal karne-wali nars-ke sTeSan-me ek-dusre ke-bare-me gupt-rup-se baat kii.</p><p>Doctor-Erg patient-ki care doing-RP nurse's station-in each-other about secretly chat did.</p><p>‘The doctor secretly spoke about each other in the station of the nurse taking care of (a/the) patient.’</p></list-item><list-item><p>Ungrammatical-Interference</p><p>Doctor-ne mariizoN-ki dekhbaal karne-wali nars-ke sTeSan-me ek-dusre ke-bare-me gupt-rup-se baat kii.</p><p>Doctor-Erg patients-Gen care doing-RP nurse's station-in each-other about secretly chat did.</p><p>‘The doctor secretly spoke about each other in the station of the nurse taking care of (the) patients.’</p></list-item></list></list-item></list><p>Inside the pre-nominal RC the distractor bore either accusative or genitive case (according to the verb's requirements). Although this increased the contrast between the nominative grammatical subject and the distractor, it is unlikely that the case difference would play a role in distinguishing appropriate from inappropriate NPs, as accusative and genitive-marked NPs can serve as antecedents for local anaphors under the right structural conditions (see, e.g., Dayal, <xref rid="B13" ref-type="bibr">1994</xref>; Mohanan, <xref rid="B33" ref-type="bibr">1994</xref>; Bhatt and Dayal, <xref rid="B9" ref-type="bibr">2007</xref>).</p><p>A second concern with the experimental materials is that there exists the potential for temporary misanalysis of the structural position of the distractor during incremental parsing. When it initially encounters the distractor, the parser has not yet encountered any information that indicates that the distractor is contained within an embedded clause. In the absence of this information, an incremental parser is likely to analyze the distractor as a constituent of the main clause. This type of temporary misparse is common in head-final languages where embedded arguments can be encountered prior to the verb that licenses them (Inoue, <xref rid="B17" ref-type="bibr">1991</xref>; Mazuka and Itoh, <xref rid="B27" ref-type="bibr">1995</xref>; Miyamoto, <xref rid="B32" ref-type="bibr">2003</xref>). The misanalysis would be disconfirmed at the relative pronoun <italic>wali</italic>, at which point the object would be correctly reanalyzed as a constituent of the relative clause. This misparse should occur across all conditions, but it may have a greater impact on processing in the <italic>Ungrammatical-Interference</italic> condition. Under this misanalysis the RC-internal object would initially be analyzed as a suitable antecedent for an upcoming reciprocal. We return to the ability of such a misparse to affect later parsing decisions in the <italic>Ungrammatical-Interference</italic> condition in the discussion.</p></sec><sec><title>Participants</title><p>32 self-reported native speakers of Hindi were recruited from the student bodies of IIT, Delhi and Jawaharlal Nehru University in New Delhi (18 male, mean age = 20.1). Participants were compensated Rs. 300 for their participation, which lasted around 35 min.</p></sec><sec><title>Procedure</title><p>Participants were run on one of two laptop PCs using the Linger software package (Doug Rohde, MIT) in a self-paced word-by-word moving window paradigm (Just et al., <xref rid="B19" ref-type="bibr">1982</xref>). Each trial began with a sentence masked by dashes appearing on the screen. Letters and punctuation marks were masked, but spaces were left unmasked so that word-boundaries were visible. As the participant pressed the spacebar, a new word appeared and the previous word was re-masked. All text appeared in Devanagari font.</p><p>A yes/no comprehension question that probed its interpretation followed each sentence (experimental materials can be found at the first author's website). Participants were instructed to read sentences at a natural pace and to respond to the comprehension questions as accurately as possible. Participants responded to questions using the f-key for “yes” and the j-key for “no.” If the question was answered incorrectly the word <italic>galat</italic> (“incorrect/wrong”) appeared briefly in the center of the screen. Each participant was randomly assigned to one of the lists and the order of the stimuli within the presentation list was randomized for each participant.</p></sec><sec><title>Analysis</title><p>Data from one participant were excluded due to failure to comply with experimental guidelines. Data from another participant were excluded because the participant's mean accuracy on comprehension questions was close to chance. This resulted in the data of 30 subjects being used for later analysis. Two items were excluded from analysis due to errors.</p><p>Statistical analyses were carried out on log-transformed reading times using linear mixed effects regression (Baayen et al., <xref rid="B5" ref-type="bibr">2008</xref>). Reading times from both correct and incorrect trials were included in the analysis. Experimental fixed effects were the simple difference sum-coded factors G<sc>rammaticality</sc> and I<sc>nterference</sc> and their interaction. All models included random intercepts for both subjects and items. Models with a maximal random effects structure were fit whenever possible (Barr et al., <xref rid="B8" ref-type="bibr">2013</xref>). If a maximal model failed to converge, a model was used that contained only by-subject random slopes for both fixed effects and their interaction.</p></sec><sec><title>Results</title><sec><title>Comprehension Question Accuracy</title><p>Comprehension question accuracy averaged 69.2%. No significant differences were found in average accuracy across conditions (logistic mixed effects model, all <italic>z</italic>s < 1).</p></sec><sec><title>Reading Time Results</title><p>Reading times from the post-reciprocal region are given in Figure <xref ref-type="fig" rid="F2">2</xref>.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Average word-by-word self-paced reading times for all items in Experiment 1</bold>.</p></caption><graphic xlink:href="fpsyg-05-01252-g0002"/></fig><p><bold><italic>Pre-reciprocal region.</italic></bold> No significant effects were found in the pre-reciprocal region.</p><p><bold><italic>Reciprocal region.</italic></bold> No significant effects were found in the reciprocal region.</p><p><bold><italic>Post-position region.</italic></bold> Average reading times were reliably faster in the grammatical conditions than in the ungrammatical conditions (main effect of G<sc>rammaticality</sc>: <inline-formula><mml:math id="M1"><mml:mover accent="true"><mml:mi>β</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:math></inline-formula> = −0.088, s.e. = 0.034, <italic>t</italic> = −2.92); see Figure <xref ref-type="fig" rid="F3">3</xref>. Although reading times in the <italic>Ungrammatical-Interference</italic> condition were numerically longer than those in the <italic>Ungrammatical-NoInterference</italic> condition, the G<sc>rammaticality</sc> × I<sc>nterference</sc> interaction was not significant (<italic>t</italic> = 1.41). No reliable pairwise differences were observed between ungrammatical conditions (<italic>t</italic> < 1).</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Average post-reciprocal self-paced reading times in Experiment 1</bold>. Error bars indicate standard error of the participant mean.</p></caption><graphic xlink:href="fpsyg-05-01252-g0003"/></fig><p><bold><italic>Reciprocal+2 region.</italic></bold> There were no significant main effects two regions after the critical reciprocal, but the model revealed a marginally significant G<sc>rammaticality</sc> × I<sc>nterference</sc> interaction (<inline-formula><mml:math id="M2"><mml:mover accent="true"><mml:mi>β</mml:mi><mml:mo>^</mml:mo></mml:mover></mml:math></inline-formula> = 0.105, s.e. = 0.054, <italic>t</italic> = 1.96) two regions after the reciprocal. This interaction reflected the fact that the <italic>Ungrammatical-Interference</italic> condition was read more slowly than any other condition, including the <italic>Ungrammatical-NoInterference</italic> condition. The pairwise comparison between the two ungrammatical conditions revealed the numerical difference between the two conditions not to be significant (<italic>t</italic> = 1.3). However, given the relatively low power of the current study, it is possible that this interaction would achieve significance with higher power. We return to this interaction effect in the discussion.</p><p><bold><italic>Reciprocal+3 till Final region.</italic></bold> No significant effects were observed in any subsequent region.</p></sec></sec><sec><title>Discussion</title><p>The SPR experiment sought to determine whether the processing of a pre-verbal reciprocal in Hindi was subject to facilitatory interference. The study manipulated the number features on a structurally appropriate antecedent for the reciprocal, as well as the features of the structurally inappropriate distractor, as a means of testing whether (equally weighted) morphological cues are used to access a local anaphor's antecedent.</p><p>When a structurally appropriate feature-matching antecedent was present to license the pre-verbal reciprocal the regions following the critical reciprocal were read more rapidly than when there was no feature-matching and structurally appropriate antecedent. In contrast to the prediction of the model simulations, we failed to find any evidence of facilitatory interference (see Figure <xref ref-type="fig" rid="F4">4</xref>). In fact, the empirical results trend in the opposite direction; there were clear inhibitory effects. The post-reciprocal region in the <italic>Ungrammatical-Interference</italic> condition was read at a comparable or slightly slower rate than the processing of the reciprocal in the <italic>Ungrammatical-NoInterference</italic> condition. Despite the fact that our study potentially lacks the power to observe an interference effect, we are more secure in our conclusion that there is a lack of facilitatory interference in light of the direction of the numerical trend toward an interaction in the post-reciprocal region.</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>Comparison of predicted interference effects from model simulations and observed interference effects from Experiment 1</bold>. Models simulated expected retrieval latencies if morphological and positional cues were assigned equal weights in antecedent retrieval. For the simulated data, error bars represent the middle 95% of the distribution of predicted interference effects. Error bars around the empirical means mark the 95% CI.</p></caption><graphic xlink:href="fpsyg-05-01252-g0004"/></fig><p>Two words downstream from the reciprocal, reading times were longest when the local subject did not match the features of the reciprocal but the features of the distractor did match the reciprocal's number features.<xref ref-type="fn" rid="fn0002"><sup>2</sup></xref> We discuss this effect below because although it is inconsistent with the predictions of our simulations, it does potentially indicate that the distractor's morphological features may affect overall processing of the reciprocal.</p><p>The mechanism by which the distractor exerts inhibitory influence on reciprocal licensing is unclear. It is commonly assumed that inhibitory interference should occur when multiple items in memory match a retrieval cue (e.g., Badecker and Straub, <xref rid="B7" ref-type="bibr">2002</xref>; Lewis and Vasishth, <xref rid="B23" ref-type="bibr">2005</xref>; Van Dyke and McElree, <xref rid="B44" ref-type="bibr">2011</xref>). Yet, we observed inhibition in the absence of a multiple-match configuration: the main subject matched the positional cue and the distractor matched the number cue. This suggests that the mechanism used to explain inhibition in multiple-match cases (e.g., the fan effect in Lewis and Vasishth's, <xref rid="B23" ref-type="bibr">2005</xref> model), is not the appropriate explanation for our finding. We consider three possible explanations of this inhibitory effect and the role that number features play in guiding initial retrieval under each scenario.</p><p>The first possible interpretation of the inhibitory effect links the slightly delayed slowdown to erroneous retrieval of the distractor during initial memory access. The increased reading times in our SPR experiment might reflect initial misretrieval of the distractor based on its morphological overlap with the probe, followed by the increased processing cost of inhibiting that distractor. This line of reasoning has been pursued by Patil et al. (<xref rid="B38" ref-type="bibr">2012</xref>) and Chen et al. (<xref rid="B10" ref-type="bibr">2012</xref>) to explain inhibitory effects in reflexive licensing. We consider this interpretation unlikely for the present data because we see no evidence of the erroneous retrieval on which the explanation is predicated. In light of subject-verb agreement and NPI licensing effects, we would expect initial misretrieval to result in some degree of facilitation, however fleeting, that would be observable in the self-paced reading times. These facilitatory interference effects consistently yield large effects on reading times in studies of other linguistic dependencies. No such facilitation was observed prior to the point of inhibitory interference in the current study.</p><p>The inhibitory effect might also be explained in terms of <italic>cue-confusability</italic>, as defined by Jäger et al. (<xref rid="B18" ref-type="bibr">2014</xref>). The proposal rests on the speculation that cues that reliably co-occur in specific retrieval contexts can be confused (less effectively deployed). In reciprocal licensing the clause and plural cues are reliably associated because both cues should be selected whenever a reciprocal is encountered. This contrasts with cue association in reflexive licensing where specific gender cues (e.g., masculine and feminine) and the clause-mate cue co-occur less reliably, e.g., it is not the case that reflexive licensing uniformly uses masculine gender. According to the proposal, confusion is more likely to occur in reciprocal licensing than in reflexive licensing. Although we note that this is a possibility in principle, we believe that the notion of cue-confusability or the mechanism by which confusion creates retrieval interference has not been sufficiently articulated to be thoroughly evaluated.</p><p>The third alternative interpretation of the effect connects the slowdown to the influence of an abandoned early garden-path parse that analyzed the distractor as a constituent of the main clause. The previous partial parse could provide an appropriately marked antecedent for the reciprocal, but would fail to provide a coherent global parse. There are no grammatical re-parses of the sentence that would allow the distractor to be reanalyzed as an appropriate antecedent for the reciprocal. We hypothesize that resolving the tension between attempting to license the reciprocal and building a globally grammatical parse of the sentence is the source of the observed interaction. The misparse is expected to intrude on the processing of the reciprocal in the <italic>Ungrammatical-Interference</italic> condition, where consideration of the reparse would result in a structurally appropriate, feature-matching antecedent for the reciprocal.</p><p>We favor the interpretation that this inhibitory effect reflects the influence of the mis-parse on repair strategies that are triggered by failure of initial antecedent retrieval (as proposed for similar effects by, Sturt, <xref rid="B41" ref-type="bibr">2003</xref>; Chow et al., <xref rid="B48" ref-type="bibr">2014</xref>). On this interpretation the failure to retrieve an appropriate antecedent for the reciprocal would initiate a more liberal search for a feature-matching phrase, or would attempt to find an alternative parse for the sentence under which the reciprocal could be grammatically bound. These repair procedures are argued to be less constrained by the structure of the previous parse (and therefore structural constraints), perhaps reflecting uncertainty in the structural analysis in light of the error signal. This scenario attributes the increase in reading times to interference, but not interference that occurs during antecedent retrieval. Rather, the locus of interference lies in retrievals associated with syntactic revision and reanalysis. It is also possible that the distractor in the mis-parsed sentence could contribute interference at retrieval time, a possibility that would be consistent with the numerical trend toward an interaction in the post-reciprocal region. We acknowledge that the present study cannot distinguish between these two options.</p><p>In sum, our SPR experiment failed to find the characteristic profile of facilitatory interference that has been found in other studies on the construction of subject-verb agreement, NPI-licensing, and control dependencies and is predicted under a cue-based retrieval model that uses morphological cues to access potential antecedents for a local anaphor. Instead, a feature-matching distractor triggered a delayed inhibitory effect when the local subject could not antecede the reciprocal in Hindi. We argued that this process was not an indication of interference during antecedent retrieval, but rather interference during a repair process subsequent to antecedent retrieval.</p></sec></sec><sec><title>General discussion</title><p>The purpose of the present study was to assess whether syntactic cues are given priority over morphological cues in the retrieval of antecedents of pre-verbal reciprocals in Hindi. Investigating the processing of Hindi reciprocals helps to establish whether the absence of facilitatory interference effects from morphologically-matched distractors in previous experiments was due to a confound of anaphor position. We hypothesized that if the absence of interference were solely due to the post-verbal position of the anaphor, and not prioritization of syntactic cues, interference would be observable in the retrieval of an antecedent for a pre-verbal anaphor in Hindi.</p><p>In our self-paced reading study native Hindi speaking participants resolved a local reciprocal dependency more quickly when the main clause subject was plural than when no grammatical antecedent was present. The presence of a feature-matching distractor did not induce reliable effects of facilitatory interference when the local subject did not match the reciprocal in features. These findings are consistent with a general lack of facilitation in the licensing of local anaphors found in previous work (e.g., Sturt, <xref rid="B41" ref-type="bibr">2003</xref>; Xiang et al., <xref rid="B47" ref-type="bibr">2009</xref>; Dillon et al., <xref rid="B14" ref-type="bibr">2013</xref>), and with lack of interference during local anaphor licensing more generally (e.g., Nicol and Swinney, <xref rid="B36" ref-type="bibr">1989</xref>; Clackson, <xref rid="B12" ref-type="bibr">2011</xref>). The presence of a feature-matching distractor produced a delayed inhibitory effect when an appropriate antecedent for the reciprocal could not be found. We reasoned that the inhibitory effect in our experiment might have arisen as a result of error-driven repair strategies, and not from participants accessing the distractor during initial antecedent retrieval.</p><p>The empirical results of our SPR experiment were compared against the results of a series of simulations that modeled latencies and error rates of a cue-based retrieval process that used equally-weighted morphological and positional cues to retrieve antecedents of a local anaphor. The empirical results did not align with the simulations' prediction that there should be facilitatory interference between ungrammatical conditions.</p><p>Overall, the results lend support to the hypothesis that the lack of facilitatory interference in local anaphor antecedent retrieval is not primarily determined by an anaphor's post-verbal position. In particular, the Hindi results appear to be incompatible with a number of the possible ways in which verbal adjacency could influence retrieval of antecedents for local anaphors discussed. The results cast doubt on explanations that rely on recent reactivation of the grammatical antecedent immediately before the reciprocal. In the Hindi materials there is no point at which retrieval of the subject is required between the distractor and when the reciprocal is encountered.</p><p>The results are consistent with models of cue-based antecedent retrieval that prioritize syntactic information in one manner or another. As noted in the introduction, a parser could be said to prioritize syntactic cues by assigning them greater weight than morphological cues, or by using syntactic cues exclusively.</p><p>Because some dependencies display facilitatory interference effects while others do not, it would appear that retrieval does not consistently prioritize positional cues. One question that arises is how the parser determines when it should prioritize syntactic cues. Rational models often assume that retrieval uses a set of cues and weights that maximizes the probability of retrieving the target, while minimizing the chances of interference. It is important to note that the optimal cue set for meeting both of these goals may change as a function of (i) the dependency being computed and (ii) the local syntactic context. Therefore, strategic considerations that take the local context into account may comprise an important part of the cue selection procedure. We term different solutions that the parser could adopt <italic>retrieval strategies</italic>.</p><p>The parser could adopt one of two strategies that make different use of morphological cues during local antecedent retrieval. First, the parser could uniformly prioritize syntactic cues for all instances of local antecedent retrieval regardless of syntactic context. Dillon et al. (<xref rid="B14" ref-type="bibr">2013</xref>) proposed that the parser implements such a retrieval strategy. According to these authors, local antecedent retrieval only uses structural cues.</p><p>An alternative to this proposal is that the parser could condition the use of morphological cues on the local syntactic context of the anaphor that triggers retrieval, as proposed by Kush (<xref rid="B21" ref-type="bibr">2013</xref>). The intuition behind this proposal stems from the observation that in certain environments structural cues alone may not suffice to identify a unique antecedent for a local anaphor. If the subject of the local clause is the anaphor's only co-argument, as it is in (15), then syntactic cues are sufficient to guarantee its retrieval. However, if there exists an additional co-argument that precedes the anaphor as in (16), a syntactic cue like the clause feature would not be able to distinguish the appropriate antecedent (<italic>the boys</italic>) from the structurally appropriate, but feature-mismatching NP <italic>Mary</italic>.</p><list list-type="simple"><list-item><p>(15) <italic>The boys</italic> spoke with <italic>each other</italic>.</p></list-item><list-item><p>(16) Mary introduced <italic>the boys</italic> to <italic>each other</italic></p></list-item></list><p>Kush (<xref rid="B21" ref-type="bibr">2013</xref>) proposed that a parser that could determine the number of clause-mates that preceded a local anaphor might use morphological cues to help guarantee retrieval of an appropriate antecedent. Determining whether the local subject is the anaphor's only clause-mate should be possible by consulting the local syntactic context. When processing English reflexives in direct object position, the anaphor's adjacency to the verb would be sufficient. In Hindi, verbal adjacency cannot be exploited to make such a determination. Kush (<xref rid="B21" ref-type="bibr">2013</xref>) proposed that the decision could be made if cue selection had access to the phrase structure rule being used to incrementally parse the input sentence. In cases where the anaphor is the first NP encountered during the incremental parse of the VP, the phrase structure predicted for the VP should not contain co-argument NPs. On the other hand, if the parser encounters a non-subject co-argument that precedes the reciprocal, the PS rule for the VP would reflect its presence and cue selection could determine that the clause index cue would no longer provide diagnostic access to the local subject.</p><p>If the parser adopts this retrieval strategy interference effects are predicted to emerge when there are non-subject clause-mates that precede a local anaphor. This proposal is consistent with recent findings from Wagers and colleagues, which suggest that that resistance to interference is, in fact, selectively conditioned on whether the anaphor is encountered after another co-argument (King et al., <xref rid="B20" ref-type="bibr">2012</xref>). Under this interpretation, interference should emerge if a co-argument preceded the reciprocal in Hindi, as in (17). We leave testing this prediction to future work.</p><list list-type="simple"><list-item><p>(17) <sup></sup>Larke-ne Mary-ko baccoN-ki party me ek-dusre ke-bare-me bataayaa.</p></list-item><list-item><p>Boy-Erg Mary-Acc kids' party in one-another about told.</p></list-item><list-item><p><sup></sup>The boy told Mary during the kids' party about <italic>each other</italic>.</p></list-item></list></sec><sec sec-type="conclusion" id="s2"><title>Conclusion</title><p>In this paper we asked whether the absence of intrusive licensing during local anaphor antecedent retrieval is restricted to post-verbal anaphors, or whether the lack of interference indicates a more general cross-linguistic state of affairs. We investigated the effect of a feature-matching distractor on the processing of unlicensed pre-verbal reciprocals in Hindi and found no indication of facilitatory interference. The results suggest that antecedent retrieval's ability to accesses the syntactically appropriate subject when licensing a local anaphor does not depend on direct verbal adjacency between the anaphor and its verb. The results appear to be better explained by a cue-based retrieval process that prioritizes, or exclusively uses, structural cues over morphological features. Finally, although we did not find evidence that a feature-matching distractor facilitates the processing of an unlicensed reciprocal, it did appear that a distractor might exert an inhibitory influence on some stage of reciprocal resolution. Future work should test whether this inhibition is a general effect, or whether its appearance is related to properties of the materials used here.</p><sec><title>Conflict of interest statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec></sec>
Writing and Low-Temperature Characterization of Oxide Nanostructures	<p>Oxide nanoelectronics is a rapidly growing field which seeks to develop novel materials with multifunctional behavior at nanoscale dimensions. Oxide interfaces exhibit a wide range of properties that can be controlled include conduction, piezoelectric behavior, ferromagnetism, superconductivity and nonlinear optical properties. Recently, methods for controlling these properties at extreme nanoscale dimensions have been discovered and developed. Here are described explicit step-by-step procedures for creating LaAlO<sub>3</sub>/SrTiO<sub>3</sub> nanostructures using a reversible conductive atomic force microscopy technique. The processing steps for creating electrical contacts to the LaAlO<sub>3</sub>/SrTiO<sub>3</sub> interface are first described. Conductive nanostructures are created by applying voltages to a conductive atomic force microscope tip and locally switching the LaAlO<sub>3</sub>/SrTiO<sub>3</sub> interface to a conductive state. A versatile nanolithography toolkit has been developed expressly for the purpose of controlling the atomic force microscope (AFM) tip path and voltage. Then, these nanostructures are placed in a cryostat and transport measurements are performed. The procedures described here should be useful to others wishing to conduct research in oxide nanoelectronics.</p>	<contrib contrib-type="author"><name><surname>Levy</surname><given-names>Akash</given-names></name><xref ref-type="aff" rid="ID1"> <sup>1</sup> </xref></contrib><contrib contrib-type="author"><name><surname>Bi</surname><given-names>Feng</given-names></name><xref ref-type="aff" rid="ID1"> <sup>1</sup> </xref></contrib><contrib contrib-type="author"><name><surname>Huang</surname><given-names>Mengchen</given-names></name><xref ref-type="aff" rid="ID1"> <sup>1</sup> </xref></contrib><contrib contrib-type="author"><name><surname>Lu</surname><given-names>Shicheng</given-names></name><xref ref-type="aff" rid="ID1"> <sup>1</sup> </xref></contrib><contrib contrib-type="author"><name><surname>Tomczyk</surname><given-names>Michelle</given-names></name><xref ref-type="aff" rid="ID1"> <sup>1</sup> </xref></contrib><contrib contrib-type="author"><name><surname>Cheng</surname><given-names>Guanglei</given-names></name><xref ref-type="aff" rid="ID1"> <sup>1</sup> </xref></contrib><contrib contrib-type="author"><name><surname>Irvin</surname><given-names>Patrick</given-names></name><xref ref-type="aff" rid="ID1"> <sup>1</sup> </xref></contrib><contrib contrib-type="author"><name><surname>Levy</surname><given-names>Jeremy</given-names></name><xref ref-type="aff" rid="ID1"> <sup>1</sup> </xref></contrib>	Journal of Visualized Experiments : JoVE	<fig id="Fig_51886" orientation="portrait" position="anchor"><alternatives><media id="Video_51886" xlink:href="jove-89-51886.mp4" mimetype="video" mime-subtype="mp4" orientation="portrait" xlink:type="simple" position="anchor"/><graphic xlink:href="jove-89-51886-thumb"/></alternatives></fig><sec><title>Introduction</title><p>Oxide heterostructures<sup>1-5</sup> exhibit a remarkably wide variety of emergent physical phenomena which are both scientifically interesting and potentially useful for applications<sup>4</sup>. In particular, the interface between LaAlO<sub>3</sub> (LAO) and SrTiO<sub>3</sub> (STO)<sup>6</sup> can exhibit insulating, conducting, superconducting<sup>7</sup>, ferroelectric-like<sup>8</sup>, and ferromagnetic<sup>9</sup> behavior. In 2006, Thiel <italic>et al</italic> showed<sup>10</sup> that there is a sharp insulator-to-metal transition as the thickness of the LAO layer is increased, with a critical thickness of 4 unit cells (4uc). It was subsequently shown that 3uc-LAO/STO structures exhibit a hysteretic transition that can be controlled locally with a conductive atomic-force microscope (c-AFM) probe<sup>11</sup>.</p><p>The properties of oxide interfaces such as LaAlO<sub>3</sub>/SrTiO<sub>3</sub> depend on the absence or presence of conducting electrons at the interface.  These electrons can be controlled using top gate electrodes <sup>12,13</sup>, back gates <sup>10</sup>, surface adsorbates<sup>14</sup>, ferroelectric layers<sup>15,16</sup> and c-AFM lithography<sup>11</sup>.  A unique feature of c-AFM lithography is that very small nanoscale features can be created.</p><p>Electrical top gating, combined with two-dimensional confinement, is often used to create quantum dots in III-V semiconductors <sup>17</sup>.  Alternatively, quasi-one-dimensional semiconducting nanowires can be electrically gated by proximity.  The methods for producing these structures are time-consuming and generally irreversible. By contrast, the c-AFM lithography technique is reversible in the sense that a nanostructure can be created for one experiment, and then “erased” (similar to a whiteboard). Generally, c-AFM writing is performed with positive voltages applied to the AFM tip, while, erasing is performed using negative voltages.  The time required to create a particular structure depends on the complexity of the device but is usually less than 30 min; most of that time is spent erasing the canvas.  The typical spatial resolution is about 10 nanometers, but with proper tuning features as small as 2 nanometers can be created<sup>18</sup>.</p><p>A detailed description of the nanoscale fabrication procedure follows. The detail provided here should be sufficient to allow similar experiments to be performed by interested researchers. The method described here has many advantages over traditional lithographic approaches used to create electronic nanostructures in semiconductors.  </p><p>The c-AFM lithography method described here is part of a much broader class of scanning-probe-based lithography efforts, including scanning anodic oxidation<sup>19</sup>, dip-pen nanolithography<sup>20</sup>, piezoelectric patterning<sup>21</sup>, and so on.  The c-AFM technique described here, coupled with the use of novel oxide interfaces, can produce some of the highest-precision electronic structures with an unprecedented variety of physical properties.  </p></sec><sec><title>Protocol</title><sec><title>1. Obtain LAO/STO Heterostructures</title><list list-type="order"><list-item><p>Obtain an oxide heterostructure consisting of 3.4 unit cells of LAO grown by pulsed laser deposition on TiO<sub>2</sub>-terminated STO substrates. Details of sample growth are described in Ref. <sup>22</sup>.</p></list-item></list></sec><sec><title>2. Photolithographic Processing of Samples</title><p>Create electrical contacts to the LAO/STO interface, with bonding pads for wiring canvases to a chip carrier. The individual processing steps are described in detail below.</p><list list-type="order"><list-item><p>Spin photoresist <list list-type="order"><list-item><p>Spin photoresist on the samples at 600 rpm for 5 sec, then at 4,000 rpm for 30 sec. The photoresist layer will be about 2 µm thick. Bake the samples at 95 °C for 1 min.</p></list-item></list> </p></list-item><list-item><p>Expose photoresist using a mask aligner with 320 nm light for 100 sec with a dose of 5 mW/cm<sup>2</sup>.</p></list-item><list-item><p>Develop the photoresist in photoresist developer for 1 min.</p></list-item><list-item><p>Ion milling <list list-type="order"><list-item><p>Use an Ar<sup>+</sup> ion mill to remove 15 nm of material (LAO and STO) in the areas not covered by photoresist. Place the samples at a 22.5° angle to the direction perpendicular to the incoming Ar+ ion beam. If the Ar+ etching rate is not calibrated, perform a calibration run to ensure that the correct amount of material is removed. Determine the etching depth using AFM or equivalent profilmetry.</p></list-item></list> </p></list-item><list-item><p>DC sputtering of Ti and Au <list list-type="order"><list-item><p>Deposit 4 nm Ti, then 25 nm Au onto the samples so that the Au makes electrical contact with the exposed STO layer. The sputtering pressure is in the range 2-6 x 10<sup>-7</sup> Torr, and the sputtering takes place with the sample at RT. Pre-sputter Ti for 10 min with shutter closed at 100 W, then open shutter and sputter for 20 sec at 100 W. Upon completion, immediately pre-sputter Au for 1 min at 50 W then sputter Au for 30 sec to the samples at 50 W. Calibrate the time to produce the desired Ti and Au thicknesses.</p></list-item></list> </p></list-item><list-item><p>Lift-off <list list-type="order"><list-item><p>Use Acetone/IPA ultrasonic wash to remove photoresist from the surface of the samples.</p></list-item></list> </p></list-item><list-item><p>Second layer <list list-type="order"><list-item><p>A second lithographic process, excluding step 4 (<italic>i.e</italic>., excluding ion milling), is used to create gold wire connections to individual bonding pads. The two patterns must be well-aligned to ensure that they do not produce electrical shorts.</p></list-item></list> </p></list-item><list-item><p>Plasma cleaning. <list list-type="order"><list-item><p>An IPC Barrel Etcher is used to remove the photoresist residue in the pattern trench. The instrument used at the 100 W and 1 Torr argon for 1 min</p></list-item></list> </p></list-item></list></sec><sec><title>3. Wire Bond a Sample to Prepare for Writing</title><list list-type="order"><list-item><p>Mount the LAO/STO sample in a chip carrier (<bold>Figure 2A</bold>) with 28 available pins.</p></list-item><list-item><p>Wire bond structure</p></list-item></list><p>NOTE: Use a wire bonder to make electrical connections between bonding pads on the sample and the chip carrier. Attach 1 mil (25 micrometer) gold wires between the electrical contacts and the chip carrier. Write nanostructures</p></sec><sec><title>4. Write Nanostructures</title><list list-type="order"><list-item><p>Create an informal sketch of the conductive nanostructure (<bold>Figure 3A</bold>).</p></list-item><list-item><p>Open the scalable vector graphics (SVG) editor (<bold>Figure 3B</bold>). <list list-type="order"><list-item><p>Use a template or define the window size to match that of the AFM image.</p></list-item><list-item><p>Load the AFM image of the sample into the SVG editor.</p></list-item><list-item><p>Create nanostructure elements overlaid on the AFM image.</p></list-item></list> </p></list-item><list-item><p>Load the SVG file into the nanolithography program.</p></list-item><list-item><p>Run the lithography software to create a conductive nanostructure. <list list-type="order"><list-item><p>Use <italic>V</italic><italic><sub>tip</sub></italic>=+10 V to create nanostructures, and <italic>V</italic><italic><sub>tip</sub></italic>=-10 V to erase nanostructures.</p></list-item><list-item><p>Move the c-AFM tip at a speed ranging from 200 nm/sec to 2 µm/sec.</p></list-item></list> </p></list-item></list></sec><sec><title>5. Cool Device and Take Measurements</title><list list-type="order"><list-item><p>Turn off all white lights and use red filters/light sources.</p></list-item><list-item><p>Extract the sample from the AFM system.</p></list-item><list-item><p>Load the sample into the dilution refrigerator (A).</p></list-item><list-item><p>Measure resistance vs. temperature (B) as the sample is cooled.</p></list-item><list-item><p>Measure transport properties at low temperatures (C).</p></list-item></list></sec></sec><sec><title>Representative Results</title><p>The results shown here are representative of the transport behavior that can be exhibited by this class of nanostructures, and has been described elsewhere in detail<sup>23-26</sup>. In this example, a nanowire cavity has been constructed (<bold>Figure 4</bold>) from a 3.3 unit cell LAO/STO heterostructure. Conductive paths (shown in green) are typically 10 nm wide, as determined by nanowire “cutting” experiments<sup>11</sup>. The tip speed and voltage for each segment is independently configurable from the lithography front panel (<bold>Figure 4B</bold>), as is the tip writing speed. “Virtual electrodes” that interface with the interfacial contacts ensure that there is a highly conductive electrical connection to the nanostructures.</p><p>After the nanostructure is written, it is transferred to the dilution refrigerator. Exposure to light at or below 550 nm will produce unwanted photoconduction, so it is important to transfer the device in darkness or with the aid of a red “darkroom” light (<bold>Figure 5A</bold>). Electrical connections should be made at RT, and as with most semiconductor nanostructures, great care should be taken when changing electrical connections at cryogenic temperatures. If the devices is subjected to electrostatic discharge, it will most likely become insulating. Remarkably, the device functionality can be recovered by “cycling” the temperature to 300 K and cooling down again.</p><p>During cooldown, it is routine to monitor the two-terminal resistance, and even the four-terminal resistance, as a function of temperature. For these measurements an ac voltage (typically ~1 mV) is applied at a low frequency (<10 Hz) to one of the electrodes, while the ac current is measured using a transimpedance amplifier. Lock-in demodulation and filtering is performed using a home-developed lock-in amplifier. The ac current is monitored as a function of temperature (<bold>Figure 5B</bold>).</p><p>Once the device is cooled to the base temperature of the dilution refrigerator (50 mK), four-terminal transport measurements are performed (<bold>Figure 5C</bold>). For these measurements, current is sourced through the main channel of the device, while voltage across the device is simultaneously measured. Instead of measuring with a lock-in amplifier, a full current-voltage (I-V) trace is measured. This method contains more information and the differential conduction can be calculated via numerical differentiation. For the particular device, the differential conduction is measured as a function of the side-gate voltage <italic>V</italic><italic><sub>sg</sub></italic>. This gate allows the chemical potential of the device to be changed. The transport through the device shows a strong non-monotonic dependence, indicating regions in which Coulomb blockade takes place for smaller values, and strong superconductivity for larger values of <italic>V</italic><italic><sub>sg</sub></italic> . Details about the physical interpretation for this class of device will be described elsewhere.</p><p><graphic xlink:href="jove-89-51886-0.jpg" position="float" orientation="portrait"/><bold>Figure 1.</bold><bold>Photolithographic processing steps.</bold> Step 1: spin photoresist. Step 2: expose photoresist using mask aligner. Step 3: develop photoresist. Step 4: ion milling. Step 5: DC sputtering to deposit Ti and Au. Step 6: lift-off. Step 7: deposit the second layer. Step 8: plasma cleaning.</p><p><graphic xlink:href="jove-89-51886-1.jpg" position="float" orientation="portrait"/><bold>Figure 2</bold>. <bold>Images of lithographically patterned LAO/STO heterostructures.</bold> <bold>(A)</bold> Image showing 5mm x 5mm sample wire bonded to a chip carrier. <bold>(B)</bold> Optical image showing bonding pads and one of the canvases. <bold>(C)</bold> Close-up of a single canvas. <ext-link ext-link-type="uri" xlink:href="https://www.jove.com/files/ftp_upload/51886/51886fig2highres.jpg">Please click here to view a larger version of this figure.</ext-link></p><p><graphic xlink:href="jove-89-51886-2.jpg" position="float" orientation="portrait"/><bold>Figure 3</bold>. <bold>(A)</bold> Informal design of LAO/STO nanostructure. <bold>(B)</bold> Precise layout of nanostructure using an open-source scalable vector graphics (SVG) editor.</p><p><graphic xlink:href="jove-89-51886-3.jpg" position="float" orientation="portrait"/><bold>Figure 4</bold>. <bold>(A)</bold> Lithography front panel for c-AFM patterning. <bold>(B)</bold> Screenshot from 3D simulator showing position and voltage of c-AFM tip. <ext-link ext-link-type="uri" xlink:href="https://www.jove.com/files/ftp_upload/52058/52058fig4large.jpg">Please click here to view a larger version of this figure.</ext-link></p><p><graphic xlink:href="jove-89-51886-4.jpg" position="float" orientation="portrait"/><bold>Figure 5</bold>. <bold>(A)</bold> LAO/STO nanostructure being inserted into dilution refrigerator. <bold>(B)</bold> Monitoring of sample resistance as it is cooled from 300 K to 50 mK.<bold> (C)</bold> Monitoring of four-terminal differential conductance of device as a function of side gate voltage Vsg and voltage across the device (V4t). Intensity graph displayed in units of siemens (S), and voltages are displayed in units of volts (V).</p></sec><sec><title>Discussion</title><p>Successful creation of nanostructures depends on several critical steps. It is important that the LAO/STO samples are grown with a thickness that is known to be at the boundary between the insulating and conductive phase. (Details of sample growth fall outside the scope of this paper, but are crucial for overall success.) Second, it is important to have relative humidity within the range 25-45% for successful c-AFM writing. Values below 25% are unlikely to produce conductive nanostructures, while too high humidity will generally produce uncontrollably large features. Also, temperature control of the AFM is important if the c-AFM tip needs to achieve precise registry over long periods of time. Once the nanostructures are created, they must be placed in a vacuum environment if experiments lasting longer than a few hours are to be performed. For the experiments described here, the structure is created and within minutes transferred to a vacuum environment.</p><p>It is recommend before writing that a “writing test” be performed on all relevant electrodes. In such a test, two virtual electrodes are first created, and a single nanowire is written while simultaneously monitoring the conductance. A similar test of erasure can be performed by “cutting” the nanowire shortly afterwards. If the nanostructure is decaying rapidly, the issue is most likely due either to the interfacial contacts or the canvas itself. To distinguish between these two effects, a four-terminal measurement of the conductance should be performed, and the two-terminal conductance should be compared with the four-terminal conductance as a function of time. If the two-terminal conductance is decaying more rapidly than the four-terminal conductance, then the issue is related to the electrical contacts to the interface. If the four-terminal conductance is decaying at a comparable rate, then most likely the canvas is not suitable and should be replaced.</p><p>There are natural limitations of the current method for creating nanostructures. Specifically, the writing speed for the smallest devices is limited to a few hundred nanometers per second. Speeds far above that value lead to unpredictable results. Use of parallel writing techniques are possible<sup>27,28</sup>, but are not highly developed and have their own drawbacks. The size of nanostructures that can be created is naturally limited by the scan range of the AFM being used. A high-quality AFM with closed-loop feedback in the two scan directions is highly recommended. Tracking of point-like objects on the sample surface should be performed to monitor temporal drift of the sample.</p><p>Once creation of conductive nanostructures at oxide interfaces has been mastered, there are a wide range of experimental directions that can be explored. Using this technique, a wide variety of nanostructures and devices have already been demonstrated, including nanowires<sup>18</sup>, tunnel barriers<sup>29</sup>, rectifying junctions<sup>30</sup>, field-effect transistors<sup>18</sup>, single-electron transistors<sup>31</sup>, superconducting nanowires<sup>32</sup>, nanoscale optical detectors<sup>33</sup>, and nanoscale THz emitters and detectors<sup>34</sup>. </p></sec><sec><title>Disclosures</title><p>The authors have nothing to disclose.</p></sec>
Alpha- and beta-band oscillations subserve different processes in reactive control of limb movements	<p>The capacity to rapidly suppress a behavioral act in response to sudden instruction to stop is a key cognitive function. This function, called reactive control, is tested in experimental settings using the stop signal task, which requires subjects to generate a movement in response to a go signal or suppress it when a stop signal appears. The ability to inhibit this movement fluctuates over time: sometimes, subjects can stop their response, and at other times, they can not. To determine the neural basis of this fluctuation, we recorded local field potentials (LFPs) in the alpha (6–12 Hz) and beta (13–35 Hz) bands from the dorsal premotor cortex of two nonhuman primates that were performing the task. The ability to countermand a movement after a stop signal was predicted by the activity of both bands, each purportedly representing a distinct neural process. The beta band represents the level of movement preparation; higher beta power corresponds to a lower level of movement preparation, whereas the alpha band supports a proper phasic, reactive inhibitory response: movements are inhibited when alpha band power increases immediately after a stop signal. Our findings support the function of LFP bands in generating the signatures of various neural computations that are multiplexed in the brain.</p>	<contrib contrib-type="author"><name><surname>Pani</surname><given-names>Pierpaolo</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/99336"/></contrib><contrib contrib-type="author"><name><surname>Di Bello</surname><given-names>Fabio</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/191144"/></contrib><contrib contrib-type="author"><name><surname>Brunamonti</surname><given-names>Emiliano</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/1738"/></contrib><contrib contrib-type="author"><name><surname>D’Andrea</surname><given-names>Valeria</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/181175"/></contrib><contrib contrib-type="author"><name><surname>Papazachariadis</surname><given-names>Odysseas</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/73506"/></contrib><contrib contrib-type="author"><name><surname>Ferraina</surname><given-names>Stefano</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="author-notes" rid="fn001"><sup>*</sup></xref><uri xlink:type="simple" xlink:href="http://community.frontiersin.org/people/u/1534"/></contrib>	Frontiers in Behavioral Neuroscience	<sec sec-type="introduction" id="s1"><title>Introduction</title><p>Reactive control is the ability to rapidly suppress an imminent behavioral act in response to a sudden instruction to stop (Stuphorn and Emeric, <xref rid="B57" ref-type="bibr">2012</xref>). It is a significant function in cognitive control that is impaired in many psychiatric diseases and genetic syndromes (Aron, <xref rid="B1" ref-type="bibr">2011</xref>; Brunamonti et al., <xref rid="B14" ref-type="bibr">2011</xref>; Takkar et al., <xref rid="B59" ref-type="bibr">2011</xref>; Pani et al., <xref rid="B46" ref-type="bibr">2013</xref>).</p><p>Reactive control is measured in experimental settings using the stop signal, or countermanding, task (Logan and Cowan, <xref rid="B35" ref-type="bibr">1984</xref>). In most trials of this task, subjects must generate a movement in response to a go signal and inhibit the movement, as instructed by randomly presented stop signals. Typically, performance on the task fluctuates over time: sometimes, the subjects can stop their response and at other times fail, with the other task conditions being equal (Nelson et al., <xref rid="B44" ref-type="bibr">2010</xref>).</p><p>A potential factor that favors this modulation in performance is the finding that stop signals are presented in brief periods that are characterized by various tonic levels of movement preparation. Changes in movement preparation are detected in oscillating reaction times (RTs) to the go signal during the task. Lower RTs are associated with greater movement preparation (or readiness to respond) and vice versa. At the neural level, a wide network that comprises frontal cortical and subcortical (especially the basal ganglia and cerebellum) structures regulates the competition between movement preparation and movement suppression, which constitute the two sides of movement control (Chambers et al., <xref rid="B16" ref-type="bibr">2009</xref>; Stuphorn and Emeric, <xref rid="B57" ref-type="bibr">2012</xref>; Brunamonti et al., <xref rid="B13" ref-type="bibr">2014</xref>).</p><p>The computations of these structures can be examined by analyzing local field potentials (LFPs). Local field potentials reflect various subthreshold integrative processes, primarily synaptic inputs, that carry information about the state of the network and the local intracortical processing in the neural volume around the electrode tip (Mitzdorf, <xref rid="B41" ref-type="bibr">1985</xref>; Logothetis, <xref rid="B36" ref-type="bibr">2003</xref>; Kajikawa and Schroeder, <xref rid="B28" ref-type="bibr">2011</xref>; Lindèn et al., <xref rid="B33" ref-type="bibr">2011</xref>; Buzsáki et al., <xref rid="B15" ref-type="bibr">2012</xref>). Local field potentials comprise several band-limited components—theta (θ: 4–7 Hz), alpha (α: 8–12 Hz), and beta (β: 13–35 Hz) (Ray et al., <xref rid="B53" ref-type="bibr">2008</xref>)—that are associated with various functions and processing pathways. Thus, analyzing LFPs allows one to examine the presence of information channels that mediate the processing of neural information (Belitski et al., <xref rid="B7" ref-type="bibr">2008</xref>; Montemurro et al., <xref rid="B42" ref-type="bibr">2008</xref>; Kayser et al., <xref rid="B29" ref-type="bibr">2009</xref>) that can not be detected by recording spiking activity (Logothetis, <xref rid="B37" ref-type="bibr">2008</xref>).</p><p>Studies on the structures in the frontal-basal ganglia network that govern movement control support the diversity of information channels (Brittain et al., <xref rid="B10" ref-type="bibr">2014</xref>). For example, at the onset of movement, motor cortices and basal ganglia structures experience a decrease in beta activity (Pfurtscheller et al., <xref rid="B49" ref-type="bibr">2003</xref>; Zhang et al., <xref rid="B66" ref-type="bibr">2008</xref>; Dejean et al., <xref rid="B18" ref-type="bibr">2011</xref>; Jenkinson and Brown, <xref rid="B26" ref-type="bibr">2011</xref>), whereas during holding-static periods, beta power rises (Baker et al., <xref rid="B4" ref-type="bibr">1999</xref>, <xref rid="B5" ref-type="bibr">2001</xref>; Williams and Baker, <xref rid="B64" ref-type="bibr">2009</xref>). Alpha activity is related to the inhibition of the sensorimotor cortices (Pfurtscheller and Neuper, <xref rid="B50" ref-type="bibr">1994</xref>; Suffczynski et al., <xref rid="B58" ref-type="bibr">2001</xref>) and is considered a sign of top-down, cognitive inhibitory processing (Klimesch et al., <xref rid="B30" ref-type="bibr">2007</xref>; Jensen and Mahazeri, <xref rid="B27" ref-type="bibr">2010</xref>; Hwang et al., <xref rid="B24" ref-type="bibr">2014</xref>). These band components thus represent distinct neural processes that coordinate motor control during a task or some phase of the task.</p><p>In this study, we examined the dynamic of the alpha and beta band components of LFPs while recording from the dorsal premotor cortex (PMd) of a monkey, an area that has significant function in the frontal basal ganglia network in motor control (Mirabella et al., <xref rid="B40" ref-type="bibr">2011</xref>; Marcos et al., <xref rid="B38" ref-type="bibr">2013</xref>). We found that the ability to countermand a movement after a stop signal is predicted by the activity of both bands: the alpha band supports a proper phasic, reactive inhibitory response, and the beta band regulates the level of movement preparation.</p></sec><sec sec-type="materials\|methods" id="s2"><title>Materials and methods</title><p>Two adult male rhesus monkeys (Macaca mulatta; monkey S ~7.5 Kg and monkey L ~8 Kg) were examined. All experimental procedures, animal care, housing, and surgical procedures conformed with European (Directive 86/609/ECC and 2010/63/UE) and Italian (D.L. 116/92 and D.L. 26/2014) laws on the use of nonhuman primates in scientific research and were approved (no. 58/2005-B) by the Italian Ministry of Health.</p><sec id="s2-1"><title>Surgery, apparatus, and recording procedures</title><p>Under general anesthesia, a head-holding device, scleral eye coil (Robinson, <xref rid="B54" ref-type="bibr">1963</xref>), and recording cylinder were implanted. The recording cylinder (18 mm in diameter) was stereotactically positioned on the left frontal lobe, over the right arm representation of the PMd (Paxinos et al., <xref rid="B48" ref-type="bibr">2000</xref>). Recording positions were confirmed by structural MRI in monkey S and by visual inspection of anatomical landmarks after opening of the dura in monkey L. Details have been reported elsewhere (Mirabella et al., <xref rid="B40" ref-type="bibr">2011</xref>; see also supplementary Figure 3).</p><p>The experiments were performed in a dim, sound-attenuated room. The monkeys were seated upright in a chair with the head fixed; the arm that was contralateral to the recorded hemisphere was free, and the other arm was restrained in a comfortable position.</p><p>A 21-inch PC monitor (CRT noninterlaced, refresh rate 85 Hz, 800 × 600 resolution, 32-bit color depth; monitor-eye distance 21 cm) that was equipped with a touchscreen (MicroTouch, sampling rate 200 Hz) was placed in front of the monkey to present stimuli and monitor touch positions. Visual stimuli consisted of red circles (2.43 cd/m<sup>2</sup>) with a diameter of 7.6° (2.8 cm) on a dark background of uniform luminance (<0.01 cd/m<sup>2</sup>). The stimuli were synchronized with the monitor refresh rate. A noncommercial software package, CORTEX <xref ref-type="fn" rid="fn0001"><sup>1</sup></xref>, was used to regulate the stimuli and behavioral responses and collect neural (single unit activity 1 kHz) and eye movement (200 Hz) data.</p><p>Eye movements were monitored using a magnetic search coil technique (Remmel Labs, Ashland, MA, USA).</p><p>Neural activity was recorded extracellularly with a 7-channel multielectrode system (Thomas Recording, Giessen, Germany). The electrodes were quartz-insulated, platinum-tungsten fibers (80-µm diameter, 0.8 to 2.5-MΩ impedance) that were inserted transdurally, one at a time, with microdrives. After filtering and amplification steps (Thomas Recording, Giessen, Germany), a copy of the raw signal was sent to a dual-window spike discriminator (BAK Electronics, Mount Airy, MD, USA) for single-unit recording (online sorting). Single-unit results have been reported in a different format (Mirabella et al., <xref rid="B40" ref-type="bibr">2011</xref>; Marcos et al., <xref rid="B38" ref-type="bibr">2013</xref>). A second copy of the unfiltered raw signal was acquired with time stamps of the behavioral events for offline analysis (Tucker Davis Technologies, FL, USA; sampling rate 24.4 kHz). Modulation of high-frequency activity has been reported by Mattia et al. (<xref rid="B39" ref-type="bibr">2013</xref>).</p></sec><sec id="s2-2"><title>Behavioral task and analysis</title><p>Each trial of the reaching countermanding task (Figure <xref ref-type="fig" rid="F1">1A</xref>) began with the appearance of a circle at the center of the screen that the monkeys had to touch and hold for varying times (500–800 ms). Then, the circle disappeared (Go signal), and simultaneously, a circle appeared at the periphery at one of two opposite positions. In the <italic>no-stop trials</italic>, the monkey had to reach for the target within an allotted time (RT-bound: 600 ms for monkey S and 750 ms for monkey M). In 33% of the trials (<italic>stop trials</italic>), the central circle reappeared unpredictably (Stop signal) after varying delays (stop signal delay, SSD), instructing the monkey to keep its hand in the resting position for at least 450 ms (650–850 ms for monkey S, 450–550 ms for monkey L).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>(A)</bold> Countermanding reaching task: each trial began with the hand (represented by the gray annulus) on the central black circle. After a random time, the central circle disappeared, and a target appeared in 1 of 2 opposite positions (Go signal; only one position is represented). In no-stop trials, the hand had to leave the resting position and touch the peripheral circle. In stop trials, after a variable delay (SSD) after the Go signal, the central circle reappeared (Stop signal), requiring the monkey to cancel the planned movement. <bold>(B)</bold> Two example channels that were selected for further analysis on the basis of their modulation. Red line represents the average activity of correct stop trials, the blue line is the average activity of latency-matched no-stop trials, and the green line is the difference between them. Horizontal green lines represent mean ± 2.5 SD of the difference between no-stop and correct stop trials in the 300 ms preceding the go signal (see text for further details). <bold>(C)</bold> Time-frequency plots showing the contribution of the beta and alpha bands in stop correct trials (top row) and no-stop latency-matched trials (middle row) and the difference between them (bottom row) for each monkey (S on, stop signal onset; SSRT is the estimated latency of the Stop Signal Reaction Time).</p></caption><graphic xlink:href="fnbeh-08-00383-g0001"/></fig><p>A reward was delivered in the <italic>no-stop</italic> trials for initiating the reaching movement before the RT-bound was up and touching the peripheral circle; in <italic>correct stop</italic> trials, the reward was given for keeping the hand still in the initial position. No reward was delivered in <italic>wrong stop trials</italic>, when the monkey moved its hand despite a stop signal, even if the monkey change its mind and tried to return to the central circle.</p><p>To elicit an overall ability to inhibit of approximately 50%, we used either one of two techniques to adapt the SSDs to the monkeys’ behavior in the recording sessions: a fixed procedure, in which one of four delays was randomly presented in each stop trial, and a staircase procedure, in which the SSD increased (if the previous stop trial was correct) or decreased (if the previous stop trial was incorrect) by a fixed amount of time (step). In the fixed procedure, the SSDs were selected to effect the likelihood of inhibiting between 0.85 (for the shortest delay) and 0.15 (for the longest delay). In the staircase procedure, the step was 58.8 ms (five times the unit refresh rate).</p><p>We estimated the time it took each subject to cancel a movement—i.e., the stop signal reaction time (SSRT; Logan and Cowan, <xref rid="B35" ref-type="bibr">1984</xref>)—by subtracting the <italic>central</italic> SSD (corresponding to a probability of inhibiting of ~50%) from the mean RT (mean method: Logan and Cowan, <xref rid="B35" ref-type="bibr">1984</xref>; Hanes and Schall, <xref rid="B23" ref-type="bibr">1995</xref>). For data on the fixed procedure, we obtained a function of inhibition, defined as the probability of inhibiting as a function of the SSDs. The inhibition function was then fitted with a Weibull cumulative function (see Mirabella et al., <xref rid="B40" ref-type="bibr">2011</xref> for further details on the same behavioral dataset) to generate the <italic>central</italic> SSD value, corresponding to 50% probability of inhibition. For data on the staircase procedure, we calculated the <italic>central SSD</italic>, representative of the overall runs, using the midrun estimation method (Wetherill and Levitt, <xref rid="B63" ref-type="bibr">1965</xref>; Wetherill, <xref rid="B62" ref-type="bibr">1966</xref>; Levitt, <xref rid="B32" ref-type="bibr">1971</xref>). For the same dataset we calculated the SSRT with the integration method. For data obtained with the staircase procedure we considered only SSDs that were presented at least 14 times. Stop signal reaction time is calculated for each of the SSDs: RTs of no stop trials are rank ordered and the nth RT is found (nth = number of RTs × probability of having wrong stop trials at that SSD). The SSD is then subtracted from the nth RT, obtaining the SSRT (Logan and Cowan, <xref rid="B35" ref-type="bibr">1984</xref>). The SSRTs obtained for each SSD are then averaged to compute a single SSRT estimate.</p></sec><sec id="s2-3"><title>Data analysis</title><p>We included every recording session that respected the following criteria in the database: overall inhibitory performance of 0.4–0.6 in at least one of the two movement directions and higher mean no-stop signal RT compared with the wrong stop RT. The latter comparison corresponds to a test of the fundamental assumption to calculate the SSRT (see Hanes et al., <xref rid="B22" ref-type="bibr">1998</xref>; Mirabella et al., <xref rid="B40" ref-type="bibr">2011</xref> for further details). A lack of this assumption renders the data dispensable in calculating the SSRT (Logan and Cowan, <xref rid="B35" ref-type="bibr">1984</xref>). For each dataset, we then considered the stop trials that corresponded to the SSD that was closer to the 50% probability to inhibit for each movement direction that respected the criteria above.</p><p>Once the data were chosen, based on the behavioral criteria, we narrowed them down on the basis of the neural signal. We included only channels that were artifact- and noise-free in the voltage domain. Moreover, we selected channels, based on their modulation of the LFP (voltage) task—i.e., signals that differed in voltage between the 300 ms preceding the Go signal (control epoch) and the RT epoch (task-related activity). Finally, we focused on channels that potentially governed the cancellation of fast movements (reactive control), because they are the only channels that can determine whether and when a movement is generated. To regulate the cancellation of fast (stimulus-driven) movements, a neural signal must have different task-related activities when a movement is generated vs. when it is canceled, and the change in activity must occur before the end of the estimated cancellation process (SSRT).</p><p>To determine whether our selected channels were involved in movement cancellation, we adopted similar criteria and methods as in previous studies (Hanes et al., <xref rid="B22" ref-type="bibr">1998</xref>; Chen et al., <xref rid="B67" ref-type="bibr">2010</xref>; Scangos and Stuphorn, <xref rid="B55" ref-type="bibr">2010</xref>; Mirabella et al., <xref rid="B40" ref-type="bibr">2011</xref>). For selected recordings, we compared LFP activity in the voltage domain of correct stop trials with that of latency-matched no-stop trials—i.e., no-stop trials with RTs longer than or equal to the sum of the SSD and SSRT. We computed the difference between signals of the two types of trials in the 300-ms epoch before the go signal. We set the threshold as the average of this difference plus 2.5 SD. The difference in average voltage between the two types of trials after the SSD and before the SSRT had to surpass the threshold for us to consider that the signal was potentially involved in movement cancellation.</p><p>This comparison controls for the level of movement preparation: the latency-matched no-stop trials are trials in which the movement would have been canceled if the stop signal had occurred, thus reflecting the same level of motor preparation in the correct stop trials. The probability of inhibition of ~50% in the stop trials permitted us to also select no-stop trial with a latency that matched those of wrong-stop trials.</p><p>Ultimately, the behavioral data set comprised 22 recording sessions (monkey <italic>L</italic> = 5; monkey <italic>S</italic> = 17) that respected the criteria above. For each of the recording sessions, we were able to record signals that passed the selection criteria—from up to six electrodes (channels), for a total of 63 channels (monkey <italic>L</italic> = 10; monkey <italic>S</italic> = 53).</p><p>Time-frequency analysis was computed using the multi-taper algorithm with the freeware toolbox “Chronux”<xref ref-type="fn" rid="fn0002"><sup>2</sup></xref>. For each trial, the mean value and linear trend were removed from the raw signal, and spectrograms were generated in a window of 300 ms with 10-ms steps, using a frequency bandwidth of 5 Hz and 2 Slepian tapers. We set the maximum frequency to 150 Hz and obtained a 131 × 97 time-frequency array, with frequency and steps of 1.5 Hz and 10 ms, respectively. Relative spectrograms were defined as the ratio (in dB) between power spectrum in each time-frequency bin and mean power spectrum across all trials of baseline activity between –300 ms and 0 ms relative to target onset. Each relative spectrogram corresponds to the average across all trials under the same conditions.</p><p>For each time-frequency bin, we computed the mean relative power spectrum across trials and analyzed the difference between mean spectra for the conditions of each pair of trials (e.g., successful stop trials vs. latency-matched no-stop trials). We examined whether the differences were significant by separate permutation tests for each bin. This test was performed by shuffling the power values across the two groups of trials, computing the difference in means between the reshuffled groups and repeating the shuffling process N times (<italic>N</italic> = 5000). For each comparison, we obtained a color-coded <italic>p</italic>-value map (<italic>α</italic> = 0.05) of differences by repeating the permutation test for all bins of the arrays. The green color of the resulting time-frequency maps indicates <italic>p</italic> > 0.05; other colors indicate sign and intensity of significant differences.</p><p>The overall significance was corrected by multiple comparison using the false-discovery rate (FDR) method (Benjamini and Yekutieli, <xref rid="B8" ref-type="bibr">2001</xref>; Durka et al., <xref rid="B19" ref-type="bibr">2004</xref>), with an <italic>α</italic>-value of 0.05. Briefly, this “step-up” method was performed by: (1) ordering the <italic>p</italic>-values in an ascending series—<italic>p</italic><sub>(1)</sub> ≤ <italic>p</italic><sub>(2)</sub> ≤ … ≤ <italic>p</italic><sub>(<italic>k</italic>)</sub>; (2) finding the largest <italic>k</italic> for which <italic>p</italic><sub>(k) ≤</sub> <italic>αk/m</italic>; and (3) rejecting the null hypothesis for all bins with <italic>p</italic> ≤ <italic>p</italic><sub>(k)</sub>.</p><p>We focused our analysis on two frequency bands: the upper part of the theta (6–7 Hz) and alpha bands (8–12 Hz)—hereafter called alpha—and the beta band (β: 13–35 Hz).</p></sec></sec><sec sec-type="results" id="s3"><title>Results</title><sec id="s3-1"><title>Reactive control of movement</title><p>Monkey L reacted faster to the targets than monkey S in the no-stop trials (RTs in ms (mean ± std): 401.8 ± 62 and 555.6 ± 77, respectively) and wrong stop trials (379.3 ± 52 and 507 ± 50 respectively). There was a significant difference in RTs between no-stop trials and wrong stop trials in both monkeys (S: <italic>t</italic>-test (8327) = 15.02, <italic>p</italic> = 0.0001; <italic>p</italic> < 0.05); monkey (L: <italic>t</italic>-test (1518) = 3.5, <italic>p</italic> = 0.0005). We then considered a total of 84 SSDS across all the sessions and channels analyzed. For 73/84 (86%) of them the RTs in wrong stop trials were faster than those of no stop trials (non-parametric K-Smirnov test <italic>p</italic> < 0.05). These data confirmed the independence assumption of the race model (Logan and Cowan, <xref rid="B35" ref-type="bibr">1984</xref>) and permitted us to reliably estimate the SSRT, yielding a behavioral measure of the reactive control of movement; for both monkeys, the probability of inhibiting when given a stop signal approached 50% (mean ± std, monkey L 0.49 ± 0.4 and monkey S 0.48 ± 0.9). The estimated speed overall in canceling the movement (SSRT) was ~140 ms (L: mean method, mean ± std132 ± 11; integration method, mean ± std 135 ± 19; S: mean method mean ± std 147 ± 17; integration method, mean ± std 149 ± 15). No difference was detected between the two estimates (K-Smirnov <italic>p</italic> > 0.1 in all comparisons). For analysis purpose we considered for each session the average of the two estimates.</p><p>We considered the modulation in voltage before the end of the SSRT to select the channels to analyze as follows. Figure <xref ref-type="fig" rid="F1">1B</xref> shows two representative channels (monkey L left, monkey S right) that were analyzed, Figure <xref ref-type="fig" rid="F1">1C</xref> shows the time-frequency power contribution of the alpha and beta bands to the observed signal in the voltage range (see Figure legend for further details).</p></sec><sec id="s3-2"><title>LFP signature of movement generation</title><p>Reaction times vary between trials. To determine the LFP correlates of such variation, we compared the RTs between the first one-third of the selected trials (ordered by RT duration) and third one-third, corresponding to the 33% fastest and 33% slowest RTs, respectively. Reaction times were longer in slowest trials (median ± std in ms: 632 ± 56 vs. 486 ± 45 in monkey S, and 432 ± 51 vs. 346 ± 50 in monkey L, <italic>p</italic> < 0.00001 in both cases, K-Smirnov test).</p><p>Figure <xref ref-type="fig" rid="F2">2</xref> shows the time-frequency maps, relative to this analysis for each monkey separately). Approximately 100 ms after the go signal, beta activity declined in both monkeys (left columns for monkey L and S). Monkey L also showed an increase in alpha activity, locked to the go signal, whereas monkey S experienced a reduction in the same frequencies, followed by a slight increase.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Local field potential signature of movement generation</bold>. Grand average time-frequency plot showing the LFP activity across all populations of recordings for each monkey. Left column for each monkey: alignment to the Go signal (go); right column: alignment to the movement onset (mov on). Top row: grand average of slowest responses. Middle row: grand average of fastest responses. Bottom row: difference between slowest and fastest responses. Dotted lines represent mean ± SD of RT in the left column and of the Go signal in the right column.</p></caption><graphic xlink:href="fnbeh-08-00383-g0002"/></fig><p>Immediately prior to the start of the movement, the power across both frequencies declined. The difference between the two trial groups demonstrates that slower RTs are characterized in both monkeys by prolonged, higher beta activity following the go signal, and greater alpha activity.</p><p>When aligned to the onset of movement, the beta activity decreased immediately before the start of the movement and alpha activity rose in both monkeys (Figure <xref ref-type="fig" rid="F2">2</xref>, right columns for monkey L and S). The two conditions did not differ significantly in the 200 ms preceding the movement onset in both beta and alpha frequencies (bottom line).</p><p>Thus, both monkeys were characterized by a more rapid decline in beta activity in the fastest trials, whereas alpha activity had opposite patterns between monkeys after the go signal and increased immediately after movement onset in both animals. Average across monkeys is illustrated in Figure <xref ref-type="fig" rid="F3">3</xref>. Trial by trial correlation analysis (Pearson correlation coefficients (Pcc) calculated separately for each channel), between each band mean power in the first RT epochs (from 50 to 250 ms after go signal) and the RT, showed a stronger positive correlation between RTs and beta activity compared to alpha (median Pcc 0.23 for beta and −0.21 for alpha, <italic>p</italic> < 0.0001, K-Smirnov test). The stronger relationship between beta activity and RTs was confirmed by the number of channels significantly modulated: 30/63 for beta and 13/63 for alpha (<italic>Z</italic>-test = 3.2, <italic>p</italic> = 0.00142, see Supplementary Figure 1).</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Grand average time-frequency plot across all channels; Left column, alignment to the Go signal (go); right column: alignment to the movement onset (mov on)</bold>. Top row: grand average of slowest responses. Middle row: grand average of fastest responses. Bottom row: difference between slowest and fastest responses. Other conventions as in Figure <xref ref-type="fig" rid="F2">2</xref>.</p></caption><graphic xlink:href="fnbeh-08-00383-g0003"/></fig><p>The same analysis was performed on the last part of the epoch (from 200 before until RT). 16/63 channels showed a significant correlation between beta and RTs, while 13/62 between alpha and RT (<italic>Z</italic>-test = −0.64; <italic>p</italic> = 0.5; median Pcc 0.21 for alpha and 0.20 for beta, supplementary Figure 1, right column). No difference was observed between the two (K-Smirnov test, <italic>p</italic> = 0.99). Overall, in the first part of the trial, beta power showed a stronger correlation with RTs compared to alpha power, while just before the movement onset no difference was detected between the two frequencies.</p></sec><sec id="s3-3"><title>LFP signature of reactive cancellation</title><p>We examined LFP activity with regard to the reactive control of movement by selecting the correct stop trials with an SSD for which the probability of inhibiting was approximately 50% (minimum number of trials in wrong and correct stop = 8) and the corresponding latency-matched no-stop trials for each recording.</p><p>Figure <xref ref-type="fig" rid="F4">4</xref> shows the contrasts between correct stop trials and latency-matched no-stop trials for monkeys L, S and for all channels recorded. The same comparison, referring to a single channel, has been shown in Figure <xref ref-type="fig" rid="F1">1C</xref>.</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>Grand average of the time-frequency plots of correct stop trials (upper panels) and latency-matched no-stop trials (middle panels) and their difference (bottom panels)</bold>. Data are presented aligned to the go signal separately for each monkey (Monkey S and Monkey L), and across all channels. Other conventions as in Figures <xref ref-type="fig" rid="F1">1</xref>, <xref ref-type="fig" rid="F2">2</xref>.</p></caption><graphic xlink:href="fnbeh-08-00383-g0004"/></fig><p>In comparing the two time-frequency maps, before the stop signal appeared, the activity was similar between correct stop trials (top row) and latency-matched no-stop trials (middle row). After the stop signal in stop correct trials, alpha band activity increased in both monkeys. Concurrently, the beta band did not decrease, as in latency-matched no-stop trials. The rise in alpha power and lack of decline in beta power occurred before the end of the SSRT, meeting the requirements for a signal to be considered as being involved in movement control. Thus in both cases, the reactive suppression of a movement was characterized by a phasic increase in alpha and a sustained beta activity before the end of the SSRT (see also supplementary Figure 2).</p></sec><sec id="s3-4"><title>LFP signature of successful vs. unsuccessful inhibition</title><p>To predict the effectiveness of reactive control, the described pattern should differ in wrong stop trials—i.e., trials in which the stop process is potentially driven but deficient in interrupting generation of the movement. Thus, we compared successful vs. unsuccessful inhibition.</p><p>As per the race model (Logan and Cowan, <xref rid="B35" ref-type="bibr">1984</xref>), the failed inhibition in wrong stop trials is attributed to the level of motor readiness by the nervous system when the stop signal is presented: greater motor readiness effects a lower probability of suppressing the movement.</p><p>In comparing correct stop trials (Figure <xref ref-type="fig" rid="F5">5</xref>, top row) with wrong stop trials (middle row), the latter were characterized by low beta and low or delayed alpha activity when the stop signal appeared or immediately after. Beta and alpha activity was higher around the time when the stop signal was presented. This observation supports the hypothesis that these frequencies are linked to the result of the reactive control of movement.</p><fig id="F5" position="float"><label>Figure 5</label><caption><p><bold>Comparison between correct stop and wrong stop, aligned to stop signal presentation</bold>. Conventions are as in Figure <xref ref-type="fig" rid="F4">4</xref>.</p></caption><graphic xlink:href="fnbeh-08-00383-g0005"/></fig><p>In summary, the ability to interrupt a movement is characterized by two phenomena: a stimulus-driven stop process, reflected by an increase in alpha, and the level of motor readiness, represented by beta activity. In successful stop trials, beta activity remains high, and the stop signal drives the increase in alpha before the end of the SSRT, thus suppressing the reaching movement. In wrong stop trials, it appears that greater motor preparation is accompanied by ineffective instantiation of the stop process, represented by reduced and delayed alpha activity(see also supplementary Figure 2).</p><p>To evaluate the relationship between the two bands along the trial, we also performed a trial by trial correlation analysis (Pearson correlation coefficients calculated separately for each channel), between alpha and beta band power in the 150 ms preceding stop signal and from the stop signal to the end of the SSRT, in correct as well in wrong stop trials. We found that the two bands were overall weakly correlated. Before the stop signal presentation both in wrong and correct stop trials 17/63 channels showed a significant correlation between alpha and beta band. After stop signal presentation this number decreased to 13/63 in correct stop trials and to 11/63 in wrong stop trials. Thus the correlation between the two bands was observed only in a minority of channels (~20%), was not modulated by the phase of the task, and was not modulated by the type of trial (correct stop trials: <italic>Z</italic>-test = 0.8388, <italic>p</italic> = 0.40; wrong stop trials: <italic>Z</italic>-test = 1.29; <italic>p</italic> = 0.19; correct vs. wrong <italic>Z</italic>-test = 0.29, <italic>p</italic> = 0.76).</p></sec><sec id="s3-5"><title>Relationship between alpha and beta bands and the probability of success in stop trials</title><p>To further test the role of alpha and beta band activity in movement suppression, we asked whether alpha and beta band could vary depending on the probability of inhibition. We organized correct and wrong stop trials into three groups, depending on the difficulty or probability of success: easy trials, characterized by high probability of success (mean ± se: probability of success = 0.82 ± 12; SSDeasy = 262 ± 60); medium trials, were the stop trials with the SSD closer to the 0.5 probability of successful inhibition (probability of success 0.49 ± 0.11; SSDmedium = 354 ± 64); difficult trials, were the low probability of success trials, with SSDS longer than the previous one (mean ± se = 0.25 ± 0.14; SSDdifficult = 387 ± 8).</p><p>We then performed the analysis separately for alpha and beta band on 60/63 channels that provided enough trials (at least eight for each trial group), measuring the mean power recorded during the SSRT.</p><p>We found that correct stop trials showed a higher alpha activity compared to wrong stop trials for medium and difficult trials (power in dB (mean ± se): −0.1 ± 0.05 vs. −0.25 ± 0.06, <italic>p</italic> = 0.002, and −0.06 ± 0.04 −0.18 ± 0.06, <italic>p</italic> = 0.01 respectively, Figure <xref ref-type="fig" rid="F6">6</xref>), but no difference was observed for easy trials (power in dB (mean ± se): −3.23 ± 0.05 vs. −0.17 ± 0.07, <italic>p</italic> = 0.32). At the same time alpha activity of correct trials was lower in easy than in medium and difficult trials (<italic>p</italic> = 0.007 and <italic>p</italic> = 0.004 respectively), while no difference was observed between the last two (<italic>p</italic> = 0.31). In wrong stop trials no differences were detected between easy and medium or difficult trials (<italic>p</italic> = 0.2 and <italic>p</italic> = 0.8; repeated measures analysis with factors probability of success (easy, medium and difficult), and stop trial accuracy (correct, wrong): 2-way interaction decomposed: <italic>F</italic><sub>(2,118)</sub> = 8.07, <italic>p</italic> = 0.0005, Newman-Keuls <italic>post hoc</italic> test).</p><fig id="F6" position="float"><label>Figure 6</label><caption><p><bold>Comparison between correct (black) and wrong (red) stop trials characterized by high (easy), medium or low (difficult) probability to inhibit the movement</bold>. The comparison is done separately for alpha and beta band.</p></caption><graphic xlink:href="fnbeh-08-00383-g0006"/></fig><p>Beta activity showed a slightly different patterns: overall beta activity was higher in correct compared to wrong stop trials (main effect: <italic>F</italic><sub>(1,59)</sub> = 13,141, <italic>p</italic> = 0.0006; power in dB (mean ± se): −0.58 ± 0.06 vs. 0.72 ± 0.1, Figure <xref ref-type="fig" rid="F6">6</xref>); at the same time beta activity was higher in easy trials compared to medium and difficult trials, (power in dB (mean ± se): −0.61 ± 0.07 vs. −0.67 ± 0.06 and −0.68 ± 0.06 respectively (<italic>F</italic><sub>(2,118)</sub> = 3.7, <italic>p</italic> = 0.028), main effect of probability of success).</p><p>Overall these data shows that alpha band was higher in correct trials compared to wrong stop trials specifically in medium and difficult trials, and that was lower in easy correct trials. Beta activity was overall higher in correct stop trials than in wrong stop trials, and specifically higher in easy stop trials compared to medium and difficult trials.</p></sec><sec id="s3-6"><title>Comparison between wrong stop and latency matched no-stop trials</title><p>To detect subtle effect of the stop signal presentation in wrong stop trials, we compared wrong stop trials with latency matched no-stop trials, that is trials too fast to be inhibited had the stop signal been presented.</p><p>In comparing the two time-frequency maps after the go signal (Figure <xref ref-type="fig" rid="F7">7</xref>, left columns for monkey S and L; Figure <xref ref-type="fig" rid="F7">7</xref> left column), the activity was similar between wrong stop trials (top row) and latency-matched no-stop trials (middle row). Toward movement onset a strong increase in alpha activity is observed for both trials type. This activity is clear when the maps are aligned to the movement onset (Figure <xref ref-type="fig" rid="F7">7</xref>, right columns for monkey S and L; Figure <xref ref-type="fig" rid="F8">8</xref> right column). Importantly this activity rises first in wrong stop trials than in no-stop trials, and it is accompanied by a higher sustained beta activity.</p><fig id="F7" position="float"><label>Figure 7</label><caption><p><bold>Grand average time-frequency plot showing the LFP activity of wrong stop trials (upper panels) and latency-matched no-stop trials (middle panels) and their difference (bottom panels)</bold>. Data are presented separately for each monkey. In left columns data are aligned to the go signal; in right colums are aligned to movement onset. Other conventions as in Figure <xref ref-type="fig" rid="F2">2</xref>.</p></caption><graphic xlink:href="fnbeh-08-00383-g0007"/></fig><fig id="F8" position="float"><label>Figure 8</label><caption><p><bold>Grand average time-frequency plot across all channels</bold>. Other conventions as in Figure <xref ref-type="fig" rid="F6">6</xref>.</p></caption><graphic xlink:href="fnbeh-08-00383-g0008"/></fig></sec><sec id="s3-7"><title>Tonic aspects of movement preparation affect the reactive control of movement</title><p>Changes in inhibitory performance are observed during the stop task. Further, correct stop trials occur at intervals of lower movement preparation in no-stop trials that immediately precede and follow them (Nelson et al., <xref rid="B44" ref-type="bibr">2010</xref>), compared with no-stop trials before and after wrong stop trials. Thus, these tonic changes on the local temporal scale (the trials around the stop trial) can affect the conflict between movement generation and suppression.</p><p>To test this hypothesis, we analyzed the behavioral data. In both monkeys, the RTs in no-stop trials that preceded correct stop trials were slower compared with no-stop trials that preceded wrong stop trials (L: <italic>p</italic> = 0.002; S: <italic>p</italic> = 0.006). The same result was observed for no-stop trials that followed correct stop vs. wrong stop trials [L: 0.009; S: <italic>p</italic> = 0.002; factorial ANOVA with monkey (L and S), trial type (no-stop, stop correct, stop wrong), and time (no-stop trials immediately preceding or immediately following the trial type) as factors; 3-way interaction decomposed: <italic>F</italic><sub>(2,9698)</sub> = 5,6240, <italic>p</italic> = 0.004, Newman-Keuls <italic>post hoc</italic> test].</p><p>We then pooled the no-stop trials that surrounded the correct stop and compared them with no-stop trials that surrounded the wrong stop trials (mean ± sem: L: 419.9 ± 3.7 and 401.7 ± 3.7, respectively, <italic>p</italic> = 0.0004; S: 572.4 ± 2.2 vs. 552.95 ± 2.12, <italic>p</italic> = 0.00002, Newman-Keuls <italic>post hoc</italic> test).</p><p>Thus, the behavioral data demonstrate that the level of movement preparation, represented by RTs, in the immediate context (i.e., preceding and following trials) contributes to determining behavioral performance. To determine the neural correlate of these fluctuations, we compared the no-stop trials that surrounded correct stop trials with those that surrounded wrong stop trials in each monkey. The decrease in beta activity characterized trials that surrounded wrong stop trials, assuming a pattern (Figures <xref ref-type="fig" rid="F9">9</xref>, <xref ref-type="fig" rid="F10">10</xref>) that was similar to that of the comparison between the slowest one-third vs. fastest one-third no-stop trials. These data support the function of beta activity in determining the readiness to respond, even on a local time scale.</p><fig id="F9" position="float"><label>Figure 9</label><caption><p><bold>Effects of tonic level of movement preparation: comparison between no-stop trials that surrounded correct stop trials and no-stop trials that surrounded wrong stop trials</bold>. Conventions are as in Figure <xref ref-type="fig" rid="F2">2</xref>. Data are shown separately for each monkey.</p></caption><graphic xlink:href="fnbeh-08-00383-g0009"/></fig><fig id="F10" position="float"><label>Figure 10</label><caption><p><bold>Grand average time-frequency plot across all channels</bold>. Other conventions as in Figure <xref ref-type="fig" rid="F9">9</xref>.</p></caption><graphic xlink:href="fnbeh-08-00383-g0010"/></fig></sec></sec><sec sec-type="discussion" id="s4"><title>Discussion</title><p>We found that the ability to reactively control an imminent movement can be predicted by the level of beta- and alpha-band activities in the PMd. Specifically, the beta band appears to represent the activity of a sustained brake that affects a tonic level of motor preparation, whereas in this context, the alpha band reflects an inhibitory signal that characterizes the suppression of the movement, mostly with a phasic dynamic. Thus, the inhibition of a movement is related to these two distinct neural computations: one that represents a level of motor preparation, codified in the beta activity during the task, and the alpha activity that represents the phasic inhibitory signal that could suppress the movement.</p><p>The presence of alpha and beta bands modulations in the cortical and subcortical structures of the limb motor system has been established. Overall, our data and interpretations are consistent with recordings from motor cortices and with models on the functions of the various frequency components of the LFP (Gilbertson et al., <xref rid="B21" ref-type="bibr">2005</xref>; Baker, <xref rid="B3" ref-type="bibr">2007</xref>).</p><p>The inverse relationship between beta activity and the readiness to respond has been observed in humans and monkeys. One interpretation is that beta activity represents, at least in motor areas, the maintenance of a postural state and movement prevention (in our case, the hand maintained on the central target). During the holding period, strong beta activity has been noted in motor cortices (Baker et al., <xref rid="B4" ref-type="bibr">1999</xref>, <xref rid="B5" ref-type="bibr">2001</xref>; Williams and Baker, <xref rid="B64" ref-type="bibr">2009</xref>). Also, an anticorrelation between beta activity and RTs has been reported in sensorimotor cortices and the supplementary motor area (SMA) in monkeys (Zhang et al., <xref rid="B66" ref-type="bibr">2008</xref>; Chen et al., <xref rid="B67" ref-type="bibr">2010</xref>). Gilbertson et al. (<xref rid="B21" ref-type="bibr">2005</xref>) found that movement acceleration was reduced when the cue that triggered the response was presented during times of elevated beta activity. Again, entraining cortical activity at 20 Hz in healthy subjects slows voluntary movement (Pogosyan et al., <xref rid="B51" ref-type="bibr">2009</xref>). These observations and our data here support that beta activity participates to the stopping result by regulating the level of motor preparation.</p><p>The increase in beta activity that we observed in slower trials is attributed primarily to the input from the cortical and subcortical regions (Belitski et al., <xref rid="B7" ref-type="bibr">2008</xref>) that affect the neural processes that promote movement generation through the cortical-basal ganglia-thalamocortical loop (Brown and Williams, <xref rid="B12" ref-type="bibr">2005</xref>; Aron, <xref rid="B1" ref-type="bibr">2011</xref>). These loops might also modulate alpha activity during the reactive control of movement. Increased alpha activity is linked to inhibition of the sensorimotor cortices (Pfurtscheller and Neuper, <xref rid="B50" ref-type="bibr">1994</xref>; Suffczynski et al., <xref rid="B58" ref-type="bibr">2001</xref>). A recent hypothesis suggests that this rhythm reflects top-down, cognitive inhibitory processing that supports the suppression of task-irrelevant processes and the competition between processes within the motor system (Klimesch et al., <xref rid="B30" ref-type="bibr">2007</xref>; Jensen and Mahazeri, <xref rid="B27" ref-type="bibr">2010</xref>). In our case, the increase in alpha activity was observed specifically after a stop signal and before the end of the SSRT.</p><p>The PMd participates in the frontal-basal ganglia-thalamofrontal network that governs movement control. Other frontal areas, such as the pre-SMA and SMA, have been implicated in forms of tonic (proactive) control, by modulating the level of responsiveness, and in reactive control, primarily by mediating the inhibition of responses. The signatures of proactive control in the pre-SMA and SMA are similar to the PMd—i.e., increased beta activity is associated with lower responsiveness. It is possible that the increase in beta is a general signal that affects many areas of the brain that participate in movement control, that probably originates in the basal ganglia (possibly striatum, Courtemanche et al., <xref rid="B17" ref-type="bibr">2003</xref>; Zandbelt and Vink, <xref rid="B65" ref-type="bibr">2010</xref>) and prefrontal cortex (Brown, <xref rid="B11" ref-type="bibr">2007</xref>). In fact, robust synchronization has been observed between the beta rhythm in the subthalamic nucleus and motor structures (Kühn et al., <xref rid="B31" ref-type="bibr">2009</xref>).</p><p>The involvement of the pre-SMA and SMA vs. PMd appears to differ in reactive control. Whereas single-cell recordings suggest that the PMd decides whether to generate a movement (Mirabella et al., <xref rid="B40" ref-type="bibr">2011</xref>), the pre-SMA and SMA are more supportive of a disparate type of inhibitory signal (Chen et al., <xref rid="B67" ref-type="bibr">2010</xref>; Scangos and Stuphorn, <xref rid="B55" ref-type="bibr">2010</xref>). Specifically, neural activity in the pre-SMA might be a signal that is related to the motivation or tendency to inhibit. The reactive control can thus be triggered by lateral prefrontal regions (the right IFC in humans, Aron et al., <xref rid="B2" ref-type="bibr">2014</xref>) that act through basal-ganglia structures (specifically, the subthalamic nucleus) on motor cortical structures, such as the PMd.</p><p>Thus, in the reactive control of movement, as evaluated by the countermanding task, a tonic level of motor readiness and a phasic inhibitory signal cohabit in the PMd.</p><p>We believe that the beta band constitutes the signature of a tonic brake that regulates the speed of movement preparation. Single-unit, as well as multiunit, activities in the PMd reflect the level of movement preparation and whether a movement will be performed (Mirabella et al., <xref rid="B40" ref-type="bibr">2011</xref>; Mattia et al., <xref rid="B39" ref-type="bibr">2013</xref>). Also, neural variability in this area is affected by the history of recent trials (Marcos et al., <xref rid="B38" ref-type="bibr">2013</xref>). The beta and alpha bands can represent the correlates of two distinct synaptic computations that, ultimately, will act on the same neuronal populations to regulate behavior. The beta band seems to be related with the fine regulation of movement preparation, as confirmed also by the stronger correlation with RTs; while the alpha band with the sudden stop of the movement. Beta and alpha band correlates to each other only in about 20% of the analyzed channels, thus supporting the idea that in most cases they are independent.</p><p>A further support to the independence of the two computations comes from the modulation of their activity in easy, medium and difficult trials; alpha power is specifically higher in medium and difficult correct stop trials, while beta power is overall higher in correct stop trials, but specifically in the easy trials.</p><p>The possibility that neural computations subtending alpha and beta activity can act on the same neural populations is supported by the finding that neurons in the saccadic system are affected both by tonic (proactive) and reactive aspects of movement control (Pouget et al., <xref rid="B52" ref-type="bibr">2011</xref>).</p><p>The stop signal task performance is normally analyzed in the frame of the race model. This model assumes that the go process, triggered by the go stimulus, and the stop process, triggered by the stop signal, race in parallel against each other. In stop trials the suppression of the movement will occur if the stop process wins the race. Attempts to find a neural instantiation of the go and stop process have identified go and stop units with movement and fixation neurons in the saccadic system (Boucher et al., <xref rid="B9" ref-type="bibr">2007</xref>; Lo et al., <xref rid="B34" ref-type="bibr">2009</xref>), and population of neurons in PMd (Mirabella et al., <xref rid="B40" ref-type="bibr">2011</xref>). The modulations we observed could correlate with the activity of stop and go units, but not being identified with them. Specifically beta activity, that is anti-correlated with RTs, shows a temporal dynamic similar to that of a model of tonic inhibition that act on saccadic neurons and prevents from moving already at the presentation of the go signal (Lo et al., <xref rid="B34" ref-type="bibr">2009</xref>). Alpha activity increases after stop signal presentation and before the end of SSRT in correct stop trials. This dynamic is similar to what observed in fixation neurons in saccadic centers and in some neurons in PMd (Hanes et al., <xref rid="B22" ref-type="bibr">1998</xref>; Paré and Hanes, <xref rid="B68" ref-type="bibr">2003</xref>; Mirabella et al., <xref rid="B40" ref-type="bibr">2011</xref>), that probably instantiates the stopping process able to act (directly or indirectly) on movement neurons to suppress their activity. Alpha activity can thus be a correlate of the fast stopping process instantiated once the stop signal is presented.</p><p>These diverse neural rhythms likely reflect disparate functional pathways—one that acts when a sudden, fast change is required and the other that finely regulates the movement.</p><p>It is possible that these activities are sustained by partially overlapping structures, meaning that the same structures participate in different aspects of motor control through disparate brain rhythms. Similar machinery could allow different neural computations to be subserved by various frequencies, thus permitting parallel dispatches of information. This multiplexing might enhance the diversity of information that is sent to the neurons that mediate the final computation of the decision, allowing them to be regulated using more complex mechanisms.</p><p>Changes in LFP power detect episodic changes in the synchrony of certain frequency bands with regard to behavior. These oscillations might have small effects on behavior or none at all—a hypothesis that is related to the general idea that the neural computations that determine the behavior reside at the firing rate level (output of the neural population), whereas the oscillations can not affect them. Alternatively, it is generally accepted that modulations in LFP power reflect changes primarily at the synaptic level that contribute to alterations in the firing rate of neurons (Buzsáki et al., <xref rid="B15" ref-type="bibr">2012</xref>).</p><p>Many studies support a causal role of oscillations on behavior. With regard to beta activity, for example, the stimulation of the motor cortex in healthy humans affects force production and the speed of responses (Pogosyan et al., <xref rid="B51" ref-type="bibr">2009</xref>; Wach et al., <xref rid="B61" ref-type="bibr">2013</xref>), and stimulation of the subthalamic nucleus in patients also decreases the speed of a response (Jahanshahi, <xref rid="B25" ref-type="bibr">2013</xref>). There is a similar effect for alpha frequencies (Timmermann et al., <xref rid="B60" ref-type="bibr">2004</xref>). Recently, a causal role of brain rhythms in regulating behavior has recently been demonstrated for higher (gamma) activities, as well. Engelhard et al. (<xref rid="B20" ref-type="bibr">2013</xref>) used a brain machine interface approach to train monkeys to move a cursor on a screen by modulating the gamma power recorded in motor cortex sites, thus showing that the volitional control of these frequencies generate sufficient information to perform a task.</p><p>We believe that the subsecond scale in which we observed the modulation constitutes sufficient time to affect behavior. Modulation in these frequencies can affect the activity of single neurons (Murthy and Fetz, <xref rid="B43" ref-type="bibr">1992</xref>; Spinks et al., <xref rid="B56" ref-type="bibr">2008</xref>), likely by modulating the firing activity or synchronized activity between neurons. At the level of the PMd, we have shown that weak inputs prime a chain reaction between neural modules (populations) from more excitable modules to others, promoting the development of a motor plan in approximately 120 ms (Mattia et al., <xref rid="B39" ref-type="bibr">2013</xref>). Anyway this is hypothetical: a causal relationship between LFPs and behavior cannot be claimed in this study.</p><p>In our data, one monkey showed an increase in alpha activity after the appearance of a peripheral target, and alpha activity also rose after the reaching movement started. Can these observations do not support our interpretation of the alpha band as a correlate of a suppressive computation? We think this is not the case for some reasons.</p><p>The alpha modulation we observed and interpreted as a sign of suppression, occurs specifically during the SSRT and it is higher in stop correct compared to both no-stop and wrong stop trials. Also it occurs first in correct than in wrong stop trials, and it increases first in wrong stop trials compared to latency matched no-stop trials, purportedly representing an attempt of late suppression triggered by the stop signal presentation. The alpha modulation observed after the movement started can be interpreted in terms of a suppressive signal too. In fact at this moment of the task the monkey is reaching the peripheral target and will have to stop the hand on it to get the reward. Neurons in PMd are known to participate both in suppression and to the online updates of reaching movements (Battaglia-Mayer et al., <xref rid="B6" ref-type="bibr">2014</xref>), and similar mechanism can be involved in inhibiting a being prepared as well as a being performed movement. We think that alpha activity we observed in monkey L is a response to the onset of the go signal instructing the reaching, a process that, is the same in both stop correct and no-stop trials. This phenomena, for frequency <10 Hz, has been already documented in both motor and premotor cortices (O’Leary and Hatsopoulos, <xref rid="B45" ref-type="bibr">2006</xref>).</p><p>More generally, with regard to the issue of combining alpha activity as an inhibitory signal with these data, we believe that alterations in alpha activity can not be considered univocally a sign of motor or cognitive suppression, nor can other measures of the nervous system, such as firing rate, be considered specific cognitive functions. In fact the interpretation of the firing rate of a neuron will depend on the function that is subserved by the same neuron in a controlled context.</p><p>These frequency bands constitute a basic computation that can be used in many cognitive contexts. The function (in terms of behavior) must be established in specific experimental settings. It is possible, for example, that beta activity in this context reflects a basal-frontal system that regulates behavior finely, whereas alpha activity is a widespread braking mechanism that is implemented to halt a movement after the stop signal.</p><p>It is possible that these bands can be used to communicate between structures, thus forming the electrophysiological hardware, the content of which will depend on the structures that are active. It has been proposed that the cortex uses a multiplexing strategy to encode different types of information simultaneously on various time scales (Panzeri et al., <xref rid="B47" ref-type="bibr">2010</xref>), thus increasing the capacity for information.</p></sec><sec id="s6"><title>Conflict of interest statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec>
CETP gene polymorphisms and risk of coronary atherosclerosis in a Chinese population	<sec><title>Background</title><p>Coronary atherosclerosis, the most common form of coronary artery disease (CAD), is characterized by accumulation of lipid in the walls of coronary arteries. Recent data from clinical trials have showed that high-density lipoprotein cholesterol (HDL-C) has causal role in the pathogenesis and development of coronary atherosclerosis. Cholesteryl ester transfer protein (CETP) is an important regulator of plasma HDL-C. Several genetic mutations in the CETP gene were found to be associated with HDL-C levels. The aim of the present study is to evaluate the association of HDL-C-related CETP polymorphisms and risk of coronary atherosclerosis.</p></sec><sec><title>Methods</title><p>We investigated the association of seven single nucleotide polymorphisms (SNP) (rs1800775, rs708272, rs5882, rs1532624, rs1864163, rs7499892, and rs9989419) in the CETP gene with the risk of coronary atherosclerosis and levels of HDL-C in a case–control study in China. Included in the study were 420 patients with coronary atherosclerosis and 424 healthy controls. SNP genotyping was performed by TaqMan allelic discrimination assay and serum lipid levels were measured by standard laboratory methods.</p></sec><sec><title>Results</title><p>Carriers of the AA and GA + AA genotypes of rs708272 had significant lower risks of coronary atherosclerosis (OR = 0.55, 95% CI: 0.36-0.85, p = 0.003; OR = 0.67, 95% CI: 0.50-0.90, p = 0.007, respectively) compared to those with GG genotype. These relations remained significant after adjustment for confounding effects of age, smoking, diabetes and hypertension. The rs1800775 polymorphism was significantly associated with serum levels of HDL-C in healthy controls (p = 0.04). Besides, rs708272 was in close linkage disequilibrium (LD) with rs1800775 in this study.</p></sec><sec><title>Conclusions</title><p>Our findings indicated that CETP rs708272 may be associated with the risk of coronary atherosclerosis and rs1800775 may influence serum HDL-C levels in healthy controls in Chinese.</p></sec>	<contrib contrib-type="author" equal-contrib="yes" id="A1"><name><surname>Wang</surname><given-names>Jun</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>zero991127@126.com</email></contrib><contrib contrib-type="author" equal-contrib="yes" id="A2"><name><surname>Wang</surname><given-names>Li Jun</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>wanglijun@medmail.com.cn</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Zhong</surname><given-names>Yong</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>zhongyongnj@163.com</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Gu</surname><given-names>Ping</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>ping19762000@yahoo.com.cn</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Shao</surname><given-names>Jia Qing</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>shaojiaq@hotmail.com</email></contrib><contrib contrib-type="author" corresp="yes" id="A6"><name><surname>Jiang</surname><given-names>Shi Sen</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>zero991127@hotmail.com</email></contrib><contrib contrib-type="author" corresp="yes" id="A7"><name><surname>Gong</surname><given-names>Jian Bin</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>agong62@126.com</email></contrib>	Lipids in Health and Disease	<sec><title>Background</title><p>Coronary atherosclerosis, a chronic inflammatory disease characterized by the accumulation of fatty materials such as cholesterol and triglyceride on the walls of the coronary arteries, is the principal cause of coronary artery disease (CAD) [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>]. HDL is believed to be a protective factor against CAD, and the inverse relationship between plasma HDL-C and the incidence of CAD is well established [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B4">4</xref>]. Preliminary studies have suggested that HDL infusions can induce atherosclerosis regression [<xref ref-type="bibr" rid="B5">5</xref>]. Protective effect of HDL on atherosclerosis may due to its role in preventing oxidation or other adverse effects of low-density lipoprotein cholesterol (LDL-C) on endothelial cell, moreover, HDL also can directly stimulate endothelial cell to produce nitric oxide, beneficial anti-inflammatory, anti-apoptotic and anti-thrombotic agents as well as promote endothelial repair processes [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref>].</p><p>Cholesteryl ester transfer protein (CETP) is a hydrophobic glycoprotein, which has an established role in transporting of cholesterol from the peripheral tissues to the liver for elimination through exchanging triglycerides of VLDL and LDL against cholesteryl esters of HDL. The possibility that increased function of CETP might be proatherogenic and that inhibition of its activity might be antiatherogenic was first raised >20 years ago [<xref ref-type="bibr" rid="B8">8</xref>]. CETP inhibitors as novel drugs have been developed to raise HDL-C concentrations and improve HDL function in patients with coronary disease, although the effect and safety still need to be confirmed [<xref ref-type="bibr" rid="B9">9</xref>].</p><p>Several mutations in the CETP gene have been identified as a cause of CETP deficiency and change of HDL-C levels, but the associations of these single nucleotide polymorphisms (SNP) and susceptibility to atherosclerosis still lacks consistency [<xref ref-type="bibr" rid="B10">10</xref>-<xref ref-type="bibr" rid="B12">12</xref>]. Besides, the relation between these SNPs and risk of coronary atherosclerosis has not been fully studied in Chinese population.</p><p>To help clarify whether the CETP SNPs which were previously shown to be associated with plasma HDL-C levels and also confirmed in a genome-wide association study [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B13">13</xref>-<xref ref-type="bibr" rid="B17">17</xref>] are associated with susceptibility of coronary atherosclerosis and plasma HDL-C levels, we examined seven SNPs in the CETP gene (rs1800775, rs708272, rs5882, rs1532624, rs1864163, rs7499892, and rs9989419) in a case–control study in Chinese population.</p></sec><sec sec-type="results"><title>Results</title><p>Our study population consisted of 420 cases and 424 healthy controls. Characteristics of the study subjects are shown in Table <xref ref-type="table" rid="T1">1</xref>. Cases and controls were comparable with respect to age and gender. Cases were more probably to smoke cigarettes (50.9% vs. 32.3%), have diabetes (21.0% vs. 12.0%) and hypertension (48.7% vs. 38.7%). Besides, cases have significant lower levels of serum HDL-C and higher levels of serum total cholesterol (TC) and LDL-C than that in controls.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Selected characteristics of cases and controls</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Characteristics</bold></th><th align="left"><bold>Cases (n = 420)</bold></th><th align="left"><bold>Controls (n = 424)</bold></th><th align="left"><bold>P value</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Age (year)<hr/></td><td align="left" valign="bottom">66(61–70)<hr/></td><td align="left" valign="bottom">66(60–70)<hr/></td><td align="left" valign="bottom">0.97<hr/></td></tr><tr><td align="left" valign="bottom">Sex (Male/female)<hr/></td><td align="left" valign="bottom">167/253<hr/></td><td align="left" valign="bottom">168/256<hr/></td><td align="left" valign="bottom">0.52<hr/></td></tr><tr><td align="left" valign="bottom">Smoking (Yes/no)<hr/></td><td align="left" valign="bottom">214/206<hr/></td><td align="left" valign="bottom">137/287<hr/></td><td align="left" valign="bottom"><0.001<hr/></td></tr><tr><td align="left" valign="bottom">Diabetes (Yes/no)<hr/></td><td align="left" valign="bottom">88/331<hr/></td><td align="left" valign="bottom">51/372<hr/></td><td align="left" valign="bottom">0.001<hr/></td></tr><tr><td align="left" valign="bottom">Hypertension (Yes/no)<hr/></td><td align="left" valign="bottom">205/215<hr/></td><td align="left" valign="bottom">164/260<hr/></td><td align="left" valign="bottom">0.003<hr/></td></tr><tr><td align="left" valign="bottom">BMI (kg/m<sup>2</sup>)<hr/></td><td align="left" valign="bottom">24.3(23.0-25.9)<hr/></td><td align="left" valign="bottom">24.2(22.9-26.6)<hr/></td><td align="left" valign="bottom">0.86<hr/></td></tr><tr><td align="left" valign="bottom">Total cholesterol (mmol/L)<hr/></td><td align="left" valign="bottom">4.72(4.11-5.48)<hr/></td><td align="left" valign="bottom">4.09(3.44-5.22)<hr/></td><td align="left" valign="bottom"><0.001<hr/></td></tr><tr><td align="left" valign="bottom">HDL-C (mmol/L)<hr/></td><td align="left" valign="bottom">1.20(1.09-1.51)<hr/></td><td align="left" valign="bottom">1.23(1.01-1.67)<hr/></td><td align="left" valign="bottom"><0.001<hr/></td></tr><tr><td align="left">LDL-C (mmol/L)</td><td align="left">2.83(2.47-3.29)</td><td align="left">2.45(2.06-3.13)</td><td align="left"><0.001</td></tr></tbody></table></table-wrap><p>The associations of CETP variants with risk of coronary atherosclerosis are presented in Table <xref ref-type="table" rid="T2">2</xref>. The genotype distributions of these seven variants showed no deviation from the expected Hardy-Weinberg equilibrium among controls (p > 0.05). Of these SNPs, carriers of the AA and GA + AA genotypes of rs708272 had significant lower risk of coronary atherosclerosis (OR = 0.55, 95% CI: 0.36-0.85, p = 0.003; OR =0.67, 95% CI: 0.50-0.90, p = 0.007, respectively) compared with carriers of the major genotype. These associations remained statistically significant after further adjustment for age, smoking, hypertension and diabetes. None of the other SNPs examined was associated with coronary atherosclerosis.</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Association of genetic variants in CETP gene with risk of coronary atherosclerosis</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>SNP</bold></th><th align="left"><bold>Controls, n</bold></th><th align="left"><bold>Cases, n</bold></th><th align="left"><bold>OR (95% CI)</bold></th><th align="left"><bold>P for trend</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">rs1800775<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">AA<hr/></td><td align="left" valign="bottom">116<hr/></td><td align="left" valign="bottom">102<hr/></td><td align="left" valign="bottom">1.00<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">AC<hr/></td><td align="left" valign="bottom">222<hr/></td><td align="left" valign="bottom">216<hr/></td><td align="left" valign="bottom">1.11(0.79-1.56)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">CC<hr/></td><td align="left" valign="bottom">83<hr/></td><td align="left" valign="bottom">100<hr/></td><td align="left" valign="bottom">1.43(0.93-2.19)<hr/></td><td align="left" valign="bottom">0.32<hr/></td></tr><tr><td align="left" valign="bottom">AC + CC<hr/></td><td align="left" valign="bottom">305<hr/></td><td align="left" valign="bottom">316<hr/></td><td align="left" valign="bottom">1.19(0.86-1.65)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">rs1532624<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">CC<hr/></td><td align="left" valign="bottom">199<hr/></td><td align="left" valign="bottom">209<hr/></td><td align="left" valign="bottom">1.00<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">CA<hr/></td><td align="left" valign="bottom">191<hr/></td><td align="left" valign="bottom">183<hr/></td><td align="left" valign="bottom">0.88(0.66-1.18)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">AA<hr/></td><td align="left" valign="bottom">34<hr/></td><td align="left" valign="bottom">29<hr/></td><td align="left" valign="bottom">0.80(0.45-1.41)<hr/></td><td align="left" valign="bottom">0.52<hr/></td></tr><tr><td align="left" valign="bottom">CA + AA<hr/></td><td align="left" valign="bottom">225<hr/></td><td align="left" valign="bottom">212<hr/></td><td align="left" valign="bottom">0.87(0.65-1.16)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">rs708272<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">GG<hr/></td><td align="left" valign="bottom">139<hr/></td><td align="left" valign="bottom">176<hr/></td><td align="left" valign="bottom">1.00<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">GA<hr/></td><td align="left" valign="bottom">207<hr/></td><td align="left" valign="bottom">192<hr/></td><td align="left" valign="bottom">0.71(0.52-0.97)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">AA<hr/></td><td align="left" valign="bottom">74<hr/></td><td align="left" valign="bottom">50<hr/></td><td align="left" valign="bottom">0.55(0.36-0.85)<hr/></td><td align="left" valign="bottom">0.04<hr/></td></tr><tr><td align="left" valign="bottom">GA + AA<hr/></td><td align="left" valign="bottom">281<hr/></td><td align="left" valign="bottom">242<hr/></td><td align="left" valign="bottom">0.67(0.50-0.90)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">rs1864163<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">GG<hr/></td><td align="left" valign="bottom">289<hr/></td><td align="left" valign="bottom">282<hr/></td><td align="left" valign="bottom">1.00<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">GA<hr/></td><td align="left" valign="bottom">116<hr/></td><td align="left" valign="bottom">115<hr/></td><td align="left" valign="bottom">1.05(0.76-1.44)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">AA<hr/></td><td align="left" valign="bottom">18<hr/></td><td align="left" valign="bottom">23<hr/></td><td align="left" valign="bottom">1.49(0.75-2.93)<hr/></td><td align="left" valign="bottom">0.92<hr/></td></tr><tr><td align="left" valign="bottom">GA + AA<hr/></td><td align="left" valign="bottom">134<hr/></td><td align="left" valign="bottom">138<hr/></td><td align="left" valign="bottom">1.10(0.81-1.49)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">rs7499892<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">GG<hr/></td><td align="left" valign="bottom">261<hr/></td><td align="left" valign="bottom">259<hr/></td><td align="left" valign="bottom">1.00<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">GA<hr/></td><td align="left" valign="bottom">136<hr/></td><td align="left" valign="bottom">138<hr/></td><td align="left" valign="bottom">1.06(0.78-1.45)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">AA<hr/></td><td align="left" valign="bottom">27<hr/></td><td align="left" valign="bottom">24<hr/></td><td align="left" valign="bottom">0.96(0.53-1.75)<hr/></td><td align="left" valign="bottom">0.88<hr/></td></tr><tr><td align="left" valign="bottom">GA + AA<hr/></td><td align="left" valign="bottom">163<hr/></td><td align="left" valign="bottom">162<hr/></td><td align="left" valign="bottom">1.05(0.78-1.40)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">rs5882<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">AA<hr/></td><td align="left" valign="bottom">142<hr/></td><td align="left" valign="bottom">135<hr/></td><td align="left" valign="bottom">1.00<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">AG<hr/></td><td align="left" valign="bottom">219<hr/></td><td align="left" valign="bottom">215<hr/></td><td align="left" valign="bottom">1.00(0.73-1.37)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">GG<hr/></td><td align="left" valign="bottom">63<hr/></td><td align="left" valign="bottom">71<hr/></td><td align="left" valign="bottom">1.13(0.73-1.74)<hr/></td><td align="left" valign="bottom">0.89<hr/></td></tr><tr><td align="left" valign="bottom">AG + GG<hr/></td><td align="left" valign="bottom">282<hr/></td><td align="left" valign="bottom">286<hr/></td><td align="left" valign="bottom">1.03(0.76-1.39)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">rs9989419<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">GG<hr/></td><td align="left" valign="bottom">198<hr/></td><td align="left" valign="bottom">201<hr/></td><td align="left" valign="bottom">1.00<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">GA<hr/></td><td align="left" valign="bottom">185<hr/></td><td align="left" valign="bottom">179<hr/></td><td align="left" valign="bottom">0.96(0.71-1.29)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">AA<hr/></td><td align="left" valign="bottom">41<hr/></td><td align="left" valign="bottom">41<hr/></td><td align="left" valign="bottom">0.96(0.59-1.58)<hr/></td><td align="left" valign="bottom">0.71<hr/></td></tr><tr><td align="left">GA + AA</td><td align="left">226</td><td align="left">220</td><td align="left">0.96(0.72-1.27)</td><td align="left"> </td></tr></tbody></table><table-wrap-foot><p>ORs: adjusted for age.</p></table-wrap-foot></table-wrap><p>Four SNPs in the CETP gene (rs1800775-rs1532624-rs708272-rs1864163) were in linkage disequilibrium with D’ ranging from 0.66 to 1.00 and r<sup>2</sup> ranging from 0.07 to 0.55. However, we did not find any CETP haplotype that was significantly associated with risk of coronary atherosclerosis (data not shown).</p><p>Finally, we investigated the associations between the seven SNPs and serum HDL-C levels in 424 healthy controls. In univariate analyses, CETP rs1800775 was significantly associated with decreased serum level of HDL-C (p = 0.04). Carriers of the mutant alleles of rs1800775 (A/C: 1.45 ± 0.62 mmol/L; C/C: 1.28 ± 0.41 mmol/L) had significantly decreased serum HDL-C levels compared with subjects with GG genotype (1.49 ± 0.72 mmol/L). When ANCOVA model was applied, rs1800775 still showed significant association with HDL-C level (p = 0.04). None of the other studied SNPs was associated with serum HDL-C level (Table <xref ref-type="table" rid="T3">3</xref>).</p><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>Association between CETP polymorphisms and serum HDL-C levels in health control subjects</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th rowspan="2" align="left" valign="top"><bold>SNP</bold></th><th rowspan="2" align="left" valign="top"><bold>M/m</bold></th><th colspan="3" align="center" valign="bottom"><bold>HDL-C (mmol/L) mean ± SD</bold><hr/></th><th rowspan="2" align="left" valign="top"><bold>P</bold></th></tr><tr><th align="left"><bold>MM</bold></th><th align="left"><bold>Mm</bold></th><th align="left"><bold>mm</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">rs1800775<hr/></td><td align="left" valign="bottom">A/C<hr/></td><td align="left" valign="bottom">1.49 ± 0.72<hr/></td><td align="left" valign="bottom">1.45 ± 0.62<hr/></td><td align="left" valign="bottom">1.28 ± 0.41<hr/></td><td align="left" valign="bottom">0.04<hr/></td></tr><tr><td align="left" valign="bottom">rs1532624<hr/></td><td align="left" valign="bottom">C/A<hr/></td><td align="left" valign="bottom">1.38 ± 0.51<hr/></td><td align="left" valign="bottom">1.45 ± 0.68<hr/></td><td align="left" valign="bottom">1.54 ± 0.80<hr/></td><td align="left" valign="bottom">0.27<hr/></td></tr><tr><td align="left" valign="bottom">rs708272<hr/></td><td align="left" valign="bottom">G/A<hr/></td><td align="left" valign="bottom">1.34 ± 0.55<hr/></td><td align="left" valign="bottom">1.46 ± 0.62<hr/></td><td align="left" valign="bottom">1.46 ± 0.70<hr/></td><td align="left" valign="bottom">0.15<hr/></td></tr><tr><td align="left" valign="bottom">rs1864163<hr/></td><td align="left" valign="bottom">G/A<hr/></td><td align="left" valign="bottom">1.43 ± 0.65<hr/></td><td align="left" valign="bottom">1.44 ± 0.57<hr/></td><td align="left" valign="bottom">1.22 ± 0.39<hr/></td><td align="left" valign="bottom">0.39<hr/></td></tr><tr><td align="left" valign="bottom">rs7499892<hr/></td><td align="left" valign="bottom">G/A<hr/></td><td align="left" valign="bottom">1.46 ± 0.65<hr/></td><td align="left" valign="bottom">1.39 ± 0.54<hr/></td><td align="left" valign="bottom">1.27 ± 0.64<hr/></td><td align="left" valign="bottom">0.23<hr/></td></tr><tr><td align="left" valign="bottom">rs5882<hr/></td><td align="left" valign="bottom">A/G<hr/></td><td align="left" valign="bottom">1.39 ± 0.61<hr/></td><td align="left" valign="bottom">1.46 ± 0.66<hr/></td><td align="left" valign="bottom">1.38 ± 0.48<hr/></td><td align="left" valign="bottom">0.49<hr/></td></tr><tr><td align="left">rs9989419</td><td align="left">G/A</td><td align="left">1.46 ± 0.67</td><td align="left">1.39 ± 0.57</td><td align="left">1.42 ± 0.59</td><td align="left">0.55</td></tr></tbody></table><table-wrap-foot><p>M indicates major allele; m indicates minor allele.</p></table-wrap-foot></table-wrap></sec><sec sec-type="discussion"><title>Discussion</title><p>Considering the crucial role of CETP in lipid metabolism, we investigated the association of seven SNPs in this gene and the risk of coronary atherosclerosis in a Chinese population. Our results showed that rs708272 polymorphism may reduce the risk of coronary atherosclerosis and rs1800775 mutation may decrease serum HDL-C levels in healthy controls in our population.</p><p>The human CETP gene is located on chromosome 16q21, and several mutations in this gene have been reported to alter the function of CETP and plasma HDL-C levels [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B13">13</xref>]. Among these SNPs, CETP TaqIB site (rs708272) is the most studied one. The results of prospective Women’s Genome Health Study (WGHS), conducted in American women, showed that carriers of B2B2 and B1B2 genotypes (56 mg/dL and 52 mg/dL) had significant higher levels of plasma HDL-C than carriers of B1B1 genotype (50 mg/dL) [<xref ref-type="bibr" rid="B10">10</xref>]. Moreover, a meta-analysis, based on the data from year 1970 to 2008, also found that individuals with TaqIB B2 allele had higher mean HDL-C concentrations and lower mean CETP activity compared with carriers of B1 allele [<xref ref-type="bibr" rid="B13">13</xref>]. Besides, the relationship between TaqIB polymorphism and HDL-C levels was also confirmed by other studies [<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B19">19</xref>]. But the association between CETP TaqIB and CAD risk lacks consistency. A case–control study among Singapore population [<xref ref-type="bibr" rid="B20">20</xref>] found that the absence of B2 allele was associated with 2.0-fold increased risk of CAD, but a meta-analysis revealed that B2B2 genotype had a 1.45-fold significant increased risk of CAD compared with B1B1 genotype in population-based studies [<xref ref-type="bibr" rid="B21">21</xref>]. In our study, we also found that B2B2 genotype conferred a significant reduced risk of coronary atherosclerosis which had the same directions of risk with the previous study done in Singapore population. In addition, controls with B1B2 or B2B2 genotype had non-significant higher level of serum HDL-C (1.46 ± 0.62 mmol/L and 1.46 ± 0.70 mmol/L) compared with the controls with B1B1 genotype (1.34 ± 0.55 mmol/L), which was in accordance with the WGHS results. CETP TaqIB is located in intron 1 of CETP gene, and the functional phenotype of this mutation is still unclear.</p><p>CETP -629C/A (rs1800775) is located in promoter region of the CETP gene, and this variation may influence the functionality of CETP gene by changing in binding site Sp1/Sp3 and in turn repress CETP promoter activity [<xref ref-type="bibr" rid="B22">22</xref>,<xref ref-type="bibr" rid="B23">23</xref>]. Rs1800775 was found to be significantly associated with plasma HDL-C levels in the WGHS with average HDL-C (mg/dL) of 49 in CC carriers, 52 in CA carriers and 55 in AA carriers [<xref ref-type="bibr" rid="B10">10</xref>]. The same tend of association was also found among Latvian population [<xref ref-type="bibr" rid="B24">24</xref>]. In our study, healthy controls carrying the CC genotype (1.28 ± 0.41 mmol/L) had significant lower level of HDL-C compared with subjects with AA genotype (1.49 ± 0.72 mmol/L), which was consistent with previous reports. Besides, we found CC genotype conferred a non-significant elevated risk of coronary atherosclerosis (OR = 1.43, 95% CI = 0.93-2.19). Rs1800775 was in high linkage disequilibrium with rs708272 in our study (D’ = 0.97, r<sup>2</sup> = 0.55), and this association was also suggested by several other studies [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B25">25</xref>].</p><p>The exact mechanisms which may link CETP genetic polymorphism to coronary atherosclerosis are largely unclear. But the linkage between CETP polymorphisms and plasma HDL-C level may provide a possible mechanism that merits further investigation. In addition, the limitation of this study was the relatively small sample size, which hampered our ability to detect some significant associations.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>In conclusion, the present study demonstrating that the A allele of the rs708272 may confer decreased risk of coronary atherosclerosis, and rs1800775 may exert effect on serum HDL-C levels in Chinese population. Future studies with larger sample sizes as well as functional studies are needed to confirm our findings.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Study subjects</title><p>A case–control study including 420 consecutive patients with coronary atherosclerosis and 424 healthy controls was carried out. Cases were collected from Jinling hospital in Nanjing between June 2008 and May 2012. Clinical diagnosis of coronary atherosclerosis was evaluated by percutaneous coronary angiography, reviewed by two experienced cardiologists. Healthy control subjects, without coronary atherosclerosis, were selected in the same period and from the same hospital, and were frequency matched to the cases by age (5-year age groups) and gender. All controls were individuals free of CAD that determined by medical history, clinical examinations, or electrocardiography. Those with coronary myocardial bridge were excluded from the study. At enrollment, demographic characteristics, anthropometric measures, medical histories were collected from each subject by a trained interviewer using a structured questionnaire. Written informed consent was obtained from all enrolled participants and this study was approved by the Ethics Committee of Jinling hospital.</p></sec><sec><title>Laboratory tests</title><p>Blood samples were drawn for measurement of serum levels of TC, HDL-C, LDL-C after a 12-hour overnight fast. Serum levels of TC (mmol/L), HDL-C (mmol/L), and LDL-C (mmol/L) were determined by colorimetric enzymatic assays with use of an Auto-Analyzer (AU 2700 Olympus, FirstChemical Ltd, Tokyo, Japan).</p></sec><sec><title>Genotyping</title><p>Genomic DNA was extracted from peripheral blood leucocytes using Promega DNA Extraction Kit (Promega, Madison, WI, USA). Genotyping was performed using the TaqMan assay on the ABI PRISM 7900HT Sequence Detection System (Applied Biosystems, Foster City, CA, USA), in a 384-well format, with dual fluorescent reporter probes VIC and FAM. The genotyping call rate was >95%, and the completion rate was >99%. The quality and potential misclassification of the genotyping were assessed by re-evaluating 5% of duplicate DNA samples (42 total samples) that were randomly selected from the whole population and placed within the same reaction plates used for the study subjects. The concordance rate for the quality control samples was 100%.</p></sec><sec><title>Statistical analysis</title><p>We used SAS software (version 9.3; SAS Institute, Inc.) for the statistical analyses. χ<sup>2</sup> statistics and the t test were used to evaluate case–control differences in the distribution of risk factors. Variables were tested for normality with Shapiro-Wilk statistics. Skewed data, including age, BMI, TC, HDL-C and LDL-C were log transformed and expressed as medians and interquartile ranges. The odds ratios (ORs) and 95% confidence intervals (CIs) for the associations between the SNPs and disease risk were estimated by unconditional logistic regression. Hardy-Weinberg equilibrium for genotypic distribution and linkage disequilibrium between loci were assessed by HaploView version 4.0 (Daly Lab at the Broad Institute, Cambridge, MA, USA) [<xref ref-type="bibr" rid="B26">26</xref>]. Associations between haplotypes (> 1% frequency) and the risk of coronary atherosclerosis were evaluated by computing OR and 95% CI using HAPSTAT, assuming an additive model, using the most common haplotype as the referent category [<xref ref-type="bibr" rid="B27">27</xref>]. Both univariate ANOVA and multivariate ANCOVA analyses adjusting for age, smoking, diabetes and hypertension were performed to determine the effects of the CETP polymorphisms on serum HDL-C levels. A two tailed P-value of 0.05 was considered statistically significant.</p></sec></sec><sec><title>Abbreviations</title><p>CAD: Coronary artery disease; HDL-C: High-density lipoprotein cholesterol; LDL-C: Low-density lipoprotein cholesterol; CETP: Cholesteryl ester transfer protein; SNP: Single nucleotide polymorphisms; TC: Total cholesterol; CHB: Han Chinese in Beijing; BMI: Body mass index.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>JW and LJW designed the molecular genetic studies and drafted the manuscript. YZ and PG carried out the genotyping experiments. JQS performed the statistical analysis. SSJ and JBG participated in the design of sample collection and improved the manuscript. All authors read and approved the final manuscript.</p></sec>
Association of angiotensin converting enzyme gene insertion/deletion polymorphism and familial hypercholesterolemia in the Saudi population	<sec><title>Background</title><p>The study of the association between genotype and phenotype is of great importance for the prediction of multiple diseases and pathophysiological conditions. The relationship between angiotensin converting enzyme (<italic>ACE</italic>) Insertion/Deletion (I/D) polymorphism and Familial Hypercholesterolemia (FH) has been not fully investigated in all the ethnicities. In this study we sought to determine the frequency of I/D polymorphism genotypes of <italic>ACE</italic> gene in Saudi patients with FH.</p></sec><sec><title>Results</title><p>This is a case–control study carried out purely in Saudi population. Genomic DNA was isolated from 128 subjects who have participated in this study. <italic>ACE</italic> gene I/D polymorphism was analyzed by polymerase chain reaction in 64 FH cases and 64 healthy controls. There was no statistically significant difference between the groups with respect to genotype distribution. Furthermore, we did not find any significant difference in the frequency of <italic>ACE</italic> I/D polymorphism in FH subjects when stratified by gender (<italic>p</italic> = 0.43).</p></sec><sec><title>Conclusion</title><p>Our data suggest that <italic>ACE</italic> gene I/D polymorphism examined in this study has no role in predicting the occurrence and diagnosis of FH.</p></sec>	<contrib contrib-type="author" equal-contrib="yes" id="A1"><name><surname>Alharbi</surname><given-names>Khalid K</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>kharbi@ksu.edu.sa</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Kashour</surname><given-names>Tarek S</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>tkashour@ksu.edu.sa</email></contrib><contrib contrib-type="author" equal-contrib="yes" id="A3"><name><surname>Al-Hussaini</surname><given-names>Wejdan</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I3">3</xref><email>wej444dan@hotmail.com</email></contrib><contrib contrib-type="author" equal-contrib="yes" id="A4"><name><surname>Al-Nbaheen</surname><given-names>May Salem</given-names></name><xref ref-type="aff" rid="I3">3</xref><xref ref-type="aff" rid="I4">4</xref><email>malnbaheen@ksu.edu.sa</email></contrib><contrib contrib-type="author" equal-contrib="yes" id="A5"><name><surname>Mohamed</surname><given-names>Sarar</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>msarar@ksu.edu.sa</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Hasanato</surname><given-names>Rana MW</given-names></name><xref ref-type="aff" rid="I6">6</xref><email>rhasanato@ksu.edu.sa</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Tamimi</surname><given-names>Waleed</given-names></name><xref ref-type="aff" rid="I7">7</xref><email>tamimiw@ngha.med.sa</email></contrib><contrib contrib-type="author" id="A8"><name><surname>Al-Naami</surname><given-names>Mohammed Yahya</given-names></name><xref ref-type="aff" rid="I8">8</xref><email>alnaami@ksu.edu.sa</email></contrib><contrib contrib-type="author" corresp="yes" equal-contrib="yes" id="A9"><name><surname>Khan</surname><given-names>Imran Ali</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>imkhan@ksu.edu.sa</email></contrib>	Lipids in Health and Disease	<sec sec-type="intro"><title>Introduction</title><p>Familial hypercholesterolemia (FH, OMIM 143890) is an inherited disorder of lipoprotein metabolism, transmitted in an autosomal dominant manner and clinically characterized by elevated serum levels of total and low-density lipoprotein (LDL) cholesterol, the presence of tendon xanthomas and premature atherosclerosis [<xref ref-type="bibr" rid="B1">1</xref>]. The genetic basis of FH is a large array of mutations in the LDL-receptor (LDLR) gene (OMIM 606945), resulting in a lack of functional receptors for LDL on the liver cell surface, giving rise to increased plasma LDL levels [<xref ref-type="bibr" rid="B2">2</xref>]. Population frequency of heterozygous familial hypercholesterolemia (hFH) is approximately 1/500. Although genetically the disease is caused by mutations in the LDLR gene, the clinical phenotype of FH can vary regardless of the mutation and this variability is assumed to be due to both environmental and genetic factors [<xref ref-type="bibr" rid="B3">3</xref>]. The clinical phenotype caused by mutations in LDLR, APOB or PCSK9 and characterized by elevated levels of plasma LDL-Cholesterol (LDL-C) is currently referred to as Autosomal Dominant Hypercholesterolemia (ADH, OMIM 603776) [<xref ref-type="bibr" rid="B1">1</xref>].</p><p>Gene polymorphisms are markers of biologic diversity, and some genotypic variations correlate with specific phenotypes relevant to the human diseases. However, it is not clear whether many of these variants are involved in the pathogenesis of diseases or are merely in proximity to other pathogenic genetic factors-a phenomenon known as linkage disequilibrium. Angiotensin-converting enzyme (ACE) gene insertion/deletion (I/D) polymorphism is responsible for inter individual differences in ACE plasma levels. It is known that the frequency of ACE D alleles varies in different ethnic groups [<xref ref-type="bibr" rid="B4">4</xref>]. A previous study by O'Malley et al. [<xref ref-type="bibr" rid="B5">5</xref>] have shown an association with FH patients who have myocardial infarction (MI) and coronary heart disease (CHD) [<xref ref-type="bibr" rid="B5">5</xref>]. I/D polymorphism have also been associated with risk of MI [<xref ref-type="bibr" rid="B6">6</xref>]. ACE gene is involved in the conversion of Angiotensin I to Angiotensin II by its metalloproteinase enzymatic activity. It plays a major role in the rennin–angiotensin (RAS) system. I/D polymorphism (Consisting of a 287-bp fragment) in intron 16 of the gene has been related to the amount of circulating enzyme. Individuals homozygous for the deletion have an approximately two-fold higher level of circulating enzyme in comparison to individuals homozygous for the insertion [<xref ref-type="bibr" rid="B7">7</xref>]. Angiotensin II increases vascular smooth muscle cell proliferation, monocyte adhesion, platelet adhesion and aggregation. ACE genotype acting through the tissue or possible plasma ACE concentrations has been implicated in the aetiology of the disease [<xref ref-type="bibr" rid="B8">8</xref>].</p><p>ACE is a zinc metalloprotease that is widely distributed on the surface of epithelial and endothelial cells. Its gene consists of 26 exons and spans 21 kb on the human chromosome 17. The sequence codes a 1306-amino-acid protein, including a signal peptide. It has a common polymorphism that consists of the presence (I allele) or the absence (D allele) of a 287-bp Alu repeat sequence within intron 16. Although the I/D polymorphism is in the intronic region of the ACE gene, studies have shown that the DD genotype is strongly associated with increased plasma or serum ACE levels [<xref ref-type="bibr" rid="B9">9</xref>]. ACE activity in individuals with DD genotype is twice that found in those with II genotypes. Subjects with ID genotype exhibit intermediate levels of ACE [<xref ref-type="bibr" rid="B7">7</xref>]. This ACE I/D gene polymorphism has been implicated as a risk factor for a number of pathologies, such as MI, stroke, Type 2 Diabetes Mellitus (T2DM), Diabetic Nephropathy (DN), and hypertension in different ethnic groups, but these findings are far from conclusive [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. The aim of this study was to analyze whether the ACE gene I/D polymorphism is associated with the FH patients in Saudi population.</p></sec><sec><title>Materials and methodology</title><p>In this study 64 subjects diagnosed with FH, based on the Deutch working group classification criteria [<xref ref-type="bibr" rid="B11">11</xref>]. Patients were recruited from outpatient clinic at King Khalid University Hospital (KKUH), Riyadh. Healthy volunteers (<italic>n</italic> = 64) were recruited from Medical and laboratory staff at KKUH, and individuals attend KKUH for routine checkup and who were not having chronic metabolic/ medical disease. The research protocol was approved by the Institutional Review Board at KKUH (E-12-829). All participants provided written informed consent prior to enrollment into the study.</p><sec><title>Blood</title><p>Participants were advised to fast overnight for at least 10 hours prior to sample collection. Five mL of blood sample was collected from every participant; 3 mL of the serum sample was used for biochemical parameters like Total Cholesterol (TC), Triglycerides (TG), HDL-C and LDLC to confirm the FH and 2 mL of EDTA blood sample was used for molecular investigation.</p></sec><sec><title>Biochemical analysis</title><p>Serum samples of TC and TG were measured by the standard enzymatic method (Pars Azmon kit, Iran) using an automated RA-1000 (Technicon, USA). Serum LDL-C and HDL-C levels were measured using commercially available enzyme assay kits (Pars Azmon kit, Iran).</p></sec><sec><title>Genomic DNA isolation and genotyping of ACE I/D polymorphism</title><p>The blood samples which were collected in the EDTA tubes were used for extraction of DNA from the peripheral blood leukocytes. Genomic DNA was extracted by standard protocol [<xref ref-type="bibr" rid="B12">12</xref>] with Norgen DNA extraction kit (Norgen Biotek corp, Canada). Extracted DNA was quantified by NanoDrop to the DNA concentration. ACE genotypes were determined by Polymerase Chain reaction (PCR) analysis as described in our previous work [<xref ref-type="bibr" rid="B12">12</xref>]. Specific oligonucleotide primers: forward primer of 5′CTGGAGACCACTCCCATCCTTTCT-3′ and the reverse primer of 5′- GATGTGGCCATCACATTCGTCACGAT-3′ were applied to acquire a 490 bp DNA fragment in intron 16. The amplification was carried out in a total volume of 20 μL reaction mixture contain 10 μM of each primer, 6 μL of sterile water and 10 μL of 2X master mix which includes MgCl<sub>2</sub>, 10x PCR buffer, dNTPs, 10 units of Taq DNA polymerase (Norgen Biotek corp, Canada) and the 60 ng template DNA. The PCR was performed in a thermal cycler (Applied Biosystems, Hercules, California, USA). The cycling and amplification conditions were as follows; an initial denaturation was set up for 5 minutes at 95°C followed by 35 cycles of denaturation for 30 seconds at 95°C, annealing for 30 seconds at 59°C, extension for 45 seconds at 72°C and the final extension was at 72°C for 5 minutes. The insertion band is considered as 490 bp which represents the I allele and 190 bp is identified as deletion band (without 287-bp Alu insertion) which represents the D allele and the I/D band was 490/190 bp which is considered as the heterozygous (Figure  <xref ref-type="fig" rid="F1">1</xref>). The amplified PCR products were clearly separated on 2.5% agarose gel with ethidium bromide stained (Cambrex, East Rutherford, NJ, USA) and visualized on a transilluminator (Dafco, USA).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Identification of ACE I/D polymorphism in 2.5% agarose gel electrophoresis.</bold> Legend: Lane 1: 50bp DNA Marker/Ladder. Lane 2&6: Homozygous II genotype (490 bp). Lane 3&5: Homozygous DD genotype (190 bp). Lane 4: Heterozygous ID genotype (490/190 bp).</p></caption><graphic xlink:href="1476-511X-12-177-1"/></fig></sec><sec><title>Statistical analysis</title><p>All the statistical analyses were carried out using SPSS (Chicago, IL, USA) software version 19.0 for Microsoft Windows®. Clinical characteristics of all subjects were expressed as mean ± SD. Continuous variables were compared between the groups using two-tailed student’s <italic>t</italic> test. Significant cutoff was set at 0.05. Crude ORs with 95% CIs were used to evaluate the strength of the association between ACE I/D polymorphism and FH. Pooled ORs were calculated for allelic contrast (D vs. I), dominant (DD + DI vs. II) and recessive (DD vs. ID + II) genetic model. Z-test was used to determine the significance of the pooled ORs, and <italic>p</italic> value < 0.05 was considered significant. The effect of I/D genotypes of ACE gene was analyzed using general linear model ANOVA for clinical characteristics. A level of <italic>p</italic> <0.05 was considered statistically significant.</p></sec></sec><sec><title>Result</title><p>The demographic and the biological data of all the patients participating in this study are shown in Table  <xref ref-type="table" rid="T1">1</xref>. The lipid profile parameters used in this study are fasting TC, TG, HDL-C and LDL-C. TG, TC and HDL-C levels in the FH group were 1.3 ± 0.50 mmol/L, 6.4 ± 0.68 mmol/L and 1.4 ± 0. 24 mmol/L compared to 1.1 ± 0.36 mmol/L, 4.6 ± 0.39 mmol/L and 0.9 ± 0.21 mmol/L in the control group (<italic>p</italic> < 0.05). The LDL-C level was 4.8 ± 0.55 mmol/L in FH individuals compared to 3.1 ± 0.32 mmol/L in the control group (<italic>p</italic> > 0.05). There was significant difference in the Age, TG, TC and LDL-C levels between the FH subjects and the healthy controls (<italic>p</italic> < 0.05), but there was no difference between FH subjects and healthy controls in HDL-C and gender (<italic>p</italic> > 0.05).</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Demographic characteristics and biochemical profile of FH patients and healthy control</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th rowspan="2" align="center"><bold>S. no</bold></th><th rowspan="2" align="center"> </th><th align="center" valign="bottom"><bold>FH cases</bold><hr/></th><th align="center" valign="bottom"><bold>Healthy controls</bold><hr/></th><th rowspan="2" align="center"><bold> <italic>p </italic> </bold><bold>value</bold></th></tr><tr><th align="center"><bold>(</bold><bold><italic>n</italic></bold> <bold>= 64)</bold></th><th align="center"><bold>(</bold><bold><italic>n</italic></bold> <bold>= 64)</bold></th></tr></thead><tbody valign="top"><tr><td align="center" valign="bottom">1<hr/></td><td align="center" valign="bottom">Age (Years)<hr/></td><td align="center" valign="bottom">52.06 ± 10.27<hr/></td><td align="center" valign="bottom">44.8 ± 6.87<hr/></td><td align="center" valign="bottom"><italic>p</italic> = 0.0001<hr/></td></tr><tr><td align="center" valign="bottom">2<hr/></td><td align="center" valign="bottom">Gender: Male/Female<hr/></td><td align="center" valign="bottom">20:44<hr/></td><td align="center" valign="bottom">22:42<hr/></td><td align="center" valign="bottom"><italic>p</italic> = 0.7<hr/></td></tr><tr><td align="center" valign="bottom">3<hr/></td><td align="center" valign="bottom">TG (mmol/L)<hr/></td><td align="center" valign="bottom">1.3 ± 0.50<hr/></td><td align="center" valign="bottom">1.1 ± 0.36<hr/></td><td align="center" valign="bottom"><italic>p</italic> = 0.01<hr/></td></tr><tr><td align="center" valign="bottom">4<hr/></td><td align="center" valign="bottom">TC (mmol/L)<hr/></td><td align="center" valign="bottom">6.4 ± 0.68<hr/></td><td align="center" valign="bottom">4.6 ± 0.39<hr/></td><td align="center" valign="bottom"><italic>p</italic> = 0.04<hr/></td></tr><tr><td align="center" valign="bottom">5<hr/></td><td align="center" valign="bottom">HDL-C (mmol/L)<hr/></td><td align="center" valign="bottom">1.4 ± 0.24<hr/></td><td align="center" valign="bottom">0.9 ± 0.21<hr/></td><td align="center" valign="bottom"><italic>p</italic> = 0.55<hr/></td></tr><tr><td align="center">6</td><td align="center">LDL-C (mmol/L)</td><td align="center">4.8 ± 0.55</td><td align="center">3.1 ± 0.32</td><td align="center"><italic>p</italic> = 0.01</td></tr></tbody></table></table-wrap><sec><title>Screening of I/D polymorphism in the ACE gene</title><p>The distribution of ACE I/D polymorphism among FH individuals is shown in Table  <xref ref-type="table" rid="T2">2</xref>. The frequencies of ACE DD, ID and II genotypes among FH patients were 48.4%, 32.8% and 18.7% respectively. The percentage of D allele was 0.65% and of that of I allele was 0.35%. In the control subjects, the distribution of ACE DD, ID and II genotypes was 39%, 32.9% and 28.1% respectively. The allele frequencies was 0.55% and 0.45% for the D and I alleles, respectively. When we compared the frequency of ACE DD genotypes and D allele in FH cases with healthy controls, the difference was not statistically significant (OR = 1.696, <italic>p</italic> = 0.21 (95% CI = 0.7385-3.893) and OR = 1.481, <italic>p</italic> = 0.12 (95% CI = 0.8953-2.449) (Table  <xref ref-type="table" rid="T3">3</xref>). Similarly, we did not find any significant difference in the D allele frequencies between patients and controls (Table  <xref ref-type="table" rid="T3">3</xref>). We also explored the association between ACE I/D polymorphism and FH according to gender and found no significant difference, OR = 0.7373, <italic>p</italic> = 0.43 (95% CI = 0.3403-1.597) (Table  <xref ref-type="table" rid="T4">4</xref>).</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Genotype and allele frequency of ACE gene</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="center"><bold>Genotypes and Alleles</bold></th><th align="center"><bold>FH cases (</bold><bold><italic>n</italic></bold> <bold>= 64)</bold></th><th align="center"><bold>Healthy controls (</bold><bold><italic>n</italic></bold> <bold>= 64)</bold></th></tr></thead><tbody valign="top"><tr><td align="center" valign="bottom">II<hr/></td><td align="center" valign="bottom">12 (18.7)<hr/></td><td align="center" valign="bottom">18 (28.1)<hr/></td></tr><tr><td align="center" valign="bottom">ID<hr/></td><td align="center" valign="bottom">21 (32.8)<hr/></td><td align="center" valign="bottom">21 (32.9)<hr/></td></tr><tr><td align="center" valign="bottom">DD<hr/></td><td align="center" valign="bottom">31 (48.4)<hr/></td><td align="center" valign="bottom">25 (39)<hr/></td></tr><tr><td align="center" valign="bottom">I<hr/></td><td align="center" valign="bottom">45 (0.35)<hr/></td><td align="center" valign="bottom">57 (0.45)<hr/></td></tr><tr><td align="center">D</td><td align="center">83 (0.65)</td><td align="center">71 (0.55)</td></tr></tbody></table></table-wrap><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>Statistical analysis for FH individuals</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>S. no</bold></th><th align="left"><bold>Genotypes</bold></th><th align="left"><bold>Odds ratio</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">1<hr/></td><td align="left" valign="bottom">DD Vs ID + II<hr/></td><td align="left" valign="bottom">OR-1.465;95% CI = 0.7265–2.956; <italic>p</italic> = 0.28<hr/></td></tr><tr><td align="left" valign="bottom">2<hr/></td><td align="left" valign="bottom">ID + DD Vs II<hr/></td><td align="left" valign="bottom">OR-1.696;95% CI = 0.7385–3.893; <italic>p</italic> = 0.21<hr/></td></tr><tr><td align="left" valign="bottom">3<hr/></td><td align="left" valign="bottom">ID Vs II + DD<hr/></td><td align="left" valign="bottom">OR-1.00;95% CI = 0.4781–2.091; <italic>p</italic> = 0.85<hr/></td></tr><tr><td align="left">4</td><td align="left">D Vs I</td><td align="left">OR-1.481;95% CI = 0.8953–2.449; <italic>p</italic> = 0.12</td></tr></tbody></table></table-wrap><table-wrap position="float" id="T4"><label>Table 4</label><caption><p>Genotype and allele frequency of ACE gene between males and females</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="center"><bold>Genotypes and alleles</bold></th><th align="center"><bold>Males (</bold><bold><italic>n</italic></bold> <bold>= 20)</bold></th><th align="center"><bold>Females (</bold><bold><italic>n</italic></bold> <bold>= 44)</bold></th><th align="center"><bold>OR (95% CI)</bold></th><th align="center"><bold> <italic>p </italic> </bold><bold>value</bold></th></tr></thead><tbody valign="top"><tr><td align="center" valign="bottom">II<hr/></td><td align="center" valign="bottom">4 (20%)<hr/></td><td align="center" valign="bottom">8 (18.2%)<hr/></td><td align="center" valign="bottom">0.6087 (0.2083–1.779)<hr/></td><td align="center" valign="bottom"><italic>p</italic> = 0.36<hr/></td></tr><tr><td align="center" valign="bottom">ID<hr/></td><td align="center" valign="bottom">8 (40%)<hr/></td><td align="center" valign="bottom">13 (29.5%)<hr/></td><td align="center" valign="bottom">1.59 (0.5629–4.797)<hr/></td><td align="center" valign="bottom"><italic>p</italic> = 0.40<hr/></td></tr><tr><td align="center" valign="bottom">DD<hr/></td><td align="center" valign="bottom">8 (40%)<hr/></td><td align="center" valign="bottom">23 (52.3%)<hr/></td><td align="center" valign="bottom">0.8889 (0.2335–3.384)<hr/></td><td align="center" valign="bottom"><italic>p</italic> = 0.86<hr/></td></tr><tr><td align="center" valign="bottom">I<hr/></td><td align="center" valign="bottom">16 (0.4)<hr/></td><td align="center" valign="bottom">29 (0.33)<hr/></td><td align="center" valign="bottom">Ref<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="center">D</td><td align="center">24 (0.6)</td><td align="center">59 (0.67)</td><td align="center">0.7373 (0.3403–1.597)</td><td align="center"><italic>p</italic> = 0.43</td></tr></tbody></table></table-wrap></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>To the best of our knowledge this is the first study to investigate the association of ACE gene I/D polymorphism and FH in Saudi population. In our study FH individuals tended to have higher DD (48.4%) genotypes and more ID (32.9%) genotypes than controls. However, there was no significant difference in the allelic frequency between the patients and controls.</p><p>Previous studies addressing the association between FH and ACE gene polymorphism focused on the potential risk of CHD coronary heart disease associated with ACE gene polymorphism in FH patients. For example, O’Malley et al. [<xref ref-type="bibr" rid="B5">5</xref>] showed that in heterozygote FH patients with ACE-DD genotype, the incidence of MI is 2.5 times higher and CHD is 2.2 higher than in those with DI or II genotypes. On the other hand, another study [<xref ref-type="bibr" rid="B13">13</xref>], did not find any association between ACE gene deletion/insertion polymorphism and increased risk of CHD in patients with FH.</p><p>A few studies have been conducted in Saudi Arabia and Middle East looking at the association between ACE gene polymorphism and certain diseases. We have reported previously an association between ACE gene polymorphism and G6PD deficiency [<xref ref-type="bibr" rid="B12">12</xref>]. El-Hazmi et al. [<xref ref-type="bibr" rid="B14">14</xref>] also showed a significant association between ACE DD genotype and obesity among Saudi patients.</p><p>In contrary to that, Dzimiri et al. [<xref ref-type="bibr" rid="B15">15</xref>] showed no association between ACE genotype and risk of coronary artery disease in Saudi patients. Another study reported by Chmaisse et al. [<xref ref-type="bibr" rid="B16">16</xref>] showed no association between T2DM and ACE gene polymorphism in s mall Lebanese cohort. Our study is limited by small sample size and the fact that our patients have no documented coronary heart disease which may explain our inability to show an association between ACE gene polymorphism and FH.</p></sec><sec sec-type="conclusions"><title>Conclusion</title><p>In conclusion, our study has not shown an association between I/D polymorphism of ACE gene and FH in the Saudi populations. However, considering the relatively small sample sizes of our study, a firm conclusion cannot be made and larger studies in more well-characterized subjects should be conducted in the future.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>AKK participated in the design of the study and main investigator of this study. KTS has confirmed the FH patients and drafted the manuscript. AW has carried out the DNA isolation. AMS was the Co-I of this study. SM was an endocrinologist, helped in the FH cases, reviewed and edited the manuscript. HRMW carried out the lipid profile analysis. AMY has collected the control samples. TW has helped in this project with statistics. IAK: Design the study, performed the genotyping, written and edited the manuscript. All authors read and approved the final manuscript.</p></sec>
Mapping the FACT-G cancer-specific quality of life instrument to the EQ-5D and SF-6D	<sec><title>Objective</title><p>To help facilitate economic evaluations of oncology treatments, we mapped responses on cancer-specific instrument to generic preference-based measures.</p></sec><sec><title>Methods</title><p>Cancer patients (n = 367) completed one cancer-specific instrument, the FACT-G, and two preference-based measures, the EQ-5D and SF-6D. Responses were randomly divided to form development (n = 184) and cross-validation (n = 183) samples. Relationships between the instruments were estimated using ordinary least squares (OLS), generalized linear models (GLM), and censored least absolute deviations (CLAD) regression approaches. The performance of each model was assessed in terms of how well the responses to the cancer-specific instrument predicted EQ-5D and SF-6D utilities using mean absolute error (MAE) and root mean squared error (RMSE).</p></sec><sec><title>Results</title><p>Physical, functional, and emotional well-being domain scores of the FACT-G best explained the EQ-5D and SF-6D. In terms of accuracy of prediction as measured in RMSE, the CLAD model performed best for the EQ-5D (RMSE = 0.095) whereas the GLM model performed best for the SF-6D (RMSE = 0.061). The GLM predicted SF-6D scores matched the observed values more closely than the CLAD and OLS.</p></sec><sec><title>Conclusion</title><p>Our results demonstrate that the estimation of both EQ-5D and SF-6D utility indices using the FACT-G responses can be achieved. The CLAD model for the EQ-5D and the GLM model for the SF-6D are recommended. Thus, it is possible to estimate quality-adjusted life years for economic evaluation from studies where only cancer-specific instrument have been administered.</p></sec>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Teckle</surname><given-names>Paulos</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><xref ref-type="aff" rid="I3">3</xref><email>pteckle@bccrc.ca</email></contrib><contrib contrib-type="author" id="A2"><name><surname>McTaggart-Cowan</surname><given-names>Helen</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>hcowan@bccrc.ca</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Van der Hoek</surname><given-names>Kim</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>kvdhoek@bccrc.ca</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Chia</surname><given-names>Stephen</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>schia@bccancer.bc.ca</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Melosky</surname><given-names>Barb</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>bmelosky@bccancer.bc.ca</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Gelmon</surname><given-names>Karen</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>kgelmon@bccancer.bc.ca</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Peacock</surname><given-names>Stuart</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><xref ref-type="aff" rid="I3">3</xref><email>speacock@bccrc.ca</email></contrib>	Health and Quality of Life Outcomes	<sec sec-type="intro"><title>Introduction</title><p>Cancer is now the leading cause of death in many developed countries including Canada [<xref ref-type="bibr" rid="B1">1</xref>]. It accounted for 30 per cent of all deaths in 2008, followed by heart disease (21%) and stroke (6%) [<xref ref-type="bibr" rid="B2">2</xref>]. Over the past 50 years, survival rates for many forms of cancer have improved markedly due to advances in surgery, radiation therapy, and chemotherapy [<xref ref-type="bibr" rid="B1">1</xref>]. However, emerging cancer therapies often come at high costs. This not only imposes an increasing financial burden on health care systems but also raises questions about how to assess best value for money [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B4">4</xref>]. Economic evaluation is used by decision-makers in Canada and other countries to inform the allocation of scarce resources across health care interventions. Specifically, Cost-Utility Analysis (CUA) is increasingly becoming the main approach used to measure and value the impacts of interventions. CUA uses quality-adjusted life years (QALYs) to measure health outcomes by combining survival and health-related quality of life (HRQoL) into a single index. The use of CUA is recommended by the Canadian Agency for Drugs and Technologies in Health (CADTH), as well as peak regulatory bodies in Australia, the UK, and elsewhere in Europe [<xref ref-type="bibr" rid="B5">5</xref>-<xref ref-type="bibr" rid="B10">10</xref>].</p><p>The Functional Assessment of Cancer Therapy-General (FACT-G) is one of the most widely used cancer-specific HRQoL instrument [<xref ref-type="bibr" rid="B11">11</xref>]. It has been validated across a wide range of different types of cancer patients, cultures, and languages, and can be used to assess the impacts of cancer and its treatment on the physical and psycho-social well-being of patients [<xref ref-type="bibr" rid="B11">11</xref>]. In spite of its widespread use in clinical trials, responses from the FACT-G cannot be readily used in economic evaluations because it does not provide a single preference-based index of HRQoL suitable for use in CUA. Health utilities, elicited using, for example, the EuroQol-5D (EQ-5D) and the Short Form-6D (SF-6D), provide preference weights from the general population that can be used to calculate QALYs for CUA; this can inform decisions of health-related resource allocation [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B13">13</xref>]. However preference-based instruments typically cover broader health dimensions, and are not often administered in cancer clinical trials because many of the dimensions may neither be relevant nor sensitive to treatment effects [<xref ref-type="bibr" rid="B14">14</xref>].</p><p>When preference-based instruments are not administered in clinical trials, comparing effectiveness across different interventions becomes cumbersome. One option to overcome this limitation is to map, or ‘cross-walk’, the responses from a disease-specific instrument into a preference-based measure using regression modeling [<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B16">16</xref>]. In recent years, there has been growing interest in mapping in the literature, with a number of publications that have mapped responses on disease-specific instruments to generic preference-based measures [<xref ref-type="bibr" rid="B17">17</xref>-<xref ref-type="bibr" rid="B26">26</xref>]; these studies are further described in two recent review articles [<xref ref-type="bibr" rid="B15">15</xref>-<xref ref-type="bibr" rid="B27">27</xref>]. In the area of oncology, we have identified 13 studies that have mapped responses from cancer-specific instruments to yield utilities [<xref ref-type="bibr" rid="B28">28</xref>].</p><p>Our review revealed that only three studies mapped FACT-G responses to preference-based instruments. Two of these studies used responses from patients with a single tumor type (e.g., colorectal [<xref ref-type="bibr" rid="B29">29</xref>], prostate [<xref ref-type="bibr" rid="B30">30</xref>]; whereas, the remaining study performed the mapping exercise using a general cancer population with the EQ-5D as the target measure [<xref ref-type="bibr" rid="B22">22</xref>]. Given that the HRQoL of patients with different types of cancer (and different stages of disease) can vary considerably, it is possible that the results of mapping exercises may differ depending on the type of cancer patients included in the study. As such, further investigation is needed to explore whether a more appropriate mapping function can be developed using responses from patients with different tumor types compared to a specific population. The resulting algorithm may better facilitate future CUAs because it may be more applicable in general cancer populations.</p><p>Thus, the objective of this study is to develop a mapping function to impute both EQ-5D and SF-6D health utility values from responses on the FACT-G using a sample of patients with cancer from three different sites (breast, colorectal, and lung) with a range of disease severity. The influence of the tumor sites, as well as disease severity, was investigated.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Study population</title><p>To participate in the study, patients had to meet the following criteria: be diagnosed with either breast, colorectal, or lung cancer; be 18 years and older; be able to speak and read English; have a life expectancy of at least six months; be without cognitive impairments; and have plans to return to an appointment with a medical oncologist. Breast, colorectal, and lung cancer were chosen as they are among the most common cancers diagnosed in British Columbia and Canada [<xref ref-type="bibr" rid="B31">31</xref>]. Recruitment and informed consent were undertaken by a medical oncologist. Consented patients were given the HRQoL and socio-demographic questionnaires at a subsequent outpatient attendance at the Vancouver Cancer Clinic. Data collection occurred between June 2008 and December 2009.</p><p>Two options were available to the patients when completing the study. The first option was that the questionnaires be completed face-to-face with a trained research assistant at the patient’s outpatient visit. Alternatively, the patients could take the questionnaires home and return the completed questionnaires in the post in a provided pre-paid envelope. For both options, researchers were available to answer questions if needed. The order of the HRQoL questionnaires was randomized for each participant. The study protocol was approved by the Research Ethics Board of the British Columbia Cancer Agency.</p></sec><sec><title>Health-related quality of life instruments</title><p>The FACT-G, EQ-5D and SF-6D were used to elicit information pertaining to HRQoL. Patients also completed a brief socio-demographic questionnaire to obtain information regarding age, sex, marital status, qualification, and ethnicity. Clinical data were obtained from the medical records of patients.</p><p>The fourth version of FACT-G consists of 27 Likert-type questions covering four domains: physical well-being (PWB, 7 items), social/family well-being (SWB, 7 items), emotional well-being (EWB, 6 items), and functional well-being (FWB, 7 items) [<xref ref-type="bibr" rid="B11">11</xref>]. Summary scores can be calculated for each of these four domains, alongside a single overall score for the instrument.</p><p>The EQ-5D consists of a general health descriptive system based on five dimensions and a 100-point visual analogue scale (VAS) [<xref ref-type="bibr" rid="B32">32</xref>]. The dimensions cover mobility, self care, usual activities, pain/discomfort, and anxiety/depression and are characterized by three levels (i.e., no problems, some problems and extreme problems). The instrument can be used to describe 243 possible health states, which are assigned utilities based on country-specific algorithms. The most widely used utility algorithm was based on a time trade-off (TTO) survey of 2997 UK respondents [<xref ref-type="bibr" rid="B13">13</xref>]. Recently, Shaw et al. developed a utility algorithm based on a TTO survey of 4048 US residents [<xref ref-type="bibr" rid="B33">33</xref>], which in lieu of a Canadian algorithm, was used to calculate EQ-5D utilities for this study.</p><p>The SF-6D was constructed from a sample of 11 items selected from the Medical Outcomes Study Short Form 36 and has been valued by a representative sample of the UK general population using the standard gamble (SG) valuation technique [<xref ref-type="bibr" rid="B34">34</xref>]. The SF-6D is based on a six-dimensional health state classifications that assesses physical functioning, role limitations due to physical health problems, bodily pain, general health, vitality, social functioning, role limitations due to emotional problems, and mental health [<xref ref-type="bibr" rid="B35">35</xref>]. Each dimension of the SF-6D has four to six levels and can be used to describe 18,000 health states. In the absence of a Canadian utility algorithm, the UK algorithm was used to calculate the SF-6D preference weights for this study.</p></sec><sec><title>Statistical analyses</title><p>The study population was randomly sub-divided into two samples: development and cross-validation. The developmental dataset was used to construct the mapping function, while the latter sample validated the developed algorithm. Demographic and clinical characteristics of the patients assigned to the development and cross-validation samples were compared using chi-squared and t-tests for categorical and continuous variables, respectively. Then Spearman correlation coefficients were calculated to assess the linear relationship between the FACT-G and the two preference-based instruments. We also tested the skewness of the EQ-5D and SF-6D utility scores.</p><p>Ordinary least squares (OLS), generalized linear model (GLM), censored least absolute deviation (CLAD), and the random effects model have been used to derived utilities from preference-based instruments [<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B16">16</xref>]. While most previous mapping studies used the OLS model [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B18">18</xref>-<xref ref-type="bibr" rid="B41">41</xref>], this approach may not be appropriate when the preference-based scores are highly skewed. (EQ-5D responses are positively skewed for this current study; this is shown in Figure <xref ref-type="fig" rid="F1">1</xref>). The presence of ceiling effects can lead to inconsistent estimates of the coefficients of independent variables. In the absence of appropriate remedial measures such as suitable transformations, the use of OLS in the presence of heteroscedasticity and non-normality may also be problematic [<xref ref-type="bibr" rid="B15">15</xref>].</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Distribution of mean SF-</bold><bold>6D, </bold><bold>EQ-</bold><bold>5D and FACT-</bold><bold>G scores.</bold></p></caption><graphic xlink:href="1477-7525-11-203-1"/></fig><p>In this study, we started from the OLS (due to its prevalent use) and calculated robust standard errors that produced consistent estimates in the presence of heteroscedasticity. We extended this to the GLM, which relaxes the assumption of the OLS, to assess whether this approach produced more accurate predictions than the OLS. Finally, we used the CLAD model to account for the ceiling effect. This approach calculates appropriate estimates of the standard error using bootstrapping techniques [<xref ref-type="bibr" rid="B42">42</xref>]. Added advantages of the CLAD estimator are its robustness to heteroscedasticity and its consistency. It is also asymptotically normal for a wide class of error distributions and can be used to model data with skewed distributions [<xref ref-type="bibr" rid="B42">42</xref>,<xref ref-type="bibr" rid="B43">43</xref>]. CLAD is a form of median regression that minimizes the sum of absolute residuals, and as such is not as sensitive to deviations from normality and homoscedasticity [<xref ref-type="bibr" rid="B42">42</xref>].</p><p>The three regression approaches (i.e., OLS, GLM, and CLAD) were ran for each of the models described below. The approach was to start with the simplest model and then work through more complicated specifications.</p></sec><sec><title>Model 1</title><p>The EQ-5D and SF-6D utility indices were each regressed on the FACT-G overall score. The overall score of the FACT-G was rescaled onto a 0-100 scale to enable easier interpretation of the results.</p></sec><sec><title>Model 2</title><p>The EQ-5D and SF-6D utility indices were each regressed on the FACT-G domain scores, which were rescaled onto a 0-100 scale.</p></sec><sec><title>Model 3</title><p>Demographic (i.e., age and sex) and clinical (i.e., stage of disease) characteristics were introduced to assess whether these variables improved the predictions of the preference-based utilities. Interaction effects (i.e., PWB x FWB, PWB x EWB, age squared, and age x sex) were also tested to examine for any non-linear relationships. However, the interaction effects did not improve the model fit and were not presented. The effect of disease stage and cancer type was explored by including dummy variables in the regression models; the results from this model are discussed in greater detail.</p><p>The performances of the models were assessed in terms of how well the responses to the FACT-G predicted the EQ-5D and SF-6D utilities. The adjusted R<sup>2</sup> describes how well the model explains the dataset it was estimated on; the higher the R<sup>2</sup>, the better the model explains the dataset. The mean absolute error (MAE) and root mean squared error (RMSE) examine the difference between the observed and predicted values, as well as provide an indication of the size of the prediction errors; the lower the values, the better the model is performing [<xref ref-type="bibr" rid="B44">44</xref>].</p><p>The distributions of the observed and predicted utility scores were evaluated using a Wilcoxon matched-pairs signed-ranks test. The performance of the preferred regression models were compared in terms of their means, standard deviations (SDs), and the 10<sup>th</sup>, 50<sup>th</sup>, and 90<sup>th</sup> percentiles of the observed and predicted values across two frequently used measures of disease severity: cancer stage (ranging from 1-4) and the Eastern Cooperative Oncology Group (ECOG) performance status (ranging from 0 to 3). For all statistical analyses, STATA version 11.1 was used [<xref ref-type="bibr" rid="B45">45</xref>].</p></sec></sec><sec sec-type="results"><title>Results</title><p>HRQoL information was obtained from 367 patients with breast (n = 140, 38%), colorectal (n = 113, 31%), and lung (n = 114, 31%) cancer. Most of the questionnaires were completed face-to-face (n = 352, 96%). The average age of participants was 58.7 years (SD 11.5 years); 67% were women (Table <xref ref-type="table" rid="T1">1</xref>). The patients reported their ethnic compositions as British/Irish (n = 164, 46%), Chinese (n = 53, 15%), or Other (n = 143, 40%). A good representation of disease severity, in terms of cancer stage and ECOG score, was observed in the study sample. There was no statistically significant differences in the demographic and clinical characteristics of the patients in the development (n = 184) and cross-validation samples (n = 183). The majority of patients answered all the items on the HRQoL instruments. The EQ-5D utility indices were negatively skewed, while the SF-6D utilities were mildly skewed; a ceiling effect was observed for the EQ-5D responses (23.4%) (Table <xref ref-type="table" rid="T2">2</xref>). The FACT-G score was highly correlated with the EQ-5D (rho = 0.649) and the SF-6D (rho = 0.714) (Table <xref ref-type="table" rid="T3">3</xref>).</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p><bold>Socio</bold>-<bold>demographic and clinical characteristics of the patients</bold></p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="left"> </th><th align="center"><bold>All </bold><bold>(n = 367) </bold><bold>count (%)</bold></th><th align="center"><bold>Development </bold><bold>(n = 184) </bold><bold>count (%)</bold></th><th align="center"><bold>Validation </bold><bold>(n = 183) </bold><bold>count (%)</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Female<hr/></td><td align="center" valign="bottom">245 (67%)<hr/></td><td align="center" valign="bottom">68%<hr/></td><td align="center" valign="bottom">69%<hr/></td></tr><tr><td align="left" valign="bottom">Mean age (sd)<hr/></td><td align="center" valign="bottom">58.7 (± 11.5)<hr/></td><td align="center" valign="bottom">58.6 (12.5)<hr/></td><td align="center" valign="bottom">58.6 (11.6)<hr/></td></tr><tr><td align="left" valign="bottom">Tumor site<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Breast<hr/></td><td align="center" valign="bottom">140 (38%)<hr/></td><td align="center" valign="bottom">36<hr/></td><td align="center" valign="bottom">35<hr/></td></tr><tr><td align="left" valign="bottom">  Lung<hr/></td><td align="center" valign="bottom">114 (31%)<hr/></td><td align="center" valign="bottom">31<hr/></td><td align="center" valign="bottom">33<hr/></td></tr><tr><td align="left" valign="bottom">  Colorectal<hr/></td><td align="center" valign="bottom">113 (31%)<hr/></td><td align="center" valign="bottom">33<hr/></td><td align="center" valign="bottom">33<hr/></td></tr><tr><td align="left" valign="bottom">Marital status<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Married or living with partner<hr/></td><td align="center" valign="bottom">251 (70%)<hr/></td><td align="center" valign="bottom">69<hr/></td><td align="center" valign="bottom">71<hr/></td></tr><tr><td align="left" valign="bottom">  Single<hr/></td><td align="center" valign="bottom">40 (11%)<hr/></td><td align="center" valign="bottom">11<hr/></td><td align="center" valign="bottom">12<hr/></td></tr><tr><td align="left" valign="bottom">  Divorced or widowed<hr/></td><td align="center" valign="bottom">68 (19%)<hr/></td><td align="center" valign="bottom">20<hr/></td><td align="center" valign="bottom">18<hr/></td></tr><tr><td align="left" valign="bottom">Education<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Primary school completed<hr/></td><td align="center" valign="bottom">27 (8%)<hr/></td><td align="center" valign="bottom">10<hr/></td><td align="center" valign="bottom">6<hr/></td></tr><tr><td align="left" valign="bottom">  Secondary school completed<hr/></td><td align="center" valign="bottom">108 (30%)<hr/></td><td align="center" valign="bottom">29<hr/></td><td align="center" valign="bottom">27<hr/></td></tr><tr><td align="left" valign="bottom">  College or university<hr/></td><td align="center" valign="bottom">211 (59%)<hr/></td><td align="center" valign="bottom">56<hr/></td><td align="center" valign="bottom">65<hr/></td></tr><tr><td align="left" valign="bottom">  Other<hr/></td><td align="center" valign="bottom">11 (3%)<hr/></td><td align="center" valign="bottom">4<hr/></td><td align="center" valign="bottom">3<hr/></td></tr><tr><td align="left" valign="bottom">Employment status<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Full time<hr/></td><td align="center" valign="bottom">104 (29%)<hr/></td><td align="center" valign="bottom">30<hr/></td><td align="center" valign="bottom">28<hr/></td></tr><tr><td align="left" valign="bottom">  Part time<hr/></td><td align="center" valign="bottom">46 (13%)<hr/></td><td align="center" valign="bottom">12<hr/></td><td align="center" valign="bottom">14<hr/></td></tr><tr><td align="left" valign="bottom">  Working at home<hr/></td><td align="center" valign="bottom">10 (3%)<hr/></td><td align="center" valign="bottom">2<hr/></td><td align="center" valign="bottom">2<hr/></td></tr><tr><td align="left" valign="bottom">  Retired<hr/></td><td align="center" valign="bottom">133 (37%)<hr/></td><td align="center" valign="bottom">39<hr/></td><td align="center" valign="bottom">37<hr/></td></tr><tr><td align="left" valign="bottom">  Unable to work<hr/></td><td align="center" valign="bottom">68 (19%)<hr/></td><td align="center" valign="bottom">17<hr/></td><td align="center" valign="bottom">18<hr/></td></tr><tr><td align="left" valign="bottom">Ethnicity<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  British/Irish<hr/></td><td align="center" valign="bottom">164 (46%)<hr/></td><td align="center" valign="bottom">42<hr/></td><td align="center" valign="bottom">45<hr/></td></tr><tr><td align="left" valign="bottom">  Chinese<hr/></td><td align="center" valign="bottom">53 (15%)<hr/></td><td align="center" valign="bottom">15<hr/></td><td align="center" valign="bottom">16<hr/></td></tr><tr><td align="left" valign="bottom">  Other<hr/></td><td align="center" valign="bottom">143 (40%)<hr/></td><td align="center" valign="bottom">44<hr/></td><td align="center" valign="bottom">38<hr/></td></tr><tr><td align="left" valign="bottom">Disease stage<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Stage 1<hr/></td><td align="center" valign="bottom">39 (11%)<hr/></td><td align="center" valign="bottom">14<hr/></td><td align="center" valign="bottom">11<hr/></td></tr><tr><td align="left" valign="bottom">  Stage 2<hr/></td><td align="center" valign="bottom">54 (15%)<hr/></td><td align="center" valign="bottom">17<hr/></td><td align="center" valign="bottom">13<hr/></td></tr><tr><td align="left" valign="bottom">  Stage 3<hr/></td><td align="center" valign="bottom">87 (24%)<hr/></td><td align="center" valign="bottom">21<hr/></td><td align="center" valign="bottom">25<hr/></td></tr><tr><td align="left" valign="bottom">  Stage 4<hr/></td><td align="center" valign="bottom">178 (50%)<hr/></td><td align="center" valign="bottom">48<hr/></td><td align="center" valign="bottom">51<hr/></td></tr><tr><td align="left" valign="bottom">ECOG<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  0<hr/></td><td align="center" valign="bottom">123 (35)<hr/></td><td align="center" valign="bottom">36<hr/></td><td align="center" valign="bottom">34<hr/></td></tr><tr><td align="left" valign="bottom">  1<hr/></td><td align="center" valign="bottom">177 (50)<hr/></td><td align="center" valign="bottom">50<hr/></td><td align="center" valign="bottom">51<hr/></td></tr><tr><td align="left" valign="bottom">  2<hr/></td><td align="center" valign="bottom">39 (11)<hr/></td><td align="center" valign="bottom">12<hr/></td><td align="center" valign="bottom">10<hr/></td></tr><tr><td align="left">  3</td><td align="center">12 (3)</td><td align="center">3</td><td align="center">3</td></tr></tbody></table></table-wrap><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Descriptive statistics for instruments used in this study</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th rowspan="2" align="left" valign="top"><bold>Statistics</bold></th><th colspan="2" align="center" valign="bottom"><bold>Preference-</bold><bold>based measures</bold><hr/></th><th align="center" valign="bottom"><bold>Cancer-</bold><bold>specific instrument</bold><hr/></th></tr><tr><th align="center"><bold>EQ-</bold><bold>5D</bold></th><th align="center"><bold>SF-</bold><bold>6D</bold></th><th align="center"><bold>FACT-</bold><bold>G</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">N<hr/></td><td align="center" valign="bottom">363<hr/></td><td align="center" valign="bottom">364<hr/></td><td align="center" valign="bottom">365<hr/></td></tr><tr><td align="left" valign="bottom">Mean<hr/></td><td align="center" valign="bottom">0.82<hr/></td><td align="center" valign="bottom">0.71<hr/></td><td align="center" valign="bottom">78.87<hr/></td></tr><tr><td align="left" valign="bottom">Standard deviation<hr/></td><td align="center" valign="bottom">0.14<hr/></td><td align="center" valign="bottom">0.11<hr/></td><td align="center" valign="bottom">15.47<hr/></td></tr><tr><td align="left" valign="bottom">Median<hr/></td><td align="center" valign="bottom">0.83<hr/></td><td align="center" valign="bottom">0.70<hr/></td><td align="center" valign="bottom">81.00<hr/></td></tr><tr><td align="left" valign="bottom">Minimum<hr/></td><td align="center" valign="bottom">0.11<hr/></td><td align="center" valign="bottom">0.44<hr/></td><td align="center" valign="bottom">36.00<hr/></td></tr><tr><td align="left" valign="bottom">Maximum<hr/></td><td align="center" valign="bottom">1.00<hr/></td><td align="center" valign="bottom">1.00<hr/></td><td align="center" valign="bottom">107<hr/></td></tr><tr><td align="left" valign="bottom">Flooring (%)<hr/></td><td align="center" valign="bottom">0.28<hr/></td><td align="center" valign="bottom">0.27<hr/></td><td align="center" valign="bottom">0.55<hr/></td></tr><tr><td align="left">Ceiling (%)</td><td align="center">23.4</td><td align="center">1.37</td><td align="center">0.60</td></tr></tbody></table></table-wrap><table-wrap position="float" id="T3"><label>Table 3</label><caption><p><bold>Spearman correlations between the preference</bold>-<bold>based measures and the FACT</bold>-<bold>G domains</bold></p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th rowspan="2" align="left" valign="top"> </th><th colspan="5" align="center" valign="bottom"><bold>FACT-</bold><bold>G</bold><hr/></th></tr><tr><th align="center"><bold>Global</bold></th><th align="center"><bold>PWB</bold></th><th align="center"><bold>FWB</bold></th><th align="center"><bold>EWB</bold></th><th align="center"><bold>SWB</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom"><bold>EQ</bold>-<bold>5D</bold><hr/></td><td align="center" valign="bottom">0.649<hr/></td><td align="center" valign="bottom">0.631<hr/></td><td align="center" valign="bottom">0.599<hr/></td><td align="center" valign="bottom">0.429<hr/></td><td align="center" valign="bottom">0.222<hr/></td></tr><tr><td align="left"><bold>SF</bold>-<bold>6D</bold></td><td align="center">0.714</td><td align="center">0.753</td><td align="center">0.678</td><td align="center">0.386</td><td align="center">0.205</td></tr></tbody></table><table-wrap-foot><p>p < 0.05.</p><p><italic>PWB</italic>: physical well-being; <italic>FWB</italic>: functional well-being; <italic>EWB</italic>: emotional well-being; <italic>SWB</italic>: social well-being.</p></table-wrap-foot></table-wrap><sec><title>Regression models</title><sec><title><bold> <italic>Mapping FACT-G onto EQ-5D</italic> </bold></title><p>Model 1: The three regression models (OLS, GLM, and CLAD) showed that the overall FACT-G global score significantly predicted the EQ-5D score (OLS-adjusted R<sup>2</sup> = 0.331, RMSE = 0.099, MAE = 0.078) (Table <xref ref-type="table" rid="T4">4</xref>A, Model 1 for OLS, GLM and CLAD).</p><table-wrap position="float" id="T4"><label>Table 4</label><caption><p><bold>Regression of the EQ</bold>-<bold>5D and SF</bold>-<bold>6D utility indices upon FACT</bold>-<bold>G</bold></p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="right"/><col align="right"/><col align="right"/><col align="right"/><col align="right"/><col align="right"/><col align="right"/><col align="right"/><col align="right"/></colgroup><thead valign="top"><tr><th rowspan="2" align="left" valign="top"> </th><th colspan="3" align="center" valign="bottom"><bold>OLS model</bold><hr/></th><th colspan="3" align="center" valign="bottom"><bold>GLM model</bold><hr/></th><th colspan="3" align="center" valign="bottom"><bold>CLAD model</bold><hr/></th></tr><tr><th align="right"><bold>1</bold></th><th align="right"><bold>2</bold></th><th align="right"><bold>3</bold></th><th align="right"><bold>1</bold></th><th align="right"><bold>2</bold></th><th align="right"><bold>3</bold></th><th align="right"><bold>1</bold></th><th align="right"><bold>2</bold></th><th align="right"><bold>3</bold></th></tr></thead><tbody valign="top"><tr><td colspan="10" align="left" valign="bottom"><bold>EQ</bold>-<bold>5D</bold><hr/></td></tr><tr><td align="left" valign="bottom">FACT-G<hr/></td><td align="right" valign="bottom">0.006<sup>§</sup><hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.008<sup>§</sup><hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.005<sup>§</sup><hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">PWB<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.009<sup>§</sup><hr/></td><td align="right" valign="bottom">0.010<sup>§</sup><hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.012<sup>§</sup><hr/></td><td align="right" valign="bottom">0.013<sup>§</sup><hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.006<sup>§</sup><hr/></td><td align="right" valign="bottom">0.012<sup>§</sup><hr/></td></tr><tr><td align="left" valign="bottom">FWB<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.008<sup>§</sup><hr/></td><td align="right" valign="bottom">0.006<sup>‡</sup><hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.010<sup>§</sup><hr/></td><td align="right" valign="bottom">0.007<sup>‡</sup><hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.007<sup>§</sup><hr/></td><td align="right" valign="bottom">0.005<sup>§</sup><hr/></td></tr><tr><td align="left" valign="bottom">EWB<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.005<sup>†</sup><hr/></td><td align="right" valign="bottom">0.006<sup>†</sup><hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.006<sup>†</sup><hr/></td><td align="right" valign="bottom">0.008<sup>†</sup><hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.005<sup>‡</sup><hr/></td><td align="right" valign="bottom">0.002<sup>§</sup><hr/></td></tr><tr><td align="left" valign="bottom">Colorectal<sup>1</sup><hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.018<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.019<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.012<hr/></td></tr><tr><td align="left" valign="bottom">Lung<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">-0.022<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">-0.023<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.019<hr/></td></tr><tr><td align="left" valign="bottom">Stage-2<sup>2</sup><hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">-0.018<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">-0.021<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">-0.030<hr/></td></tr><tr><td align="left" valign="bottom">Stage-3<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.019<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.025<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.016<hr/></td></tr><tr><td align="left" valign="bottom">Stage-4<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.026<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.034<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">-0.009<hr/></td></tr><tr><td align="left" valign="bottom">Constant<hr/></td><td align="right" valign="bottom">0.345<hr/></td><td align="right" valign="bottom">0.391<hr/></td><td align="right" valign="bottom">0.267<hr/></td><td align="right" valign="bottom">-0.798<hr/></td><td align="right" valign="bottom">-0.746<hr/></td><td align="right" valign="bottom">-0.867<hr/></td><td align="right" valign="bottom">0.380<hr/></td><td align="right" valign="bottom">0.481<hr/></td><td align="right" valign="bottom">0.281<hr/></td></tr><tr><td align="char" valign="bottom">Adjusted R<sup>2</sup><hr/></td><td align="right" valign="bottom">0.331<hr/></td><td align="right" valign="bottom">0.396<hr/></td><td align="right" valign="bottom">0.469<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">LL<hr/></td><td align="right" valign="bottom">125.736<hr/></td><td align="right" valign="bottom">136.033<hr/></td><td align="right" valign="bottom">144.088<hr/></td><td align="right" valign="bottom">125.924<hr/></td><td align="right" valign="bottom">136.068<hr/></td><td align="right" valign="bottom">143.054<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">RMSE<hr/></td><td align="right" valign="bottom">0.120<hr/></td><td align="right" valign="bottom">0.113<hr/></td><td align="right" valign="bottom">0.104<hr/></td><td align="right" valign="bottom">0.120<hr/></td><td align="right" valign="bottom">0.113<hr/></td><td align="right" valign="bottom">0.111<hr/></td><td align="right" valign="bottom">0.122<hr/></td><td align="right" valign="bottom">0.116<hr/></td><td align="right" valign="bottom">0.095<hr/></td></tr><tr><td align="left" valign="bottom">MAE<hr/></td><td align="right" valign="bottom">0.090<hr/></td><td align="right" valign="bottom">0.086<hr/></td><td align="right" valign="bottom">0.077<hr/></td><td align="right" valign="bottom">0.089<hr/></td><td align="right" valign="bottom">0.085<hr/></td><td align="right" valign="bottom">0.077<hr/></td><td align="right" valign="bottom">0.924<hr/></td><td align="right" valign="bottom">0.086<hr/></td><td align="right" valign="bottom">0.078<hr/></td></tr><tr><td align="left" valign="bottom">N<hr/></td><td align="right" valign="bottom">180<hr/></td><td align="right" valign="bottom">180<hr/></td><td align="right" valign="bottom">162<hr/></td><td align="right" valign="bottom">180<hr/></td><td align="right" valign="bottom">180<hr/></td><td align="right" valign="bottom">162<hr/></td><td align="right" valign="bottom">180<hr/></td><td align="right" valign="bottom">180<hr/></td><td align="right" valign="bottom">164<hr/></td></tr><tr><td colspan="10" align="right" valign="bottom"><bold>SF</bold>-<bold>6D</bold><hr/></td></tr><tr><td align="left" valign="bottom">FACT-G<hr/></td><td align="right" valign="bottom">0.005<sup>§</sup><hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.008<sup>§</sup><hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.005<sup>§</sup><hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">PWB<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.011<sup>§</sup><hr/></td><td align="right" valign="bottom">0.011<sup>§</sup><hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.016<sup>§</sup><hr/></td><td align="right" valign="bottom">0.0159<sup>§</sup><hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.009<sup>§</sup><hr/></td><td align="right" valign="bottom">0.012<sup>§</sup><hr/></td></tr><tr><td align="left" valign="bottom">FWB<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.007<sup>§</sup><hr/></td><td align="right" valign="bottom">0.007<sup>§</sup><hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.011<sup>§</sup><hr/></td><td align="right" valign="bottom">0.0096<sup>§</sup><hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.008<sup>§</sup><hr/></td><td align="right" valign="bottom">0.007<sup>§</sup><hr/></td></tr><tr><td align="left" valign="bottom">EWB<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.001<hr/></td><td align="right" valign="bottom">0.002<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.001<hr/></td><td align="right" valign="bottom">0.003<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.005<sup>‡</sup><hr/></td><td align="right" valign="bottom">-0.004<sup>§</sup><hr/></td></tr><tr><td align="left" valign="bottom">Colorectal<sup>1</sup><hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.006<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.002<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">-0.027<hr/></td></tr><tr><td align="left" valign="bottom">Lung<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">-0.009<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">-0.016<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">-0.050<hr/></td></tr><tr><td align="left" valign="bottom">Stage-2<sup>2</sup><hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">-0.015<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">-0.023<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">-0.031<hr/></td></tr><tr><td align="left" valign="bottom">Stage-3<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.028<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.039<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">0.016<hr/></td></tr><tr><td align="left" valign="bottom">Stage-4<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">-0.006<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">-0.005<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom">-0.022<hr/></td></tr><tr><td align="left" valign="bottom">Constant<hr/></td><td align="right" valign="bottom">0.278<hr/></td><td align="right" valign="bottom">0.343<hr/></td><td align="right" valign="bottom">0.281<hr/></td><td align="right" valign="bottom">-0.982<hr/></td><td align="right" valign="bottom">-0.895<hr/></td><td align="right" valign="bottom">-0.956<hr/></td><td align="right" valign="bottom">0.331<hr/></td><td align="right" valign="bottom">0.282<hr/></td><td align="right" valign="bottom">0.438<hr/></td></tr><tr><td align="left" valign="bottom">Adj_R^2<hr/></td><td align="right" valign="bottom">0.508<hr/></td><td align="right" valign="bottom">0.628<hr/></td><td align="right" valign="bottom">0.651<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">LL<hr/></td><td align="right" valign="bottom">207.344<hr/></td><td align="right" valign="bottom">233.806<hr/></td><td align="right" valign="bottom">222.229<hr/></td><td align="right" valign="bottom">209.645<hr/></td><td align="right" valign="bottom">237.754<hr/></td><td align="right" valign="bottom">224.711<hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td><td align="right" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">RMSE<hr/></td><td align="right" valign="bottom">0.077<hr/></td><td align="right" valign="bottom">0.066<hr/></td><td align="right" valign="bottom">0.062<hr/></td><td align="right" valign="bottom">0.076<hr/></td><td align="right" valign="bottom">0.065<hr/></td><td align="right" valign="bottom">0.061<hr/></td><td align="right" valign="bottom">0.077<hr/></td><td align="right" valign="bottom">0.069<hr/></td><td align="right" valign="bottom">0.076<hr/></td></tr><tr><td align="left" valign="bottom">MAE<hr/></td><td align="right" valign="bottom">0.062<hr/></td><td align="right" valign="bottom">0.056<hr/></td><td align="right" valign="bottom">0.051<hr/></td><td align="right" valign="bottom">0.061<hr/></td><td align="right" valign="bottom">0.054<hr/></td><td align="right" valign="bottom">0.050<hr/></td><td align="right" valign="bottom">0.062<hr/></td><td align="right" valign="bottom">0.056<hr/></td><td align="right" valign="bottom">0.057<hr/></td></tr><tr><td align="left">N</td><td align="right">182</td><td align="right">182</td><td align="right">164</td><td align="right">182</td><td align="right">182</td><td align="right">164</td><td align="right">182</td><td align="right">181</td><td align="right">166</td></tr></tbody></table><table-wrap-foot><p>*Model 3 -adjusted for age and sex Reference group – <sup>1</sup>Breast Cancer; <sup>2</sup>Cancer Stage 1; <sup>†</sup>p < 0.05, <sup>‡</sup>p < 0.01, <sup>§</sup>p < 0.001; <italic>LL</italic> = Log-likelihood; <italic>MAE</italic> = Mean Absolute Error; <italic>RMSE</italic> = Root Mean Square Error. <italic>FACT</italic>-<italic>G</italic>: Functional Assessment of Cancer Therapy; <italic>PWB</italic>: Physical well-being; <italic>FWB</italic>: Functional well-being; <italic>EWB</italic>: Social well-being; <italic>EWB</italic>: Emotional well-being.</p></table-wrap-foot></table-wrap><p>Model 2: The PWB, FWB, and EWB domains of the FACT-G significantly predicted the EQ-5D utility indices using the three regression models. Coefficients for the PWB and FWB were highly significant (p < 0.001) compared to the EWB (p < 0.05) (Table <xref ref-type="table" rid="T4">4</xref>A, OLS Model 2).</p><p>Model 3: After adjusting for the demographic and clinical characteristics of the patients, the three sub-scale scores remained significant predictors of the EQ-5D index score and improved the data fit as explained by higher adjusted R<sup>2</sup> = 0.469, and lower RMSE = 0.104. In terms of accuracy of prediction, the CLAD model (RMSE = 0.095) performed better compared to the OLS (RMSE = 0.104) and GLM (RMSE = 0.111) (Table <xref ref-type="table" rid="T4">4</xref>-A, CLAD Model 3). Adjusting for the type of cancer and disease stage improved the prediction of the model as demonstrated by higher R<sup>2</sup> (Table <xref ref-type="table" rid="T4">4</xref>, OLS Model-3).</p></sec><sec><title><bold> <italic>Mapping FACT-G onto SF-6D</italic> </bold></title><p>Model 1: A positive and significant correlation was detected between the overall FACT-G global score and the SF-6D utility score using the OLS (adjusted R<sup>2</sup> = 0.508, RMSE = 0.077, MAE = 0.051). Similar result was found when we used the GLM and CLAD models (Table <xref ref-type="table" rid="T4">4</xref>, Model 1).</p><p><italic>Model 2</italic>: The PWB and FWB domains of the FACT-G significantly predicted the SF-6D utility indices (PWB, β = 0.011, p < 0.001; FWB, β = 0.007, p < 0.001) (Table <xref ref-type="table" rid="T4">4</xref>, OLS Model 2). The relatively weak correlation between EWB and the SF-6D utility index could be that the emotional aspect is already captured by the other two domains. A significant increase in the amount of variations explained by the sub-scale scores as compared to the overall FACT-G score was observed (increase in the proportion of the total variance explained from adjusted R<sup>2</sup> = 0.508 to 0.628). Results from the GLM model and the CLAD do not vary much from the OLS model (Table <xref ref-type="table" rid="T4">4</xref>, GLM Model; CLAD Model).</p><p>Model 3: This model has a higher adjusted R<sup>2</sup> (0.651) when compared to OLS Models 1 and 2 (Table <xref ref-type="table" rid="T4">4</xref>-B). Adjusting for patient characteristics (age and sex) and clinical characteristics (stage of disease in terms of cancer stage) improved the model as demonstrated by a higher adjusted R<sup>2</sup> and lower RMSE (0.062). Similar results were observed for the GLM and CLAD models. In terms of accuracy of prediction as measured in RMSE, the GLM model (0.061) performed better compared to the CLAD (0.071) and the OLS (0.062) models (Table <xref ref-type="table" rid="T4">4</xref>, GLM Model 3). We did not find statistically significant differences in the SF-6D utility index by disease stage and cancer type.</p></sec></sec><sec><title>Comparison of observed and predicted utilities</title><p>When comparing predictions using the three different regression methods, the CLAD model better predicted the utility scores more closely to the observed values when compared to the predictions obtained using the OLS and GLM models (Table <xref ref-type="table" rid="T5">5</xref>). The SF-6D utilities based on GLM followed the observed values more closely. The Wilcoxon matched-pairs signed-ranks test showed no statistically significant differences between the distribution of the observed utility indices and the values predicted by the OLS, GLM, and CLAD models (all p > 0.05). All three models tended to over predict the lower (10<sup>th</sup> percentile) utility scores and under predict the high (upper 90<sup>th</sup> percentile) scores (Table <xref ref-type="table" rid="T2">2</xref>). Accuracy of predictions across the disease stage measures used in this study (i.e., cancer stage and ECOG performance status) is presented graphically (Figures <xref ref-type="fig" rid="F2">2</xref>-A and <xref ref-type="fig" rid="F2">2</xref>-B).</p><table-wrap position="float" id="T5"><label>Table 5</label><caption><p><bold>Descriptive summary of utility indices derived from observed SF</bold>-<bold>6D</bold>/<bold>EQ</bold>-<bold>5D and OLS</bold>/<bold>GLM</bold>/<bold>CLAD regression models</bold></p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="left"> </th><th align="center"><bold>Mean</bold></th><th align="center"><bold>SD</bold></th><th align="center"><bold>Minimum</bold></th><th colspan="3" align="center"><bold>Percentiles</bold></th><th align="center"><bold>Maximum</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"><bold>10</bold><sup>%</sup><hr/></td><td align="center" valign="bottom"><bold>50</bold>%<hr/></td><td align="center" valign="bottom"><bold>90</bold><sup>%</sup><hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">EQ-5D<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Observed<hr/></td><td align="center" valign="bottom">0.823<hr/></td><td align="center" valign="bottom">0.149<hr/></td><td align="center" valign="bottom">0.358<hr/></td><td align="center" valign="bottom">0.768<hr/></td><td align="center" valign="bottom">0.827<hr/></td><td align="center" valign="bottom">1.000<hr/></td><td align="center" valign="bottom">1.000<hr/></td></tr><tr><td align="left" valign="bottom">  OLS<hr/></td><td align="center" valign="bottom">0.831<hr/></td><td align="center" valign="bottom">0.103<hr/></td><td align="center" valign="bottom">0.557<hr/></td><td align="center" valign="bottom">0.75<hr/></td><td align="center" valign="bottom">0.84<hr/></td><td align="center" valign="bottom">0.918<hr/></td><td align="center" valign="bottom">0.951<hr/></td></tr><tr><td align="left" valign="bottom">  GLM<hr/></td><td align="center" valign="bottom">0.832<hr/></td><td align="center" valign="bottom">0.102<hr/></td><td align="center" valign="bottom">0.583<hr/></td><td align="center" valign="bottom">0.755<hr/></td><td align="center" valign="bottom">0.836<hr/></td><td align="center" valign="bottom">0.921<hr/></td><td align="center" valign="bottom">0.956<hr/></td></tr><tr><td align="left" valign="bottom">  CLAD<hr/></td><td align="center" valign="bottom">0.828<hr/></td><td align="center" valign="bottom">0.106<hr/></td><td align="center" valign="bottom">0.511<hr/></td><td align="center" valign="bottom">0.761<hr/></td><td align="center" valign="bottom">0.831<hr/></td><td align="center" valign="bottom">0.905<hr/></td><td align="center" valign="bottom">0.972<hr/></td></tr><tr><td align="left" valign="bottom">SF-6D<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Observed<hr/></td><td align="center" valign="bottom">0.720<hr/></td><td align="center" valign="bottom">0.111<hr/></td><td align="center" valign="bottom">0.467<hr/></td><td align="center" valign="bottom">0.630<hr/></td><td align="center" valign="bottom">0.702<hr/></td><td align="center" valign="bottom">0.799<hr/></td><td align="center" valign="bottom">0.887<hr/></td></tr><tr><td align="left" valign="bottom">  OLS<hr/></td><td align="center" valign="bottom">0.723<hr/></td><td align="center" valign="bottom">0.091<hr/></td><td align="center" valign="bottom">0.452<hr/></td><td align="center" valign="bottom">0.658<hr/></td><td align="center" valign="bottom">0.734<hr/></td><td align="center" valign="bottom">0.793<hr/></td><td align="center" valign="bottom">0.839<hr/></td></tr><tr><td align="left" valign="bottom">  GLM<hr/></td><td align="center" valign="bottom">0.723<hr/></td><td align="center" valign="bottom">0.092<hr/></td><td align="center" valign="bottom">0.476<hr/></td><td align="center" valign="bottom">0.657<hr/></td><td align="center" valign="bottom">0.730<hr/></td><td align="center" valign="bottom">0.792<hr/></td><td align="center" valign="bottom">0.845<hr/></td></tr><tr><td align="left">  CLAD</td><td align="center">0.730</td><td align="center">0.102</td><td align="center">0.422</td><td align="center">0.663</td><td align="center">0.744</td><td align="center">0.810</td><td align="center">0.852</td></tr></tbody></table></table-wrap><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Distribution of observed and predicted utility scores. (A)</bold> - by cancer stage and <bold>(B)</bold> – by ECOG.</p></caption><graphic xlink:href="1477-7525-11-203-2"/></fig></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>Our results demonstrate that the estimation of both EQ-5D and SF-6D utility indices using the FACT-G responses from breast, lung and colorectal cancer patients can be achieved. We had a relatively large sample of patients with different disease severity levels. Our findings suggested that the CLAD and GLM models are best used to predict EQ-5D and SF-6D utilities, respectively.</p><p>Comparing predictions of the preference-based scores using FACT-G scales indicated that a better performance was observed for the SF-6D (adjusted R<sup>2</sup> = 0.651, MAE = 0.052, RMSE = 0.062) as compared to the EQ-5D (adjusted R<sup>2</sup> = 0.469, MAE = 0.062, RMSE = 0.104); these results are in agreement with previous studies [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B25">25</xref>-<xref ref-type="bibr" rid="B39">39</xref>]. Such studies reported that the physical and functional subscales were significant predictors of the utility scores. For both preference-based indices, including patients’ demographic and clinical characteristics improved explanatory power of the models.</p><p>Taking the RSME criteria as measures of prediction, the GLM model using FACT-G domain scores plus patient demographic (age, sex) and clinical (cancer stage) characteristics fitted the SF-6D data well. The GLM model was chosen despite non-Gaussian residuals as they produced the best predictions of utility scores on a natural scale. On the other hand, the CLAD model fitted the EQ-5D data well as compared to the OLS and GLM and it was investigated further.</p><p>The PWB and FWB sub-scales of the FACT-G were statistically significant predictors for both preference-based instruments. The EWB reached statistical significance only in the EQ-5D model. The SWB score of the FACT-G was not significantly associated with the EQ-5D and SF-6D and therefore, was not included in the regression model. Similar approaches have been used in previous studies and the similar results have been reported [<xref ref-type="bibr" rid="B22">22</xref>]. It is likely that the other dimensions can capture most of the information of other dimensions; as a result, additional information would not be needed. It is possible that only a negligible relationship between those sub-scales and HRQoL exists, and other aspects of illness are captured by them.</p><p>For both instruments, Model 3 produced more accurate predictions than Models 1 and 2 demonstrating that including patient characteristics improves predictions. There was a strong correlation between predicted and observed utilities for all three models. Utility scores derived from the FACT-G using the GLM approach were highly correlated with the observed scores (rho = 0.831 for the SF-6D).</p><p>The purpose of the current study is to examine the differential validity and prediction of the utility scores using cancer-specific instrument from a sample of breast, lung, and colorectal cancer patients. Our results showed that this is feasible. We also explored this further by assessing differences between observed and predicted utilities by disease severity. Our findings demonstrate that there are similar patterns of prediction by cancer stage, though there is a trend of over predicting values for patients with greater disease severity. Future research is needed to discuss utilization of mixed modeling for patients with different disease severities. While we anticipated that constructing a mapping algorithm using patients with three different tumor sites would be more applicable to facilitate future CUAs in general cancer populations, we found that adjusting for the type and stage of cancer increased the prediction of the model demonstrated by higher R<sup>2</sup>.</p><p>The results from this current study are comparable to the one other study that mapped FACT-G responses to EQ-5D in a general cancer population [<xref ref-type="bibr" rid="B22">22</xref>]. Similar mean values for the EQ-5D and the FACT-G were observed in the current and previous studies (EQ-5D 0.82 versus 0.81; FACT-G 79.0 versus 81.1, respectively). For the current study, we observed a lower ceiling effect for the EQ-5D (23.4% versus 33.3%) and higher floor effect 0.28%. We observed similar goodness-of-fit values for our OLS Model 3 (R<sup>2</sup> = 0.47) compared to the previous study (R<sup>2</sup> = 0.45). In terms of accuracy, we reached same conclusion that predictions using CLAD was more accurate for the EQ-5D compared to the OLS; this was further confirmed when we ran the GLM model and found that the CLAD performs better.</p><p>We examined accuracy of predictions across the disease stage measures used in this study (cancer stage and the ECOG). Mean value predictions indicated that overall the GLM model performs better for the SF-6D and the CLAD model performs better for the EQ-5D compared to the alternate methods used in this study. All regression models tend to over predict lower observed utility scores and under predict higher scores.</p><p>The poor ability of the FACT-G to predict EQ-5D utilities may be influenced by the brevity of the preference-based instrument (i.e., three levels to define the five dimensions). Our previous work revealed that the EQ-5D was less discriminative between different levels of disease severities when compared to the FACT-G and SF-6D [<xref ref-type="bibr" rid="B46">46</xref>]. While the FACT-G, EQ-5D, and SF-6D assesses an individual’s HRQoL, the instruments are composed of different dimensions. For example, the FACT-G and SF-6D has a dimension describing vitality (i.e., an important health outcome for cancer patients), whereas, the EQ-5D does not. To adequately map disease-specific instruments to preference-based measures, a certain degree of overlap is required between the two descriptive systems [<xref ref-type="bibr" rid="B15">15</xref>]. If important dimensions are not covered in the preference-based measures, the mapping functions may be compromised. Performing the mapping exercise only using similar dimensions – for example, pain – would not permit the calculation of utilities to be used in CUAs.</p><p>The scoring functions of the preference-based instruments were derived from responses of the general population. The EQ-5D and SF-6D consist of dimensions related to individuals’ perceptions of their abilities within a social context (e.g., “I am able to perform my usual activities”). These dimensions capture the ability of the patients to adapt to their health state, which may not be a phenomenon that members of the general population are aware of when faced with an impaired health state [<xref ref-type="bibr" rid="B47">47</xref>]. The process of adaptation has been reported to enhance the difference between the utilities for health states provided by patients and members of the general population [<xref ref-type="bibr" rid="B48">48</xref>]. Lower utilities tend to be reported by the general population because they do not anticipate their ability to adapt when faced with life with an impaired health state, such as cancer [<xref ref-type="bibr" rid="B49">49</xref>]; this may be reflected by a lower magnitude of disutility in the population tariffs. While the use of general population tariffs raises concerns as to whether mapping functions can be estimated accurately for the population used in this study, generated results could be compared independently of the disease. This is one of the reasons why generic preference-based measurements of HRQoL are required for economic evaluations.</p><p>The ability of the cancer-specific instruments to predict utilities may be influenced by the psychometric properties of the preference-based measures. While responses on the FACT-G were obtained from Canadian cancer patients, the scoring functions of the EQ-5D and SF-6D are based on non-Canadian populations. These tariffs have been demonstrated that they are valid in a Canadian population [<xref ref-type="bibr" rid="B50">50</xref>-<xref ref-type="bibr" rid="B53">53</xref>]. However, the constructed mapping algorithms from this study may not be transferrable to patients of other countries. While it is anticipated that the use of a different country’s tariff may only have a marginal effect on the resulting mapping function, the utilities generated may not be appropriate for information resource allocation decisions for the country of study. As such, there is a need to explore the ability of the FACT-G to predict individual dimensions of the EQ-5D and SF-6D; the predicted dimensions could be used to calculate an overall utility score. This has considerable advantages because the derived results are not country-specific and may be more applicable for guiding resource allocation decisions, especially for country that do not have country-specific tariffs for the EQ-5D or the SF-6D.</p><p>This study demonstrates that it is possible to facilitate economic evaluations for specific health conditions when a preference-based measure has not been administered and when it would be impractical to conduct a valuation survey using methods such as SG or TTO. This mapping approach offers a shortcut for policy-makers and researchers who need utility values for use in economic evaluation but do not have access to information on preference-based measures. The methodology presented in this study may be applied to other disease-specific instruments.</p></sec><sec><title>Abbreviations</title><p>FACT-G: Functional assessment of cancer therapy - general; EQ-5D: EuroQol 5D; SF-6D: Short form 36 health survey; OLS: Ordinary least squares; GLM: Generalized linear models; CLAD: Censored least absolute deviations; MAE: Mean absolute error; RMSE: Root mean squared error; CUA: Cost-utility analysis; QALY: Quality-adjusted life years; HRQoL: Health-related quality of life; CADTH: Canadian agency for drugs and technologies in health; ECOG: Eastern cooperative oncology group; PWB: Physical well-being; SWB: Social/family well-being; EWB: Emotional well-being; FWB: Functional well-being; VAS: Visual analogue scale; TTO: Time trade-off; SG: Standard gamble.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>PT conceived and designed the study, oversaw all stages of data collection and entry, done the analysis, and drafted the manuscript. SP gave feedback on design and analysis. SP, HM, KvH, SC, BM, KG reviewed the manuscript. All authors read and approved the final manuscript.</p></sec>
Correlation of bevacizumab-induced hypertension and outcomes of metastatic colorectal cancer patients treated with bevacizumab: a systematic review and meta-analysis	<sec><title>Background</title><p>With the wide application of targeted drug therapies, the relevance of prognostic and predictive markers in patient selection has become increasingly important. Bevacizumab is commonly used in combination with chemotherapy in the treatment of metastatic colorectal cancer. However, there are currently no predictive or prognostic biomarkers for bevacizumab. Several clinical studies have evaluated bevacizumab-induced hypertension in patients with metastatic colorectal cancer. This meta-analysis was performed to better determine the association of bevacizumab-induced hypertension with outcome in patients with metastatic colorectal cancer, and to assess whether bevacizumab-induced hypertension can be used as a prognostic factor in these patients.</p></sec><sec><title>Methods</title><p>We performed a systematic review and meta-analysis on seven published studies to investigate the relationship between hypertension and outcome of patients with metastatic colorectal cancer treated with bevacizumab. Our primary endpoint was progression-free survival (PFS). Secondary endpoints were overall survival (OS) and overall response rate (ORR). Hazard ratios (HRs) for PFS and OS were extracted from each trial, and the log of the relative risk ratio (RR) was estimated for ORR.</p></sec><sec><title>Results</title><p>The occurrence of bevacizumab-induced hypertension in patients was highly associated with improvements in PFS (HR = 0.57, 95% CI: 0.46–0.72; <italic>P</italic> <0.001), OS (HR = 0.50; 95% CI: 0.37–0.68; <italic>P</italic> <0.001), and ORR (RR = 1.57, 95% CI: 1.07–2.30, <italic>P</italic> <0.05), as compared to patients without hypertension.</p></sec><sec><title>Conclusions</title><p>Bevacizumab-induced hypertension may represent a prognostic factor in patients with metastatic colorectal cancer.</p></sec>	<contrib contrib-type="author" equal-contrib="yes" id="A1"><name><surname>Cai</surname><given-names>Jun</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>junqcai@163.com</email></contrib><contrib contrib-type="author" corresp="yes" equal-contrib="yes" id="A2"><name><surname>Ma</surname><given-names>Hong</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>wudajianzhu2004@163.com</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Huang</surname><given-names>Fang</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>fangwhuang@163.com</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Zhu</surname><given-names>Dichao</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>dichaoczhu@163.com</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Bi</surname><given-names>Jianping</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>jianpingcbi@163.com</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Ke</surname><given-names>Yang</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>yangqke@163.com</email></contrib><contrib contrib-type="author" corresp="yes" id="A7"><name><surname>Zhang</surname><given-names>Tao</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>taowzhang@163.com</email></contrib>	World Journal of Surgical Oncology	<sec><title>Background</title><p>Colorectal cancer (CRC) is the fourth most common malignancy, and the second most frequent cause of cancer-related death in the United States, given that as many as 20–25% of patients have already developed metastases at initial diagnosis [<xref ref-type="bibr" rid="B1">1</xref>]. Vascular endothelial growth factor (VEGF) is the major factor involved in tumor angiogenesis [<xref ref-type="bibr" rid="B2">2</xref>]. It promotes endothelial cell survival, migration, and permeability, and stimulates the growth of blood vessels supplying the tumor. Poor prognosis and an increased relapse rate are often correlated with angiogenesis and increased blood vessel density in the primary tumor. Thus, anti-angiogenesis is a major topic of current research.</p><p>The VEGF signaling pathway is a target for cancer therapy. A recombinant humanized monoclonal antibody against VEGF, bevacizumab, has been developed to treat metastatic CRC (mCRC), breast cancer, non-squamous non-small cell lung cancer, renal cell carcinoma, ovarian cancer, glioblastoma, and metastatic melanoma [<xref ref-type="bibr" rid="B3">3</xref>-<xref ref-type="bibr" rid="B11">11</xref>]. Treatment with bevacizumab, however, is associated with various adverse reactions such as gastrointestinal perforations, wound healing complications, hemorrhage, arterial thrombotic events, infection, proteinuria, and hypertension. Nevertheless, the benefits of bevacizumab treatment may still outweigh potential adverse events [<xref ref-type="bibr" rid="B12">12</xref>]. Furthermore, bevacizumab has been demonstrated to be relatively safe in association with either irinotecan [<xref ref-type="bibr" rid="B13">13</xref>] or oxaliplatin-containing chemotherapy regimens [<xref ref-type="bibr" rid="B14">14</xref>], while its specific toxicity profile appears manageable by applying appropriate clinical selection criteria [<xref ref-type="bibr" rid="B15">15</xref>].</p><p>As mentioned above, arterial hypertension is a common side effect of bevacizumab treatment usually easily managed by standard anti-hypertensive therapy. Interestingly, many clinical trials have found that patients with mCRC treated with bevacizumab who developed hypertension had a better prognosis than those without hypertension [<xref ref-type="bibr" rid="B16">16</xref>-<xref ref-type="bibr" rid="B22">22</xref>]. These results were obtained through retrospective analysis of a relatively small dataset, but the findings are statistically significant and supported by other studies [<xref ref-type="bibr" rid="B6">6</xref>]. Throughout the course of treatment for mCRC, hypertension severity can be evaluated objectively and thus may be useful when making an early decision on whether to alter the course of disease treatment. The potential advantages of such a predictor include the ability to estimate the efficacy and activity of anti-VEGF agents in patients with mCRC.</p><p>Thus, the purpose of this study was to perform a systematic review and conduct a meta-analysis to determine if the occurrence of hypertension is a prognostic factor of response and survival for bevacizumab treatment in patients with mCRC.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Data sources</title><p>The study was performed using a pre-specified search strategy with a strict eligibility criteria. We did an extensive search of PubMed to retrieve relevant literature that reported the predictive value of hypertension regarding response and/or progression and/or survival in mCRC patients treated with bevacizumab. The search end date was January 2013, with no specified start date. Search term combinations were “b<italic>evacizumab</italic>”, “<italic>avastin</italic>”, and “<italic>hypertension</italic>” in all fields. There were no limits for language, methodological characteristics, or year of publication. All reference lists from the relevant articles and reviews were also examined for additional eligible studies. This study is approved by the Ethic Commity of Cancer Center of Union Hospital. And written informed consent was obtained from the patient for the publication of this report and any accompanying images.</p></sec><sec><title>Selection of studies</title><p>Two reviewers (JC, HM) independently carried out a literature search and examined the relevant studies for further assessment. The reference lists of all traced articles were examined manually. Citations selected from this initial search were subsequently screened for eligibility using the following criteria: i) patients with mCRC; ii) combined chemotherapy with bevacizumab, irrespective of chemotherapy used; iii) studies involving the use of other targeted agents were excluded to avoid bias related to drug interactions; iv) curative effect comparison between bevacizumab-induced hypertension arm with no hypertension arm; v) data available for analysis including the incidence of hypertension and sample size.</p></sec><sec><title>Primary and secondary outcomes</title><p>The primary outcome was progression-free survival (PFS), defined as the time between randomization and any progression or death from any cause, in relation to the severity of hypertension in patients treated with bevacizumab. Secondary endpoints were overall survival (OS), the time between randomization and any death, and overall response rate (ORR), the sum of partial and complete response rates according to the Response Evaluation Criteria in Solid Tumors [<xref ref-type="bibr" rid="B23">23</xref>] with hypertension occurrence as a predictor. Hypertension was graded according to the National Cancer Institute Common Terminology Criteria [<xref ref-type="bibr" rid="B24">24</xref>] for Adverse Events (version 3.0, 2003). Grade 1 toxicity is defined as an asymptomatic, transient increase (<24 h) greater than 20 mmHg diastolic or to greater than 150/100 mmHg. Grade 2 is recurrent or persistent (>24 h) or a symptomatic increase greater than 20 mmHg diastolic or to greater than 150/100 mmHg. Grade 3 is hypertension requiring therapy or more intensive therapy than previously provided. Grade 4 is a hypertensive crisis. Outcomes or responses were evaluated by either a comparison between no-hypertension (G0) and all grades of hypertension (G1–4), or a comparison between low-grade hypertension (G0–1) and high-grade hypertension (G2–4), depending on the data available.</p></sec><sec><title>Data extraction</title><p>Two reviewers (JC, HM) retrieved data independently and reached a consensus on all examined items. The following information was retrieved: first author, year of publication, number of patients, number of patients eligible for response, and median OS, PFS, ORR, and hazard ratio (HR). For trials included in this meta-analysis, if the log HR and its variance were not explicitly presented, the methods reported by Parmar et al. [<xref ref-type="bibr" rid="B25">25</xref>] were used to extract estimates of these statistics. In the case of any disagreement between the two reviewers, a third reviewer (DCZ) would review the data, and the results were attained by consensus. We contacted the authors of trials for the missing data when necessary. Data of study characteristics (concurrent treatment, number of patients, bevacizumab dose, and publication time) and clinical endpoints (PFS, OS, ORR) were then retrieved.</p></sec><sec><title>Data analysis and statistical methods</title><p>We calculated relative risk ratios (RRs) and confidence interval (CI) for ORR relating to hypertension severity in patients with bevacizumab-induced hypertension versus controls in the same trial. If the study reported HRs for survival in patients with G0 vs. G1 or higher-grade hypertension, then the comparison was made for the higher grade of hypertension (for example G0 vs. G2 or G3 [or G3–4]). Otherwise, if no other subgroups were reported, the comparison was performed for G0 vs. G1–4 hypertension. HRs were extracted from each trial for PFS and OS, and the log of relative RR was estimated for ORR, and 95% CIs were derived. The HR of each study was either directly collected from the original article, or calculated as suggested by Parmar [<xref ref-type="bibr" rid="B25">25</xref>] and Tierney [<xref ref-type="bibr" rid="B26">26</xref>]. The number of events (ORRs) was extracted from each study or calculated from the percentages provided.</p><p>A meta-analysis of both RRs and HRs was performed, and both fixed-effect and random-effect models were considered depending on the heterogeneity of the included studies. Statistical heterogeneity among trials included in the meta-analysis was assessed by using the Cochran Q statistic, and inconsistency was quantified with the I<sup>2</sup> statistic that estimates the percentage of total variation across studies due to heterogeneity rather than chance [<xref ref-type="bibr" rid="B27">27</xref>]. When substantial heterogeneity was not observed, the pooled estimate was calculated based on the fixed-effects model using the inverse variance method. Otherwise, the pooled estimate was calculated based on the random-effects model using the DerSimonian and Laird method [<xref ref-type="bibr" rid="B28">28</xref>].</p><p>Publication bias was evaluated using funnel plots for RR (plots of study results against precision), and with the Begg’s [<xref ref-type="bibr" rid="B29">29</xref>] and Egger’s [<xref ref-type="bibr" rid="B30">30</xref>] tests. Additionally, sensitivity analyses were performed to assess the influence of each study on overall estimate for RR by sequential removal of individual studies. A HR of less than one and a RR value of more than one meant a benefit for patients with bevacizumab-induced hypertension. A two-tailed <italic>P</italic> value <0.05 was considered statistically significant. All statistical analyses were performed using STATA version 11.0 software (STATA, College Station, TX, USA).</p></sec></sec><sec sec-type="results"><title>Results</title><p>There were 520 publications retrieved from the PubMed search. Among them, seven met the inclusion criteria for this review. The study flow diagram is shown in Figure  <xref ref-type="fig" rid="F1">1</xref>. The main characteristics of the included articles (author/year, reference, line of treatment, bevacizumab dose, number of patients, PFS, OS, ORR) are presented in Table  <xref ref-type="table" rid="T1">1</xref>. Patients were enrolled according to pre-specified eligibility criteria for each trial. Data regarding the predictive role of hypertension for PFS were available for all seven studies. Secondary outcome data, i.e., OS and ORR, were available for five studies, respectively.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p>Study flow diagram.</p></caption><graphic xlink:href="1477-7819-11-306-1"/></fig><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Characteristics of the seven selected studies</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left" valign="bottom"><bold>Author/year [Ref.]</bold><hr/></th><th align="left" valign="bottom"><bold>Line of treatment</bold><hr/></th><th align="left" valign="bottom"><bold>Bevacizumab dose</bold><hr/></th><th align="left" valign="bottom"><bold>No. of patients</bold><hr/></th><th align="left" valign="bottom"><bold>Median PFS (m)</bold><hr/></th><th align="left" valign="bottom"><bold>Median OS (m)</bold><hr/></th><th align="left" valign="bottom"><bold>ORR (%)</bold><hr/></th></tr><tr><th align="left"> </th><th align="left"> </th><th align="left"> </th><th align="left"> </th><th align="left"><bold>HTN vs. No HTN</bold></th><th align="left"><bold>HTN vs. No HTN</bold></th><th align="left"><bold>HTN vs. No HTN</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Scartozzi M/2009 [<xref ref-type="bibr" rid="B20">20</xref>]<hr/></td><td align="left" valign="bottom">First-line<hr/></td><td align="left" valign="bottom">5 mg/kg/2w<hr/></td><td align="left" valign="bottom">39<hr/></td><td align="left" valign="bottom">14.5 vs. 3.1<hr/></td><td align="left" valign="bottom">NA vs. 15.1<hr/></td><td align="left" valign="bottom">75% vs. 32%<hr/></td></tr><tr><td align="left" valign="bottom">Rebekah/2009 [<xref ref-type="bibr" rid="B21">21</xref>]<hr/></td><td align="left" valign="bottom">First-line<hr/></td><td align="left" valign="bottom">NA<hr/></td><td align="left" valign="bottom">52<hr/></td><td align="left" valign="bottom">NA vs. NA<hr/></td><td align="left" valign="bottom">NA vs. NA<hr/></td><td align="left" valign="bottom">NA vs. NA<hr/></td></tr><tr><td align="left" valign="bottom">De Stefano/2011 [<xref ref-type="bibr" rid="B19">19</xref>]<hr/></td><td align="left" valign="bottom">First-line<hr/></td><td align="left" valign="bottom">5 mg/kg/2w or 7.5 mg/kg/3w<hr/></td><td align="left" valign="bottom">74<hr/></td><td align="left" valign="bottom">15.1 vs. 8.3<hr/></td><td align="left" valign="bottom">35.5 vs. 26.7<hr/></td><td align="left" valign="bottom">84.6% vs. 42.6%<hr/></td></tr><tr><td align="left" valign="bottom">Osterlund P/2011 [<xref ref-type="bibr" rid="B17">17</xref>]<hr/></td><td align="left" valign="bottom">First- or second-line<hr/></td><td align="left" valign="bottom">5 mg/kg/2w<hr/></td><td align="left" valign="bottom">101<hr/></td><td align="left" valign="bottom">10.5 vs. 5.3<hr/></td><td align="left" valign="bottom">25.8 vs. 11.7<hr/></td><td align="left" valign="bottom">52.6% vs. 45.5%<hr/></td></tr><tr><td align="left" valign="bottom">Horinouchi Y/2011 [<xref ref-type="bibr" rid="B18">18</xref>]<hr/></td><td align="left" valign="bottom">First-line<hr/></td><td align="left" valign="bottom">NA<hr/></td><td align="left" valign="bottom">36<hr/></td><td align="left" valign="bottom">16.25 vs. 10<hr/></td><td align="left" valign="bottom">NA vs. NA<hr/></td><td align="left" valign="bottom">60% vs. 23.1%<hr/></td></tr><tr><td align="left" valign="bottom">Dewdney A/2011 [<xref ref-type="bibr" rid="B22">22</xref>]<hr/></td><td align="left" valign="bottom">First-line<hr/></td><td align="left" valign="bottom">7.5 mg/kg/3w<hr/></td><td align="left" valign="bottom">45<hr/></td><td align="left" valign="bottom">NA vs. NA<hr/></td><td align="left" valign="bottom">NA vs. NA<hr/></td><td align="left" valign="bottom">71% vs. 78%<hr/></td></tr><tr><td align="left">Tahover E/2013 [<xref ref-type="bibr" rid="B16">16</xref>]</td><td align="left">First- or second-line</td><td align="left">2. 5 mg/kg/w</td><td align="left">181</td><td align="left">29.9 vs. 17.2</td><td align="left">NA vs. 36.8</td><td align="left">NA vs. NA</td></tr></tbody></table><table-wrap-foot><p>ORR: Overall response rate; NA: Information not available; PFS: Progression-free survival; OS: Overall survival; HTN: Hypertension group; N0 HTN: No hypertension.</p></table-wrap-foot></table-wrap><sec><title>Efficacy</title><sec><title>Median PFS</title><p>The occurrence of hypertension induced by bevacizumab resulted in a statistically significant improvement in PFS compared with no hypertension (HR = 0.57; 95% CI: 0.46–0.72, <italic>P</italic> <0.001; heterogeneity χ<sup>2</sup> = 1.45, <italic>P</italic> for heterogeneity = 0.963; I<sup>2</sup> = 0.0%) (Figure  <xref ref-type="fig" rid="F2">2</xref>). There was no heterogeneity between trials.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p>Forest plot for meta-analysis of hypertension occurrence and progression-free survival.</p></caption><graphic xlink:href="1477-7819-11-306-2"/></fig></sec><sec><title>Median OS</title><p>Among the seven trials selected, five [<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B22">22</xref>] included relevant data. The pooled analysis showed that the occurrence of hypertension induced by bevacizumab also resulted in a statistically significant improvement in OS compared with no hypertension (HR = 0.50; 95% CI: 0.37–0.68, <italic>P</italic> <0.001; heterogeneity χ<sup>2</sup> = 5.12, <italic>P</italic> for heterogeneity = 0.275; I<sup>2</sup> = 21.9%) (Figure  <xref ref-type="fig" rid="F3">3</xref>). Once again, there was no heterogeneity between trials.</p><fig id="F3" position="float"><label>Figure 3</label><caption><p>Forest plot for meta-analysis of hypertension occurrence and overall survival.</p></caption><graphic xlink:href="1477-7819-11-306-3"/></fig></sec><sec><title>ORR</title><p>Two studies [<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B21">21</xref>] did not access this outcome, and were thus excluded from the analysis. The remaining five studies [<xref ref-type="bibr" rid="B17">17</xref>-<xref ref-type="bibr" rid="B20">20</xref>,<xref ref-type="bibr" rid="B22">22</xref>] contained pertinent data. Analysis indicated hypertension induced by bevacizumab was associated with an increase in ORR (RR = 1.57, 95% CI: 1.07–2.30, <italic>P</italic> <0.05) (Figure  <xref ref-type="fig" rid="F4">4</xref>). Because heterogeneity was significant between trials (I<sup>2</sup> = 63.7%, <italic>P</italic> = 0.026), a combined effects model was used. Funnel plots and the Egger’s test were used to assess publication bias. As reflected in Figures  <xref ref-type="fig" rid="F5">5</xref>, <xref ref-type="fig" rid="F6">6</xref>, and <xref ref-type="fig" rid="F7">7</xref>, the shape of the funnel plots appeared symmetrical.</p><fig id="F4" position="float"><label>Figure 4</label><caption><p>Forest plot for meta-analysis of hypertension occurrence and risk ratio.</p></caption><graphic xlink:href="1477-7819-11-306-4"/></fig><fig id="F5" position="float"><label>Figure 5</label><caption><p>Funnel plot for progression-free survival meta-analysis.</p></caption><graphic xlink:href="1477-7819-11-306-5"/></fig><fig id="F6" position="float"><label>Figure 6</label><caption><p>Funnel plot for overall survival meta-analysis.</p></caption><graphic xlink:href="1477-7819-11-306-6"/></fig><fig id="F7" position="float"><label>Figure 7</label><caption><p>Funnel plot for overall response rate meta-analysis.</p></caption><graphic xlink:href="1477-7819-11-306-7"/></fig></sec></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>Bevacizumab is widely used as a standard treatment for mCRC; the combined treatment of chemotherapy and bevacizumab has significantly increased the PFS and OS in patients with mCRC. Arterial hypertension is the most common side effect of bevacizumab plus chemotherapy treatment, with an overall incidence of 22–32%, and grade 3/4 events in 11–16% of patients [<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B32">32</xref>]. While the hypertension-causing mechanism of bevacizumab is unclear, it is fortunate that bevacizumab-induced hypertension rarely induces severe or life-threatening outcomes. To date, no specific predictive or prognostic biomarkers for bevacizumab treatment have been identified. Some studies have suggested that bevacizumab-induced hypertension could represent a valuable prognostic factor of clinical outcome in advanced-stage CRC patients [<xref ref-type="bibr" rid="B16">16</xref>-<xref ref-type="bibr" rid="B21">21</xref>]. Thus, it would be interesting to see if hypertension could be a predictive factor in patients with mCRC.</p><p>To our knowledge, this is the first meta-analysis that systematically evaluates the correlation of hypertension with survival and response in mCRC patients treated with bevacizumab. Our results indeed demonstrate that bevacizumab-induced hypertension in mCRC patients is significantly associated with PFS and OS. Also, our meta-analysis indicate that the occurrence of hypertension induced by bevacizumab is associated with a statistically significant improvement in ORR, in line with previous studies [<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>].</p><p>An outstanding benefit of our study is that patients who are more suitable to bevacizumab treatment could eventually be screened and selected for targeted therapy. Furthermore, it would be of extreme benefit to the fight against cancer if these results were comparable with the outcomes of anti-EGFR monoclonal antibody and KRAS status in CRC [<xref ref-type="bibr" rid="B33">33</xref>,<xref ref-type="bibr" rid="B34">34</xref>].</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>Caution is certainly needed before we may conclude that bevacizumab-induced hypertension is a reliable bio/clinical marker for early screening and diagnosis of patients with mCRC due to the limitation on available data and the relatively small sample size of our study. However, our results should undoubtedly lead to larger sample size, multiple-center clinical studies as well as analyses to further elucidate the correlation between bevacizumab-induced hypertension and mCRC. Thus, feasible and efficient methods to diagnose and treat patients with mCRC at earliest possible stages could be developed.</p></sec><sec><title>Abbreviations</title><p>CRC: Colorectal cancer; CI: Confidence interval; HR: Hazard ratio; mCRC: Metastatic CRC; ORR: Overall response rate; OS: Overall survival; PFS: Progression-free survival; RR: Risk ratio; VEGF: Vascular endothelial growth factor.</p></sec><sec><title>Competing interests</title><p>The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.</p></sec><sec><title>Authors’ contributions</title><p>JC and HM are co-first authors. All authors read and approved the final manuscript.</p></sec>
Prognostic and predictive value of Phospho-p44/42 and pAKT in HER2-positive locally advanced breast cancer patients treated with anthracycline-based neoadjuvant chemotherapy	<sec><title>Background</title><p>To evaluate the predictive and prognostic value of various molecular factors associated with the Ras/MAPK and PI3K/Akt signaling pathways in HER2-positive locally advanced breast cancer patients treated with anthracycline-based neoadjuvant chemotherapy (NAC).</p></sec><sec><title>Methods</title><p>A total of 113 patients were recruited in this retrospective study. Core needle biopsies and excision samples were assessed through immunohistochemistry for various biomarkers, including IGF-1R, Phospho-p44/42, Ki67, pAKT, PTEN, p27, and cyclinD1. The changes in these biomarkers after NAC and their predictive and prognostic values were investigated.</p></sec><sec><title>Results</title><p>Significant decreases in Ki67, Phospho-p44/42, and pAKT expression were observed after treatment (30.7% vs. 18.1%, 36.4% vs. 18.9%, and 35.1% vs. 16.4%, respectively). The decreases in Phospho-p44/42, pAKT, and Ki67 expression were strongly associated with the response to anthracycline treatment (<italic>P</italic> = 0.027, <italic>P</italic> = 0.031, and <italic>P</italic> = 0.008, respectively). In a multivariate survival analysis, Phospho-p44/42 expression after neoadjuvant chemotherapy and lymph node status were significant independent prognostic factors of both relapse-free survival and overall survival.</p></sec><sec><title>Conclusions</title><p>Reductions in Ki-67, Phospho-p44/42, and pAKT expression are related to the clinical response to anthracycline-based NAC in HER2-positive breast cancer patients. High pAKT expression prior to NAC had a better clinical response. Phospho-p44/42 expression and lymph node status after NAC could be useful for determining relapse-free survival and overall survival.</p></sec>	<contrib contrib-type="author" equal-contrib="yes" id="A1"><name><surname>Huang</surname><given-names>Liang</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>huangliang58234085@163.com</email></contrib><contrib contrib-type="author" equal-contrib="yes" id="A2"><name><surname>Chen</surname><given-names>Tianwen</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>0556175@fudan.edu.cn</email></contrib><contrib contrib-type="author" corresp="yes" id="A3"><name><surname>Chen</surname><given-names>Canming</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>fdhlyx@gmail.com</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Chen</surname><given-names>Sheng</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>0456177@fudan.edu.cn</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Liu</surname><given-names>Yin</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>0556252@fudan.edu.cn</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Wu</surname><given-names>Jiong</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>lovecqcat@gmail.com</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Shao</surname><given-names>Zhiming</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>573050723@qq.com</email></contrib>	World Journal of Surgical Oncology	<sec><title>Background</title><p>Human epidermal growth factor receptor 2 (HER2) is a tyrosine kinase receptor; up to 25% of women with early breast cancer are HER2 positive. HER2 is associated with a more aggressive biological behavior, a higher likelihood of recurrence after initial treatment, and poorer prognosis [<xref ref-type="bibr" rid="B1">1</xref>]. Three publications have suggested that HER2 positivity is associated with a relative benefit from anthracycline-containing chemotherapy compared with non-anthracycline-containing regimens, which is in agreement with the results of two meta-analyses [<xref ref-type="bibr" rid="B2">2</xref>-<xref ref-type="bibr" rid="B6">6</xref>]. Since trastuzumab was approved for use in HER2-positive breast cancer patients, the prognosis of breast cancer patients has improved. When used as a single agent, overall response rates ranging from 15% to 30% have been reported [<xref ref-type="bibr" rid="B7">7</xref>].</p><p>Neoadjuvant chemotherapy (NAC) has been used in locally advanced breast cancer to convert previously unresectable cancer into operable cancer. More recently, it has been widely administered in primarily operable breast cancer to reduce tumor volume and allow conservative surgery. The complete pathological response rate for patients with HER2-positive tumors is nearly 23%, but the rate has been shown to increase to 40% with trastuzumab [<xref ref-type="bibr" rid="B8">8</xref>]. The additional use of anthracyclines in combination with trastuzumab is thought to explain the higher complete pathological response rates observed in the GeParQuinto trial [<xref ref-type="bibr" rid="B9">9</xref>].</p><p>HER2 overexpression may lead to increased receptor homodimerization and heterodimerization, which causes intrinsic receptor tyrosine kinase activity and induces phosphorylation of the intracellular domain. The growth factor receptors utilize several signaling pathways, including the Ras/MAPK pathway, which is important for mitogenic stimulation. They also activate the PI3K/Akt cascade, which has been shown to be important for cell survival and inhibiting apoptosis. The PI3K pathway is downstream of HER2 and is activated to catalyze the phosphorylation of inositol lipids to produce PIP3 from PIP2. PIP3 recruits protein kinases and activates the protein kinase B/AKT pathway. AKT phosphorylation can inhibit cell cycle arrest. The mitogen-activated protein kinase (MAPK) signaling pathway is known to be activated in breast cancer. Extracellular signal-related kinase (ERK), a member of the MAPK pathway, promotes cell proliferation, angiogenesis, cell differentiation, and cell survival.</p><p>Insulin-like growth factor receptor-1 (IGFR-1) is a transmembrane heterotetrameric protein. It is the dominant receptor for the IGF family of molecules, and it promotes the oncogenic transformation, growth and survival of cancer cells. IGF-I/II ligand binding induces intracellular tyrosine kinase activity and triggers a cascade of reactions involving signal transduction pathways, including the Ras, Raf, MAPK, and PI3K–AKT pathways. In breast cancer, IGFR-1 expression and activation have been linked to disease progression and poor prognosis [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B11">11</xref>].</p><p>To improve the efficacy of treatment in HER2-positive breast cancer patients, it is critical to study the correlation between HER2 signaling pathway activity and the efficacy of adjuvant treatment. Therefore, we initiated a retrospective study to collect serial samples of HER2-positive breast cancer for molecular analyses in patients undergoing anthracycline-based neoadjuvant chemotherapy.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Ethics statement</title><p>The retrospective study was approved by the Ethics Committee of Shanghai Cancer Center. Written informed consent was obtained from each patient involved in the study.</p></sec><sec><title>Patients and clinical samples</title><p>From May 2002 to September 2007, 113 patients with HER2-overexpressing (defined as either 3+ or 2+ with confirmed c-erbB2 gene amplification by fluorescence <italic>in situ</italic> hybridization) stage II to III breast cancer were retrospectively recruited [<xref ref-type="bibr" rid="B12">12</xref>]. Core needle biopsy was performed for every patient to confirm the diagnosis of invasive cancer. A complete history, including patient characteristics, clinical and imaging examinations, and pathologic assessments of the morphologic and biologic features of the cancers, was collected. Patients with metastatic diseases, inflammatory breast cancer or male breast cancer were not included in this study. All patients were treated with CEF (cyclophosphamide 600 mg/m<sup>2</sup>, epirubicin 80 mg/m<sup>2</sup> and fluorouracil 500 mg/m<sup>2</sup>, q3w) or NE (vinorelbine 25 mg/m<sup>2</sup> on days 1 and 8 and epirubicin 60 mg/m<sup>2</sup> on day 1, q3w). Following completion of neoadjuvant treatment, all patients underwent breast surgery. For adjuvant chemotherapy, 74.3% of cases had an anthracycline-based regimen, and 10.6% of cases had a paclitaxel regimen; two patients received trastuzumab treatment for 1 year. Other standard therapies, including radiation therapy and endocrine therapy, were administered at the discretion of the treating clinician following NCCN guidelines. All patients were followed-up every 3 months for the first year and then every 6 months until death.</p></sec><sec><title>Assessment of the response to neoadjuvant chemotherapy</title><p>All surgical specimens were submitted for pathological evaluation. A complete pathological response was defined as no residual invasive carcinoma in the breast or lymph nodes. The clinical stage and size of the primary tumor measured by MRI or ultrasonography were recorded before treatment. The primary tumor was measured as the product of its greatest diameter. The clinical response was evaluated at each cycle of chemotherapy and prior to definitive surgery on day 21 of the last cycle of chemotherapy as the product of the primary tumor diameters and the axillary clinical status and classified as a complete response, partial response, stable disease, or progressive disease according to the solid tumors criteria (RECIST 1.1).</p></sec><sec><title>Immunochemistry</title><p>Immunohistochemistry was performed on deparaffinized sections of all core needle biopsies and surgical tumor samples. Antibodies for P27, Phospho-p44/42 and pAKT required antigen retrieval in a pressure cooker. Briefly, after antigen retrieval, endogenous peroxidases and biotin were blocked with 3% hydrogen peroxide and an Avidin-Biotin Blocking Kit (Vector, CA), respectively. This was followed by incubation with the primary antibody for 60 minutes at room temperature, the appropriate secondary antibody (Dako), labeled streptavidin-horseradish-peroxidase (Dako), DAB chromogen, and 0.2% osmium tetroxide (Sigma Chemicals, St Louis, MO), followed by counterstaining with light hematoxylin. Appropriate positive controls for each antibody and negative controls using species-matched immunoglobulin to replace the primary antibody were run with each batch. Positive tumor cells were quantified by evaluating at least 1,000 cells and expressed as percentages. Samples were evaluated by two trained pathologists (Dr. Zou, Dr. Li) who were blinded to the patient background and clinical outcome. If the difference between the two results was more than 10%, a third pathologist (Dr. Zhou) was consulted.</p><p>The cut-off for estrogen receptor (clone 1D5, Dako) and progesterone receptor (clone PgR 636, Dako) positivity was 1% of tumor cells with positive nuclear staining. For Phospho-p44/42 (clone 20G11, Cell Signaling) and pAKT (clone 736E11, Cell Signaling), the cut-off for positive expression was 20% of cells with nuclear and cytoplasmic expression [<xref ref-type="bibr" rid="B13">13</xref>]. The cut-off for Ki67 (clone MIB-1, Dako) positivity was 14% of tumor cells with positive nuclear staining. The cut-off for other markers was 10%, including nuclear staining for P27 (clone SX53G8, Dako) and cyclinD1 (clone EP12, Dako), membranous and cytoplasmic staining for IGF-1R (clone 3027, Cell Signaling), and cytoplasmic and nuclear staining for PTEN (clone 6H2.1, Dako) [<xref ref-type="bibr" rid="B14">14</xref>-<xref ref-type="bibr" rid="B16">16</xref>]. The reduction between the pre-NAC average percentage and post-NAC average percentage was defined as the reduction cut-off for Ki67, pAKT, and Phospho-p44/42 (13%, 19%, and 18%, respectively).</p></sec><sec><title>Statistical analysis</title><p>Descriptive statistics were calculated to summarize patient characteristics, tumor size, and the biomarker levels in the core needle biopsies and surgical tumor samples. Biomarker expression levels in pre- and post-chemotherapy tumor samples were compared using a paired <italic>t</italic> test. The initial biomarker levels were compared between responders and non-responders using the chi-square test. Fisher’s exact test was performed when necessary. A simultaneous analysis of the biomarkers that were significantly predictive of tumor response in the univariate analysis was performed using a multivariate logistic regression.</p><p>Survival results were last updated in September 2012. Relapse-free survival was defined as the elapsed time between the date of first diagnosis and the date of first relapse. Overall survival was calculated from the date of diagnosis to the date of death or last follow-up. Patients without events or death were censored at the last follow-up. Survival curves were established according to the Kaplan-Meier method. The log-rank test was used for univariate comparison of survival endpoints. A Cox regression was used to assess the relative influence of prognostic factors on relapse-free survival and overall survival. All tests were considered significant at a two-sided <italic>P</italic> < 0.05. All analyses were performed using SPSS 17.0 (SPSS, Chicago, IL).</p></sec></sec><sec sec-type="results"><title>Results</title><sec><title>Clinical characteristics and responses to NAC</title><p>A total of 113 HER2-positive breast cancer patients were recruited in this retrospective study. The average age of patients at diagnosis was 49 (range 26 to 78) years; 69 patients were premenopausal at presentation. There were 67 patients with a baseline tumor size greater than 5 cm. The CEF regimen was given to 55 patients, and the other patients received the NE regimen. The mean number of NAC cycles was 3.52 (range 1 to 8), and the objective response (complete response + partial response) and non-response rates (progressive disease + stable disease) were 70.8% and 29.2%, respectively. Eight patients underwent breast-conserving surgery. The clinical characteristics of these patients are shown in Table <xref ref-type="table" rid="T1">1</xref>.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Clinical characteristics of HER2-positive breast cancer patients and the univariate analysis of predictive biomarkers of the response to anthracyclines</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left" valign="bottom"><bold>Factors</bold><hr/></th><th align="left" valign="bottom"><bold>Number of patients (%)</bold><hr/></th><th colspan="3" align="left" valign="bottom"><bold>Clinical response</bold><hr/></th><th colspan="3" align="left" valign="bottom"><bold>Pathological response</bold><hr/></th></tr><tr><th align="left"> </th><th align="left"> </th><th align="left"><bold>Stable disease + progressive disease</bold></th><th align="left"><bold>Partial response + complete response</bold></th><th align="left"><bold> <italic>P</italic> </bold></th><th align="left"><bold>Complete pathological response</bold></th><th align="left"><bold>Incomplete pathological response</bold></th><th align="left"><bold> <italic>P</italic> </bold></th></tr></thead><tbody valign="top"><tr><td colspan="8" align="left" valign="bottom">Age<hr/></td></tr><tr><td align="left" valign="bottom">45 years<hr/></td><td align="left" valign="bottom">34 (30%)<hr/></td><td align="left" valign="bottom">10<hr/></td><td align="left" valign="bottom">24<hr/></td><td align="left" valign="bottom">0.975<hr/></td><td align="left" valign="bottom">3<hr/></td><td align="left" valign="bottom">31<hr/></td><td align="left" valign="bottom">0.450<hr/></td></tr><tr><td align="left" valign="bottom">≥45 years<hr/></td><td align="left" valign="bottom">79 (70%)<hr/></td><td align="left" valign="bottom">23<hr/></td><td align="left" valign="bottom">56<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">11<hr/></td><td align="left" valign="bottom">68<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td colspan="8" align="left" valign="bottom">Menopausal status<hr/></td></tr><tr><td align="left" valign="bottom">Postmenopausal<hr/></td><td align="left" valign="bottom">69 (61%)<hr/></td><td align="left" valign="bottom">20<hr/></td><td align="left" valign="bottom">49<hr/></td><td align="left" valign="bottom">0.949<hr/></td><td align="left" valign="bottom">7<hr/></td><td align="left" valign="bottom">62<hr/></td><td align="left" valign="bottom">0.364<hr/></td></tr><tr><td align="left" valign="bottom">Premenopausal<hr/></td><td align="left" valign="bottom">44 (39%)<hr/></td><td align="left" valign="bottom">13<hr/></td><td align="left" valign="bottom">31<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">7<hr/></td><td align="left" valign="bottom">37<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td colspan="8" align="left" valign="bottom">Regimen<hr/></td></tr><tr><td align="left" valign="bottom"> CEF<hr/></td><td align="left" valign="bottom">55 (49%)<hr/></td><td align="left" valign="bottom">15<hr/></td><td align="left" valign="bottom">40<hr/></td><td align="left" valign="bottom">0.660<hr/></td><td align="left" valign="bottom">9<hr/></td><td align="left" valign="bottom">46<hr/></td><td align="left" valign="bottom">0.212<hr/></td></tr><tr><td align="left" valign="bottom"> NE<hr/></td><td align="left" valign="bottom">58 (51%)<hr/></td><td align="left" valign="bottom">18<hr/></td><td align="left" valign="bottom">40<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">5<hr/></td><td align="left" valign="bottom">53<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td colspan="8" align="left" valign="bottom">Tumor size<hr/></td></tr><tr><td align="left" valign="bottom">≤5 cm<hr/></td><td align="left" valign="bottom">46 (41%)<hr/></td><td align="left" valign="bottom">21<hr/></td><td align="left" valign="bottom">25<hr/></td><td align="left" valign="bottom">0.001<hr/></td><td align="left" valign="bottom">4<hr/></td><td align="left" valign="bottom">42<hr/></td><td align="left" valign="bottom">0.323<hr/></td></tr><tr><td align="left" valign="bottom">>5 cm<hr/></td><td align="left" valign="bottom">67 (59%)<hr/></td><td align="left" valign="bottom">12<hr/></td><td align="left" valign="bottom">55<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">10<hr/></td><td align="left" valign="bottom">57<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td colspan="8" align="left" valign="bottom">Lymph node status<hr/></td></tr><tr><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">43 (38%)<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td></tr><tr><td align="left" valign="bottom">1 to 3<hr/></td><td align="left" valign="bottom">31 (27%)<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td></tr><tr><td align="left" valign="bottom">4 to 9<hr/></td><td align="left" valign="bottom">25 (22%)<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td></tr><tr><td align="left" valign="bottom">≥10<hr/></td><td align="left" valign="bottom">14 (13%)<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td></tr><tr><td colspan="8" align="left" valign="bottom">Pre-estrogen receptor<hr/></td></tr><tr><td align="left" valign="bottom">Negative<hr/></td><td align="left" valign="bottom">78 (69%)<hr/></td><td align="left" valign="bottom">21<hr/></td><td align="left" valign="bottom">57<hr/></td><td align="left" valign="bottom">0.426<hr/></td><td align="left" valign="bottom">10<hr/></td><td align="left" valign="bottom">68<hr/></td><td align="left" valign="bottom">0.835<hr/></td></tr><tr><td align="left" valign="bottom">Positive<hr/></td><td align="left" valign="bottom">35 (31%)<hr/></td><td align="left" valign="bottom">12<hr/></td><td align="left" valign="bottom">23<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">4<hr/></td><td align="left" valign="bottom">31<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td colspan="8" align="left" valign="bottom">Pre-progesterone receptor<hr/></td></tr><tr><td align="left" valign="bottom">Negative<hr/></td><td align="left" valign="bottom">83 (73%)<hr/></td><td align="left" valign="bottom">24<hr/></td><td align="left" valign="bottom">59<hr/></td><td align="left" valign="bottom">0.911<hr/></td><td align="left" valign="bottom">9<hr/></td><td align="left" valign="bottom">74<hr/></td><td align="left" valign="bottom">0.518<hr/></td></tr><tr><td align="left" valign="bottom">Positive<hr/></td><td align="left" valign="bottom">30 (27%)<hr/></td><td align="left" valign="bottom">9<hr/></td><td align="left" valign="bottom">21<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">5<hr/></td><td align="left" valign="bottom">25<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td colspan="8" align="left" valign="bottom">Pre-pMAPK<hr/></td></tr><tr><td align="left" valign="bottom">Negative<hr/></td><td align="left" valign="bottom">34 (30%)<hr/></td><td align="left" valign="bottom">13<hr/></td><td align="left" valign="bottom">21<hr/></td><td align="left" valign="bottom">0.166<hr/></td><td align="left" valign="bottom">8<hr/></td><td align="left" valign="bottom">26<hr/></td><td align="left" valign="bottom">0.855<hr/></td></tr><tr><td align="left" valign="bottom">Positive<hr/></td><td align="left" valign="bottom">79 (70%)<hr/></td><td align="left" valign="bottom">20<hr/></td><td align="left" valign="bottom">59<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">6<hr/></td><td align="left" valign="bottom">73<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td colspan="8" align="left" valign="bottom">Pre-pAKT<hr/></td></tr><tr><td align="left" valign="bottom">Negative<hr/></td><td align="left" valign="bottom">30 (27%)<hr/></td><td align="left" valign="bottom">14<hr/></td><td align="left" valign="bottom">16<hr/></td><td align="left" valign="bottom">0.014<hr/></td><td align="left" valign="bottom">4<hr/></td><td align="left" valign="bottom">26<hr/></td><td align="left" valign="bottom">0.855<hr/></td></tr><tr><td align="left" valign="bottom">Positive<hr/></td><td align="left" valign="bottom">83 (73%)<hr/></td><td align="left" valign="bottom">19<hr/></td><td align="left" valign="bottom">64<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">10<hr/></td><td align="left" valign="bottom">73<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td colspan="8" align="left" valign="bottom">Pre-PTEN<hr/></td></tr><tr><td align="left" valign="bottom">Negative<hr/></td><td align="left" valign="bottom">40 (35%)<hr/></td><td align="left" valign="bottom">13<hr/></td><td align="left" valign="bottom">27<hr/></td><td align="left" valign="bottom">0.568<hr/></td><td align="left" valign="bottom">5<hr/></td><td align="left" valign="bottom">35<hr/></td><td align="left" valign="bottom">0.979<hr/></td></tr><tr><td align="left" valign="bottom">Positive<hr/></td><td align="left" valign="bottom">73 (65%)<hr/></td><td align="left" valign="bottom">20<hr/></td><td align="left" valign="bottom">53<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">9<hr/></td><td align="left" valign="bottom">64<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td colspan="8" align="left" valign="bottom">Pre-P27<hr/></td></tr><tr><td align="left" valign="bottom">Negative<hr/></td><td align="left" valign="bottom">35 (31%)<hr/></td><td align="left" valign="bottom">15<hr/></td><td align="left" valign="bottom">20<hr/></td><td align="left" valign="bottom">0.432<hr/></td><td align="left" valign="bottom">5<hr/></td><td align="left" valign="bottom">40<hr/></td><td align="left" valign="bottom">0.737<hr/></td></tr><tr><td align="left" valign="bottom">Positive<hr/></td><td align="left" valign="bottom">68 (69%)<hr/></td><td align="left" valign="bottom">18<hr/></td><td align="left" valign="bottom">50<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">9<hr/></td><td align="left" valign="bottom">59<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td colspan="8" align="left" valign="bottom">Pre-IGF-1R<hr/></td></tr><tr><td align="left" valign="bottom">Negative<hr/></td><td align="left" valign="bottom">47 (42%)<hr/></td><td align="left" valign="bottom">12<hr/></td><td align="left" valign="bottom">35<hr/></td><td align="left" valign="bottom">0.469<hr/></td><td align="left" valign="bottom">5<hr/></td><td align="left" valign="bottom">42<hr/></td><td align="left" valign="bottom">0.634<hr/></td></tr><tr><td align="left" valign="bottom">Positive<hr/></td><td align="left" valign="bottom">66 (58%)<hr/></td><td align="left" valign="bottom">21<hr/></td><td align="left" valign="bottom">45<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">9<hr/></td><td align="left" valign="bottom">57<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td colspan="8" align="left" valign="bottom">Pre-cyclinD1<hr/></td></tr><tr><td align="left" valign="bottom">Negative<hr/></td><td align="left" valign="bottom">42 (37%)<hr/></td><td align="left" valign="bottom">14<hr/></td><td align="left" valign="bottom">28<hr/></td><td align="left" valign="bottom">0.458<hr/></td><td align="left" valign="bottom">4<hr/></td><td align="left" valign="bottom">38<hr/></td><td align="left" valign="bottom">0.477<hr/></td></tr><tr><td align="left" valign="bottom">Positive<hr/></td><td align="left" valign="bottom">71 (63%)<hr/></td><td align="left" valign="bottom">19<hr/></td><td align="left" valign="bottom">52<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">10<hr/></td><td align="left" valign="bottom">61<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td colspan="8" align="left" valign="bottom">Pre-Ki67<hr/></td></tr><tr><td align="left" valign="bottom">Negative<hr/></td><td align="left" valign="bottom">43 (38%)<hr/></td><td align="left" valign="bottom">15<hr/></td><td align="left" valign="bottom">28<hr/></td><td align="left" valign="bottom">0.298<hr/></td><td align="left" valign="bottom">2<hr/></td><td align="left" valign="bottom">41<hr/></td><td align="left" valign="bottom">0.076<hr/></td></tr><tr><td align="left">Positive</td><td align="left">70 (62%)</td><td align="left">18</td><td align="left">52</td><td align="left"> </td><td align="left">12</td><td align="left">58</td><td align="left"> </td></tr></tbody></table><table-wrap-foot><p>IGF-1R, insulin-like growth factor 1 receptor; pAKT, phosphorylated AkT; pMAPK, phosphorylated mitogen-activated protein kinase.</p></table-wrap-foot></table-wrap></sec><sec><title>Changes in biomarker expression after NAC</title><p>The expressions of various biomarkers (in Figure <xref ref-type="fig" rid="F1">1</xref>) before and after NAC were compared in 99 patients who did not achieve complete pathological response after NAC. A paired <italic>t</italic> test analysis found no significant changes in p27, PTEN, IGF-1R, or cyclinD1 expression (15.7% vs. 16.5%, 17.0% vs. 18.3%, 49.0% vs. 51.0%, and 18.2% vs. 16.9%, respectively). A significant decrease in Ki67, Phospho-p44/42, and pAKT expression was observed after treatment (30.7% vs. 18.1%, 36.4% vs. 18.9%, and 35.1% vs. 16.4%, respectively). We defined positive-staining tumor cells decreasing by more than the cut-off value as positive decrease in biomarker expression. The details are shown in Table <xref ref-type="table" rid="T2">2</xref>. Ki67, Phospho-p44/42, and pAKT expression were all significantly decreased after anthracycline-based neoadjuvant chemotherapy, as shown in Figure <xref ref-type="fig" rid="F2">2</xref>.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p>Positive immunohistochemical expression: (A) Ki67, (B) Phospho-p44/42, (C) pAKT, (D) p27, (E) PTEN, (F) IGF-1R, (G) cyclinD1.</p></caption><graphic xlink:href="1477-7819-11-307-1"/></fig><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Univariate analysis of relapse-free survival and overall survival</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="left"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="left" valign="bottom"><bold>Factor</bold><hr/></th><th colspan="2" align="center" valign="bottom"><bold>Relapse-free survival</bold><hr/></th><th colspan="2" align="center" valign="bottom"><bold>Overall survival</bold><hr/></th><th align="left" valign="bottom"><bold>Factor</bold><hr/></th><th colspan="2" align="center" valign="bottom"><bold>Relapse-free survival</bold><hr/></th><th colspan="2" align="center" valign="bottom"><bold>Overall survival</bold><hr/></th></tr><tr><th align="left"> </th><th align="center"><bold> <italic>P</italic> </bold></th><th align="center"><bold>Hazard ratio (95% confidence interval)</bold></th><th align="center"><bold> <italic>P</italic> </bold></th><th align="center"><bold>Hazard ratio (95% confidence interval)</bold></th><th align="left"> </th><th align="center"><bold> <italic>P</italic> </bold></th><th align="center"><bold>Hazard ratio (95% confidence interval)</bold></th><th align="center"><bold> <italic>P</italic> </bold></th><th align="center"><bold>Hazard ratio (95% confidence interval)</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Age<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"><45y vs. ≥45y<hr/></td><td align="center" valign="bottom">0.724<hr/></td><td align="center" valign="bottom">1.2 (0.5 to 2.7)<hr/></td><td align="center" valign="bottom">0.256<hr/></td><td align="center" valign="bottom">1.0 (0.9 to 1.0)<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Menopausal status<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Premenopausal vs. Postmenopausal<hr/></td><td align="center" valign="bottom">0.663<hr/></td><td align="center" valign="bottom">0.9 (0.5 to 1.5)<hr/></td><td align="center" valign="bottom">0.822<hr/></td><td align="center" valign="bottom">1.0 (0.6 to 2.0)<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Regimen<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">CEF vs. NE<hr/></td><td align="center" valign="bottom">0.479<hr/></td><td align="center" valign="bottom">1.2 (0.7 to 2.0)<hr/></td><td align="center" valign="bottom">0.568<hr/></td><td align="center" valign="bottom">1.2 (0.6 to 2.2)<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Tumor size<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">≤5 cm vs. >5 cm<hr/></td><td align="center" valign="bottom">0.784<hr/></td><td align="center" valign="bottom">1.1 (0.6 to 2.0)<hr/></td><td align="center" valign="bottom">0.456<hr/></td><td align="center" valign="bottom">1.3 (0.6 to 2.8)<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Lymph node status<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Negative<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">1 to 3 nodes +<hr/></td><td align="center" valign="bottom">0.007<hr/></td><td align="center" valign="bottom">2.7 (1.3 to 5.6)<hr/></td><td align="center" valign="bottom">0.012<hr/></td><td align="center" valign="bottom">4.3 (1.4 to 13.3)<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">4 to 9 nodes +<hr/></td><td align="center" valign="bottom"><0.001<hr/></td><td align="center" valign="bottom">4.6 (2.2 to 9.5)<hr/></td><td align="center" valign="bottom">0.001<hr/></td><td align="center" valign="bottom">7.2 (2.4 to 21.9)<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">≥10 nodes +<hr/></td><td align="center" valign="bottom"><0.001<hr/></td><td align="center" valign="bottom">6.8 (3.0 to 15.9)<hr/></td><td align="center" valign="bottom"><0.001<hr/></td><td align="center" valign="bottom">13.7 (4.3 to 44.4)<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Pre-estrogen receptor<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="left" valign="bottom">Post-estrogen receptor<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Negative vs. positive<hr/></td><td align="center" valign="bottom">0.252<hr/></td><td align="center" valign="bottom">1.4 (0.8 to 2.3)<hr/></td><td align="center" valign="bottom">0.620<hr/></td><td align="center" valign="bottom">1.2 (0.6 to 2.2)<hr/></td><td align="left" valign="bottom">Negative vs. positive<hr/></td><td align="center" valign="bottom">1.000<hr/></td><td align="center" valign="bottom">1.0 (0.5 to 1.8)<hr/></td><td align="center" valign="bottom">0.540<hr/></td><td align="center" valign="bottom">0.8 (0.4 to 1.7)<hr/></td></tr><tr><td align="left" valign="bottom">Pre-progesterone receptor<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="left" valign="bottom">Post-progesterone receptor<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Negative vs. positive<hr/></td><td align="center" valign="bottom">0.724<hr/></td><td align="center" valign="bottom">0.9 (0.5 to 1.6)<hr/></td><td align="center" valign="bottom">0.334<hr/></td><td align="center" valign="bottom">0.7 (0.3 to 1.5)<hr/></td><td align="left" valign="bottom">Negative vs. positive<hr/></td><td align="center" valign="bottom">0.841<hr/></td><td align="center" valign="bottom">1.0 (0.6 to 2.0)<hr/></td><td align="center" valign="bottom">0.754<hr/></td><td align="center" valign="bottom">0.9 (0.4 to 2.0)<hr/></td></tr><tr><td align="left" valign="bottom">Pre-Ki67<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="left" valign="bottom">Post-Ki67<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Negative vs. positive<hr/></td><td align="center" valign="bottom">0.840<hr/></td><td align="center" valign="bottom">1.1 (0.6 to 1.8)<hr/></td><td align="center" valign="bottom">0.647<hr/></td><td align="center" valign="bottom">0.9 (0.5 to 1.6)<hr/></td><td align="left" valign="bottom">Negative vs. positive<hr/></td><td align="center" valign="bottom">0.260<hr/></td><td align="center" valign="bottom">1.4 (0.8 to 2.3)<hr/></td><td align="center" valign="bottom">0.434<hr/></td><td align="center" valign="bottom">1.3 (0.7 to 2.4)<hr/></td></tr><tr><td align="left" valign="bottom">Pre-pMAPK<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="left" valign="bottom">Post-pMAPK<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Negative vs. positive<hr/></td><td align="center" valign="bottom">0.174<hr/></td><td align="center" valign="bottom">0.7 (0.4 to 1.2)<hr/></td><td align="center" valign="bottom">0.788<hr/></td><td align="center" valign="bottom">1.1 (0.6 to 2.2)<hr/></td><td align="left" valign="bottom">Negative vs. positive<hr/></td><td align="center" valign="bottom">0.019<hr/></td><td align="center" valign="bottom">2.0 (1.1 to 3.5)<hr/></td><td align="center" valign="bottom"><0.001<hr/></td><td align="center" valign="bottom">3.7 (1.9 to 7.0)<hr/></td></tr><tr><td align="left" valign="bottom">Pre-pAKT<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="left" valign="bottom">Post-pAKT<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Negative vs. positive<hr/></td><td align="center" valign="bottom">0.224<hr/></td><td align="center" valign="bottom">0.7 (0.4 to 1.2)<hr/></td><td align="center" valign="bottom">0.104<hr/></td><td align="center" valign="bottom">0.6 (0.3 to 1.1)<hr/></td><td align="left" valign="bottom">Negative vs. positive<hr/></td><td align="center" valign="bottom">0.551<hr/></td><td align="center" valign="bottom">0.8 (0.5 to 1.5)<hr/></td><td align="center" valign="bottom">0.548<hr/></td><td align="center" valign="bottom">0.8 (0.4 to 1.7)<hr/></td></tr><tr><td align="left" valign="bottom">Pre-PTEN<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="left" valign="bottom">Post-PTEN<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Negative vs. positive<hr/></td><td align="center" valign="bottom">0.450<hr/></td><td align="center" valign="bottom">0.8 (0.5 to 1.4)<hr/></td><td align="center" valign="bottom">0.538<hr/></td><td align="center" valign="bottom">0.8 (0.4 to 1.5)<hr/></td><td align="left" valign="bottom">Negative vs. positive<hr/></td><td align="center" valign="bottom">0.622<hr/></td><td align="center" valign="bottom">0.9 (0.5 to 1.6)<hr/></td><td align="center" valign="bottom">0.571<hr/></td><td align="center" valign="bottom">1.3 (0.6 to 2.7)<hr/></td></tr><tr><td align="left" valign="bottom">Pre-P27<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="left" valign="bottom">Post-P27<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Negative vs. positive<hr/></td><td align="center" valign="bottom">0.929<hr/></td><td align="center" valign="bottom">1.0 (0.6 to 1.6)<hr/></td><td align="center" valign="bottom">0.439<hr/></td><td align="center" valign="bottom">0.8 (0.4 to 1.5)<hr/></td><td align="left" valign="bottom">Negative vs. positive<hr/></td><td align="center" valign="bottom">0.220<hr/></td><td align="center" valign="bottom">1.4 (0.8 to 2.5)<hr/></td><td align="center" valign="bottom">0.317<hr/></td><td align="center" valign="bottom">1.4 (0.7 to 2.9)<hr/></td></tr><tr><td align="left" valign="bottom">Pre-IGF-1R<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="left" valign="bottom">Post-IGF-1R<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Negative vs. positive<hr/></td><td align="center" valign="bottom">0.728<hr/></td><td align="center" valign="bottom">1.1 (0.7 to 1.8)<hr/></td><td align="center" valign="bottom">0.874<hr/></td><td align="center" valign="bottom">1.1 (0.6 to 2.0)<hr/></td><td align="left" valign="bottom">Negative vs. positive<hr/></td><td align="center" valign="bottom">0.388<hr/></td><td align="center" valign="bottom">0.8 (0.5 to 1.3)<hr/></td><td align="center" valign="bottom">0.760<hr/></td><td align="center" valign="bottom">1.1 (0.6 to 2.2)<hr/></td></tr><tr><td align="left" valign="bottom">Pre-cyclinD1<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="left" valign="bottom">Post-cyclinD1<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left">Negative vs. positive</td><td align="center">0.606</td><td align="center">1.2 (0.7 to 2.0)</td><td align="center">0.814</td><td align="center">1.1 (0.6 to 2.1)</td><td align="left">Negative vs. positive</td><td align="center">0.602</td><td align="center">1.2 (0.7 to 2.1)</td><td align="center">0.912</td><td align="center">1.0 (0.5 to 2.1)</td></tr></tbody></table><table-wrap-foot><p>IGF-1R, insulin-like growth factor 1 receptor; pAKT, phosphorylated AKT; pMAPK, phosphorylated mitogen-activated protein kinase.</p></table-wrap-foot></table-wrap><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Ki67, Phospho-p44/42 and pAKT expression were significantly decreased after anthracycline treatment. (A)</bold> Ki67, <bold>(B)</bold> Phospho-p44/42, <bold>(C)</bold> pAKT.</p></caption><graphic xlink:href="1477-7819-11-307-2"/></fig></sec><sec><title>Predictors and response to NAC</title><p>Biomarkers and clinical characteristics were examined to investigate their value in predicting the NAC response (Table <xref ref-type="table" rid="T2">2</xref>). Univariate analysis demonstrated that tumor size and pre-pAKT were predictive factors of the response to anthracyclines (<italic>P</italic> = 0.001 and <italic>P</italic> = 0.014, respectively). Multivariate analysis demonstrated that primary tumor size and pAKT expression remained independent predictive factors of the clinical response to anthracycline-based NAC (<italic>P</italic> = 0.012 and <italic>P</italic> = 0.031, respectively). However, no biomarker could predict a pathologic complete response. We also found that the clinical response was coincident with decreased biomarker expression, including Ki67, pAKT, and Phospho-p44/42 (<italic>P</italic> = 0.001, <italic>P</italic> = 0.002, and <italic>P</italic> = 0.007, respectively), as shown in Figure <xref ref-type="fig" rid="F3">3</xref>.</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Correlation between pretreatment and post-treatment markers and the pathological response.</bold> Values below the line indicate the percentage decrease compared to the cut-off value. <bold>(A)</bold> Ki67, <bold>(B)</bold> Phospho-p44/42, <bold>(C)</bold> pAKT.</p></caption><graphic xlink:href="1477-7819-11-307-3"/></fig></sec><sec><title>Prognostic markers</title><p>The median follow-up time was 60 months (ranging from 14 to 123 months). The overall 5-year relapse-free survival was 50.4%, and the overall survival was 72.6%. A univariate analysis (Table <xref ref-type="table" rid="T2">2</xref>) demonstrated that the number of positive lymph nodes and post-Phospho-p44/42 expression were prognostic factors for relapse-free survival. In the multivariate analysis, the number of positive lymph nodes (hazard ratio, 2.0; 95% confidence interval, 1.6 to 2.6; <italic>P</italic> < 0.001) and post-Phospho-p44/42 expression (hazard ratio, 2.3; 95% confidence interval 1.3 to 4.1; <italic>P</italic> < 0.001) remained significantly independent prognostic factors. In the univariate analysis (Table <xref ref-type="table" rid="T2">2</xref>), lymph node status (<italic>P</italic> < 0.001) and post-Phospho-p44/42 expression (<italic>P</italic> = 0.019) showed clear associations with overall survival. In the multivariate analysis, lymph node status (hazard ratio, 2.3; 95% confidence interval, 1.7 to 3.3; <italic>P</italic> < 0.001) and post-Phospho-p44/42 expression (hazard ratio, 4.3; 95% confidence interval, 2.2 to 8.4; <italic>P</italic> < 0.001) were also significant predictors of overall survival. Representative survival curves are shown in Figure <xref ref-type="fig" rid="F4">4</xref>.</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>Kaplan-Meier curves.</bold> Curves for relapse-free survival according to <bold>(A)</bold> post-Phospho-p44/42 and <bold>(B)</bold> lymph node status. Curves for overall survival according to <bold>(C)</bold> post-Phospho-p44/42 and <bold>(D)</bold> lymph node status.</p></caption><graphic xlink:href="1477-7819-11-307-4"/></fig></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>To our knowledge, this is the largest analysis of PI3K-Akt and MAPK pathway activation in HER2-overexpressing breast cancer patients who received an anthracycline-based neoadjuvant chemotherapy regimen without trastuzumab.</p><p>Owing to its aggressive nature and poor prognosis, a number of preclinical and clinical studies have focused on the HER2-positive subtype. The use of anthracyclines in neoadjuvant treatments for HER2-positive breast cancer in addition to trastuzumab is still controversial [<xref ref-type="bibr" rid="B17">17</xref>].</p><p>HER2 signaling activates pathways (PI3K/Akt and Ras/MAPK) regulating cell cycle progression and cell proliferation. In HER2-overexpressing MBC group, Gori (<italic>et al</italic>. found that Phospho-p44/42 and pAKT were not associated with the clinical outcome, although low Phospho-p44/42 expression showed a trend for association with a longer overall survival [<xref ref-type="bibr" rid="B18">18</xref>]. In our study, the pre-pAKT and pre-Phospho-p44/42 expression rates were high, and this finding has been confirmed by other recent studies [<xref ref-type="bibr" rid="B19">19</xref>-<xref ref-type="bibr" rid="B21">21</xref>]. The results indicate that the Ras/MAPK and PI3K/Akt pathways are universally active in locally advanced breast cancer with HER2 overexpression. However, the PI3K/Akt and Ras/MAPK pathways have been related to resistance to doxorubicin and paclitaxel in breast cancer cells [<xref ref-type="bibr" rid="B22">22</xref>,<xref ref-type="bibr" rid="B23">23</xref>]. The level of pre-pAKT expression has been shown to have a significant correlation with the objective response rates to anthracycline treatment. It is possible that Akt isoforms have a distinct impact on the cellular resistance to a given drug and, in fact, Akt activity does not confer equal resistance to different chemotherapeutic agents. For example, the overexpression of constitutively active Akt isoforms in HeLa cells has been shown to induce isoform-specific sensitivity to doxorubicin [<xref ref-type="bibr" rid="B24">24</xref>]. The role of pAKT in the neoadjuvant setting is still controversial, owing to limited investigations, even in large clinical trials. However, decreases in Phospho-p44/42 and pAKT expression are related to the response to anthracyclines. Higher levels of active MAPK may have aggressive biological behavior, which has been found to be associated with lymph node metastasis [<xref ref-type="bibr" rid="B25">25</xref>]. In the TNBC subgroup, high ERK protein expression levels and shorter survival times have been observed [<xref ref-type="bibr" rid="B26">26</xref>]. After anthracycline-based adjuvant treatment, a higher score was significantly associated with poorer survival following relapse compared to a lower expression score among patients with MAPK overexpression [<xref ref-type="bibr" rid="B21">21</xref>].</p><p>Furthermore, few studies have presented precise values for pAKT and Phospho-p44/42 in neoadjuvant chemotherapy. In our study, low Phospho-p44/42 expression after neoadjuvant chemotherapy was a strong prognostic factor for these patients. Patients with high post-Phospho-p44/42 expression had a higher recurrence rate (up to 68%) in the first 5 years, while patients with low post-Phospho-p44/42 expression had a lower recurrence rate (45%), and the survival difference between the two groups was highly significant. It is possible that decreased pAKT and Phospho-p44/42 expression promotes tumor apoptosis and inhibits tumor proliferation, resulting in a survival benefit for HER2-positive breast cancer patients treated with anthracyclines. It is also possible that topoisomerase IIα expression is regulated by Ras pathways and tumor proliferation status. The activation of the Ras/Raf/MAPK pathway has been shown to be involved in the induction of MRP-1 activity and topoisomerase IIα downregulation, which are the main mechanisms of anthracycline resistance [<xref ref-type="bibr" rid="B27">27</xref>].</p><p>Higher levels of the proliferation marker Ki67 are associated with poorer survival in breast cancer patients, but we found no prognostic value for pre-NAC Ki67, post-NAC Ki67, or the Ki67 fold change. Other studies have reached controversial conclusions, and it is therefore difficult to choose reasonable predictive and prognostic factors among pre-NAC Ki67, post-NAC Ki67, and Ki67 reduction [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B28">28</xref>-<xref ref-type="bibr" rid="B30">30</xref>].</p><p>This study has limitations common to all retrospective analyses, and it lacked a control group, such as patients treated with trastuzumab-containing NAC. However, based on results of big clinical trials, such as HERA, BCIRG 006, NCCTG N9831, and NSABP B-31, trastuzumab was approved for adjuvant chemotherapy after 2005. In our study, 71% patients were treated before 2005 when most patients did not receive trastuzumab in developing countries. Additionally, the number of patients was small. However, the scientific and clinical community must establish and evaluate these biomarkers and standardize cut-off levels. Although a measurement of topoisomerase II-α amplification was not within the scope of this study, a possible explanation for our findings may be that other confounding molecular factors are involved in the mechanism of the anthracycline response in HER2-positive patients.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>In HER2-positive breast cancer patients treated with anthracyclines, the expression of pAKT, Phospho-p44/42, and Ki67 decreased significantly after treatment. Furthermore, patients with high pAKT expression before NAC had higher objective response rates to anthracyclines. The results of our study demonstrate that Phospho-p44/42 expression after neoadjuvant chemotherapy is a strong predictor of outcome. It will be necessary and valuable to further evaluate potential therapeutic targets of the PI3K and MEK signaling pathways in HER2-positive breast cancer patients.</p></sec><sec><title>Abbreviations</title><p>ERK: Extracellular signal-related kinase; HER2: Human epidermal growth factor Receptor 2; IGF-1R: Insulin-like growth factor 1 receptor; MAPK: Mitogen-activated protein kinase; NAC: Neoadjuvant chemotherapy; pAKT: Phospho-protein kinase B; Phospho-p44/42: Phospho-mitogen-activated protein kinase; pMAPK: Phosphorylated mitogen-activated protein kinase.</p></sec><sec><title>Competing interests</title><p>All authors declare that they have no potential conflict of interest.</p></sec><sec><title>Authors’ contributions</title><p>LH and T-WC have made substantial contributions to the conception and design of the study, and the acquisition of data. SC and YL analyzed and interpreted the data. JW and Z-MS revised the manuscript critically for important intellectual content. C-MC gave final approval of the version to be published. All authors read and approved the final manuscript.</p></sec>
Diagnosis and treatment of solid pseudopapillary tumor of the pancreas: experience of one single institution from Turkey	<sec><title>Background</title><p>Solid pseudopapillary neoplasia (SPN) of the pancreas is an extremely rare epithelial tumor of low malignant potential. SPN accounts for less than 1% to 2% of exocrine pancreatic tumors. The aim of this study is to report our experience with SPN of the pancreas. It includes a summary of the current literature to provide a reference for the management of this rare clinical entity.</p></sec><sec><title>Methods</title><p>A retrospective analysis was performed of all patients diagnosed and treated for SPN in our hospital over the past 15 years (1998 to 2013). A database of the characteristics of these patients was developed, including age, gender, tumor location and size, treatment, and histopathological and immunohistochemical features.</p></sec><sec><title>Results</title><p>During this time period, 255 patients with pancreatic malignancy (which does not include ampulla vateri, distal choledocal and duodenal tumor) were admitted to our department, only 10 of whom were diagnosed as having SPN (2.5%). Nine patients were women (90%) and one patient was a man (10%). Their median age was 38.8 years (range 18 to 71). The most common symptoms were abdominal pain and dullness. Seven patients (70%) presented with abdominal pain or abdominal dullness and three patient (30%) were asymptomatic with the diagnosis made by an incidental finding on routine examination. Abdominal computed tomography and/or magnetic resonance imaging showed the typical features of solid pseudopapillary neoplasm in six (60%) of the patients. Four patients underwent distal pancreatectomy with splenectomy, one patient underwent a total mass excision, and one patient underwent total pancreatic resection. Two required extended distal pancreatectomy with splenectomy. Two underwent spleen-preserving distal pancreatectomy.</p></sec><sec><title>Conclusions</title><p>SPN is a rare neoplasm that primarily affects young women. The prognosis is favorable even in the presence of distant metastasis. Although surgical resection is generally curative, a close follow-up is advised in order to diagnose a local recurrence or distant metastasis and choose the proper therapeutic option for the patient.</p></sec>	<contrib contrib-type="author" equal-contrib="yes" id="A1"><name><surname>Yagcı</surname><given-names>Ayşe</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>drayseyagci@mynet.com</email></contrib><contrib contrib-type="author" equal-contrib="yes" id="A2"><name><surname>Yakan</surname><given-names>Savas</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>savasyakan@gmail.com</email></contrib><contrib contrib-type="author" corresp="yes" id="A3"><name><surname>Coskun</surname><given-names>Ali</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>dralicoskun@hotmail.com</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Erkan</surname><given-names>Nazif</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>naziferkan@gmail.com</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Yıldırım</surname><given-names>Mehmet</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>mehmetyildi@gmail.com</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Yalcın</surname><given-names>Evrim</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>evrmylcn@hotmail.com</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Postacı</surname><given-names>Hakan</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>postacih@yahoo.com</email></contrib>	World Journal of Surgical Oncology	<sec><title>Background</title><p>Solid pseudopapillary neoplasia (SPN) of the pancreas is an extremely rare epithelial tumor of low malignant potential. SPN accounts for less than 1% to 2% of exocrine pancreatic tumors [<xref ref-type="bibr" rid="B1">1</xref>]. Until it was defined by the World Health Organization (WHO) in 1996 as ‘solid pseudopapillary tumor’ of the pancreas, this tumor was described by using various names including ‘solid cystic tumor’, ‘papillary cystic tumor’, ‘papillary epithelial neoplasia’, ‘solid and papillary epithelial neoplasia’, ‘papillary epithelial tumor’ and ‘Frantz’s tumor’, ‘solid and papillary tumor’, ‘solid-cysticpapillary epithelial neoplasm’, ‘benign or malignant papillary tumor of the pancreas’ [<xref ref-type="bibr" rid="B2">2</xref>]. These tumors typically occur in young women during the second to fourth decade of life and are histologically characterized by cystic areas and solid pseudopapillary arranged cells. The origin of these tumors is still a matter of controversy.</p><p>In this study, we report our experience with SPN of the pancreas and include a summary of the current literature to provide a reference for the management of this rare clinical entity.</p></sec><sec sec-type="methods"><title>Methods</title><p>A retrospective analysis was carried out of all patients diagnosed and treated for SPN in our hospital over the past 15 years (1998 to 2013). A database of the characteristics of these patients was developed, including age, gender, tumor location (data were derived from radiological investigations or surgical records) and size (data were derived from radiological investigations or surgical records and finally confirmed by pathology), treatment (data were derived from the medical records, including the types of surgery), and histopathological and immunohistochemical features. Pre-operative fine needle aspiration cytology FNAC) was performed in one patient. All the patients who underwent resection were followed up every six months. The investigations performed included routine blood studies, chest X-ray, CA-19-9 level and either an ultrasound or computed tomography (CT) scan of the abdomen.</p><p>This study was approved by the Local Institutional Review Board of Izmir Bozyaka Education and research hospital.</p></sec><sec sec-type="results"><title>Results</title><p>During this time period, of 255 patients with pancreatic malignancy (which does not include ampulla vateri, distal choledocal and duodenal tumor) admitted to our department, only 10 were diagnosed as having SPN (2.5%). Nine patients were women (90%) and one patient was a man (10%). The patients had a median age of 38.8 years (range 18 to 71). The most common symptoms were abdominal pain and dullness. Seven patients (70%) presented with abdominal pain or abdominal dullness and three patient (30%) were asymptomatic with the diagnosis made by an incidental finding on routine examination. Abdominal CT and/or magnetic resonance imaging (MRI) showed the typical features of solid pseudopapillary neoplasm in six (60%) of the patients (Figure <xref ref-type="fig" rid="F1">1</xref>). Tumor markers (AFP, CEA, CA 19–9 and CA 125) were normal preoperatively in all patients. Usually, the tumors appeared as well-circumscribed lesions with a mixed cystic and solid component but were almost entirely solid or else cystic with thick walls. In one patient the tumor was located in the pancreatic head (10%), in four patients in the body (40%) and in the remaining five patients in the tail (50%). Four patients underwent distal pancreatectomy with splenectomy, one patient underwent a total mass excision and one patient underwent total pancreatic resection. Two required extended distal pancreatectomy with splenectomy. Two underwent spleen-preserving distal pancreatectomy. The mean diameter of the tumor was 8 cm (range 3 to 13 cm). Patient characteristics are summarized in Table <xref ref-type="table" rid="T1">1</xref>.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Magnetic resonance imaging shows that the tumor is a well</bold>-<bold>marginated</bold>, <bold>large</bold>, <bold>encapsulated</bold>, <bold>solid and cystic mass with areas of hemorrhagic degeneration</bold>, <bold>as revealed by high signal intensity.</bold></p></caption><graphic xlink:href="1477-7819-11-308-1"/></fig><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Patients characteristics</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"> </th><th align="left"><bold>Patient 1</bold></th><th align="left"><bold>Patient 2</bold></th><th align="left"><bold>Patient 3</bold></th><th align="left"><bold>Patient 4</bold></th><th align="left"><bold>Patient 5</bold></th><th align="left"><bold>Patient 6</bold></th><th align="left"><bold>Patient 7</bold></th><th align="left"><bold>Patient 8</bold></th><th align="left"><bold>Patient 9</bold></th><th align="left"><bold>Patient 10</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom"><bold>Age/Gender</bold><hr/></td><td align="left" valign="bottom">30/F<hr/></td><td align="left" valign="bottom">18/F<hr/></td><td align="left" valign="bottom">21/F<hr/></td><td align="left" valign="bottom">18/F<hr/></td><td align="left" valign="bottom">62/F<hr/></td><td align="left" valign="bottom">50/F<hr/></td><td align="left" valign="bottom">40/F<hr/></td><td align="left" valign="bottom">33/F<hr/></td><td align="left" valign="bottom">71/F<hr/></td><td align="left" valign="bottom">45/M<hr/></td></tr><tr><td align="left" valign="bottom"><bold>Operation</bold><hr/></td><td align="left" valign="bottom">Distal pancreatectomy + splenectomy<hr/></td><td align="left" valign="bottom">Distal pancreatectomy + splenectomy<hr/></td><td align="left" valign="bottom">Total mass exicion<hr/></td><td align="left" valign="bottom">Spleen preser - ving distal pancrea -tectomy<hr/></td><td align="left" valign="bottom">Distal pancreatectomy + splenectomy<hr/></td><td align="left" valign="bottom">Subtotal distal pancreatectomy + splenectomy<hr/></td><td align="left" valign="bottom">Total pancrea- tectomy<hr/></td><td align="left" valign="bottom">Distal pancreatectomy + splenectomy<hr/></td><td align="left" valign="bottom">Subtotal distal pancreatectomy + splenectomy<hr/></td><td align="left" valign="bottom">Spleen preser - ving distal pancrea -tectomy<hr/></td></tr><tr><td align="left" valign="bottom"><bold>Tumor location</bold><hr/></td><td align="left" valign="bottom">Body<hr/></td><td align="left" valign="bottom">Tail<hr/></td><td align="left" valign="bottom">Head<hr/></td><td align="left" valign="bottom">Tail<hr/></td><td align="left" valign="bottom">Tail<hr/></td><td align="left" valign="bottom">Body + Tail<hr/></td><td align="left" valign="bottom">Head + Body<hr/></td><td align="left" valign="bottom">Tail<hr/></td><td align="left" valign="bottom">Body + Tail<hr/></td><td align="left" valign="bottom">Tail<hr/></td></tr><tr><td align="left" valign="bottom"><bold>Size(cm)</bold><hr/></td><td align="left" valign="bottom">9x7x5<hr/></td><td align="left" valign="bottom">13x9x6.5<hr/></td><td align="left" valign="bottom">13x6x5.5<hr/></td><td align="left" valign="bottom">6.5x5.7x3.6<hr/></td><td align="left" valign="bottom">4.5x3.5x3.2<hr/></td><td align="left" valign="bottom">12.5x11x6<hr/></td><td align="left" valign="bottom">4.3x3x3<hr/></td><td align="left" valign="bottom">4x3x2<hr/></td><td align="left" valign="bottom">11x7x7<hr/></td><td align="left" valign="bottom">3x3x2<hr/></td></tr><tr><td align="left" valign="bottom"><bold>İnvasion</bold><hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">Capsule and spleen<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">Capsule<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">Capsule<hr/></td><td align="left" valign="bottom">(-)<hr/></td></tr><tr><td align="left" valign="bottom"><bold>Nodal status</bold><hr/></td><td align="left" valign="bottom">0/14<hr/></td><td align="left" valign="bottom">0/10<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">0/6<hr/></td><td align="left" valign="bottom">0/14<hr/></td><td align="left" valign="bottom">0/4<hr/></td><td align="left" valign="bottom">0/7<hr/></td><td align="left" valign="bottom">0/11<hr/></td><td align="left" valign="bottom">0/5<hr/></td></tr><tr><td align="left"><bold>Follow-up</bold></td><td align="left">Healthy</td><td align="left">Healthy</td><td align="left">Healthy</td><td align="left">Healthy</td><td align="left">Healthy</td><td align="left">7th month Liver and omental Metastasis 9th month exitus</td><td align="left">29th day biliary and pancrea-tic fistula, 41th day exitus</td><td align="left">Healthy</td><td align="left">20th month Liver and omental Metastasis 24th month exitus</td><td align="left">Healthy</td></tr></tbody></table></table-wrap><p>In eight cases lymph node dissection was done in a number between 4 and 14, whereas no dissection was needed for two patients. No lymph node metastasis was present in any patient. Macroscopically, there was diffuse hemorrhage and minimal necrosis between solid and cystic areas (Figure <xref ref-type="fig" rid="F2">2</xref>). At histopathological examination, tumor mass separated from pancreas with a fibrous capsula was seen. Pseudopapillary, cystic and solid growth patterns were seen in the tumor mass. Tumor cells had an ovally shaped, small and centrally localized nucleus and large eosinophilic cytoplasm. Tumors consisted of pseudopapillary structures made of cells aligned around fine vessels, solid areas, hemorrhagic areas and cystic areas of different size (Figure <xref ref-type="fig" rid="F3">3</xref>). No mitosis was seen in five cases, whereas minimal mitosis was present in two cases (2/10 per high powered field) and multiple mitosis were present in two cases (20/10 per high powered field; case numbers 6 and 9) (Table <xref ref-type="table" rid="T2">2</xref>). The immunohistochemistry profiles are summarized in Table <xref ref-type="table" rid="T3">3</xref>. Capsular invasion was present in three cases (case numbers 2, 6 and 9), spleen invasion was also present in case number 2. Along with capsular invasion, mitosis, nuclear polymorphism and necrosis were also significant in case numbers 6 and 9 at the time of diagnosis. These two cases were considered as malignant SPN and treated with six courses of gemcitabine + cis-platinum chemotherapy. Multiple liver and omentum metastases developed in case number 2 at the seventh postoperative month; this patient died at the ninth month. Multiple liver and omentum metastases developed in case number 9 at the 20th postoperative month and she died at the 24th month. The other eight cases have been followed up closely for an average of 7.9 years (between 1 and 16 years) and no recurrence or metastasis has been seen.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Macroscopic appearance of distal pancreatectomy</bold> <bold>+ splenectomy specimen by SPN showing the solid and cystic component with hemorrhagic areas.</bold> SPN, solid pseudopapillary neoplasia.</p></caption><graphic xlink:href="1477-7819-11-308-2"/></fig><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Histologic appearance of solid pseudopapillar tumors. (A)</bold> Solid pseudopapillar tumors exhibit a pseudopapillary pattern.<bold> (B)</bold> A portion of the tumor tissue shows a collection of hyaline globules. <bold>(C)</bold> Tumor cells typically show strong immunoreactivity for vimentin in the cytoplasm. <bold>(D)</bold> CD56 shows positive cytoplasmic membranous staining.</p></caption><graphic xlink:href="1477-7819-11-308-3"/></fig><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Histopathologic features</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"> </th><th align="left"><bold>Patient 1</bold></th><th align="left"><bold>Patient 2</bold></th><th align="left"><bold>Patient 3</bold></th><th align="left"><bold>Patient 4</bold></th><th align="left"><bold>Patient 5</bold></th><th align="left"><bold>Patient 6</bold></th><th align="left"><bold>Patient 7</bold></th><th align="left"><bold>Patient 8</bold></th><th align="left"><bold>Patient 9</bold></th><th align="left"><bold>Patient 10</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Development pattern<hr/></td><td align="left" valign="bottom">Solid, cystic, papillary<hr/></td><td align="left" valign="bottom">Cystic, papillary<hr/></td><td align="left" valign="bottom">Cystic, solid, papillary<hr/></td><td align="left" valign="bottom">Cystic, papillary<hr/></td><td align="left" valign="bottom">Cystic, solid, papillary<hr/></td><td align="left" valign="bottom">Solid, cystic, papillary<hr/></td><td align="left" valign="bottom">Solid, cystic, papillary<hr/></td><td align="left" valign="bottom">Solid, cystic, papillary<hr/></td><td align="left" valign="bottom">Solid, cystic, papillary<hr/></td><td align="left" valign="bottom">Cystic + papillary<hr/></td></tr><tr><td align="left" valign="bottom">Necrosis<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">minimal<hr/></td><td align="left" valign="bottom">minimal<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(-)<hr/></td></tr><tr><td align="left" valign="bottom">Mitosis<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">2/10 per HPF<hr/></td><td align="left" valign="bottom">2/10 per HPF<hr/></td><td align="left" valign="bottom">2/10 per HPF<hr/></td><td align="left" valign="bottom">20/10 per HPF<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">20/10 per HPF<hr/></td><td align="left" valign="bottom">(-)<hr/></td></tr><tr><td align="left">Pleomorphism</td><td align="left">minimal</td><td align="left">(-)</td><td align="left">minimal</td><td align="left">minimal</td><td align="left">minimal</td><td align="left">manifest</td><td align="left">minimal</td><td align="left">(-)</td><td align="left">manifest</td><td align="left">(-)</td></tr></tbody></table></table-wrap><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>The immunohistochemistry study</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"> </th><th align="left"><bold>Patient 1</bold></th><th align="left"><bold>Patient 2</bold></th><th align="left"><bold>Patient 3</bold></th><th align="left"><bold>Patient 4</bold></th><th align="left"><bold>Patient 5</bold></th><th align="left"><bold>Patient 6</bold></th><th align="left"><bold>Patient 7</bold></th><th align="left"><bold>Patient 8</bold></th><th align="left"><bold>Patient 9</bold></th><th align="left"><bold>Patient 10</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Cytokeratin<hr/></td><td align="left" valign="bottom">Focal (+)<hr/></td><td align="left" valign="bottom">Focal (+)<hr/></td><td align="left" valign="bottom">Focal (+)<hr/></td><td align="left" valign="bottom">Focal (+)<hr/></td><td align="left" valign="bottom">Focal (+)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">Focal (+)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td></tr><tr><td align="left" valign="bottom">CEA<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td></tr><tr><td align="left" valign="bottom">Vimentin<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">Focal (+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td></tr><tr><td align="left" valign="bottom">Chromogranin<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">Focal strong (+)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td></tr><tr><td align="left" valign="bottom">Neuron specific enolase<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">Focal slight (+)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td></tr><tr><td align="left" valign="bottom">CD10<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">Slight (+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">Focal (+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td></tr><tr><td align="left" valign="bottom">CD56<hr/></td><td align="left" valign="bottom">Slight (+)<hr/></td><td align="left" valign="bottom">Slight (+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td></tr><tr><td align="left" valign="bottom">Synaptophysin<hr/></td><td align="left" valign="bottom">Focal (+)<hr/></td><td align="left" valign="bottom">Focal (+)<hr/></td><td align="left" valign="bottom">Focal (+)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">Slight focal (+)<hr/></td><td align="left" valign="bottom">Slight focal (+)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td></tr><tr><td align="left" valign="bottom">P53<hr/></td><td align="left" valign="bottom">Focal slight (+)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">10% (+)<hr/></td><td align="left" valign="bottom">5% (+)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">10% (+)<hr/></td><td align="left" valign="bottom">(-)<hr/></td></tr><tr><td align="left" valign="bottom">Kİ67<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">Under 1% (+)<hr/></td><td align="left" valign="bottom">10% (+)<hr/></td><td align="left" valign="bottom">3% (+)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">10% (+)<hr/></td><td align="left" valign="bottom">(-)<hr/></td></tr><tr><td align="left" valign="bottom">Progesterone<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(+)<hr/></td><td align="left" valign="bottom">(-)<hr/></td><td align="left" valign="bottom">(+)<hr/></td></tr><tr><td align="left">EMA</td><td align="left">Rare (+)</td><td align="left">Rare cells (+)</td><td align="left">Rare cells (+)</td><td align="left">Rare cells (+)</td><td align="left">(-)</td><td align="left">Rare cells (+)</td><td align="left">(-)</td><td align="left">(-)</td><td align="left">(-)</td><td align="left">(-)</td></tr></tbody></table></table-wrap></sec><sec sec-type="discussion"><title>Discussion</title><p>SPN is very rare; in fact, they only constitute about 5% of cystic pancreatic tumors and about 1 to 2% of exocrine pancreatic neoplasms [<xref ref-type="bibr" rid="B3">3</xref>]. They present mainly in the second and third decades of life [<xref ref-type="bibr" rid="B4">4</xref>]. Our series presented with a median age of 38.8 years, which is significantly older than in the literature (median age of 26 years) [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B6">6</xref>]. The origin of solid pseudopapillary tumors still remains unclear. These neoplasms have been suggested to have a ductal epithelial, neuroendocrine, multipotent primordial cell, or even an extra-pancreatic genital ridge angle-related cell origin [<xref ref-type="bibr" rid="B7">7</xref>].</p><p>The clinical presentation of the tumor is usually nonspecific. Abdominal discomfort or vague pain is the most common symptom, followed by a gradually enlarging mass and compression signs induced by the tumor. Some patients are completely asymptomatic, with the tumor detected incidentally by imaging studies or routine physical examination. Usually there is no evidence of pancreatic insufficiency, abnormal liver function tests, cholestasis, elevated pancreatic enzymes or an endocrine syndrome. Tumor markers are also generally unremarkable [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B8">8</xref>]. In our series, seven patients (70%) presented with abdominal pain or abdominal dullness, three patients (30%) were asymptomatic with the diagnosis made by an incidental finding on routine examination and preoperative tumor markers (AFP, CEA, CA 19–9 and CA 125) were within normal limits in all patients.</p><p>SPN can occur in every part of the pancreas but they are slightly more common in the tail [<xref ref-type="bibr" rid="B3">3</xref>]. Grossly, it appears as a large and encapsulated mass, generally well-demarcated from the remaining pancreas. In fact, invasion of the adjacent organs, such as the spleen or the duodenal wall, is rare. Depending on the tumor position (head, body or tail of the pancreas), the differential diagnosis includes adrenal mass, pancreatic endocrine tumor, liver cyst or tumor, or a pseudocyst [<xref ref-type="bibr" rid="B9">9</xref>].</p><p>Abdominal ultrasound and CT show a well encapsulated, complex mass with both solid and cystic components and displacement of nearby structures. There may be calcifications at the periphery of the mass and intravenous contrast enhancement inside the mass suggesting hemorrhagic necrosis [<xref ref-type="bibr" rid="B10">10</xref>]. However, when compared with MR imaging, CT has inherent limitations in showing certain tissue characteristics, such as hemorrhage, cystic degeneration, or the presence of a capsule. These features may, as shown at pathology, be suggestive of specific lesions such as SPN of the pancreas. Therefore, MR imaging may further aid in showing these characteristics and in the differential diagnosis of complex cystic masses within the pancreas [<xref ref-type="bibr" rid="B11">11</xref>]. Despite the technological improvements, preoperative diagnosis is difficult because of the similarity of findings among cystic lesions. Some studies advocate preoperative endosonography guided fine-needle aspiration biopsy for preoperative detection of the tumor, but this may not be accepted by others because of the uncertainty in diagnosis and the possible tumor spread [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B13">13</xref>]. In our series, preoperative endosonography guided fine-needle aspiration biopsy was performed in one out of ten patients and histology confirmed SPN.</p><p>In approximately 85% of the patients, SPN is limited to the pancreas, while about 10% to 15% of tumors have already metastasized at the time of presentation [<xref ref-type="bibr" rid="B14">14</xref>]. The most common sites for metastasis are the liver, regional lymph nodes, mesentery, omentum and peritoneum.</p><p>Once the diagnosis of SPN is made, surgery is the first choice of treatment. SPN is usually surrounded by a pseudocapsule and exhibits benign or low-grade malignancy. Conservative resection with preservation of as much pancreatic tissue as possible is the treatment of choice. According to the location of the tumor, distal pancreatectomy with or without splenectomy, pylorous preserving pancreatoduodenectomy, Whipple operation or enucleation can be performed. In our series, four patients underwent distal pancreatectomy with splenectomy, one patient underwent a total mass excision and one patient underwent total pancreatic resection. Two required extended distal pancreatectomy with splenectomy. Two underwent spleen-preserving distal pancreatectomy. Many studies have demonstrated that less aggressive surgical procedures could be preferred for the treatment of SPN [<xref ref-type="bibr" rid="B15">15</xref>]. Extensive lymphatic dissection or more radical approaches are not indicated when the disease is localized. Local invasion and metastases are not contraindications for resection. Portal vein resection is advocated when there is evidence of tumor invasion. For the metastases, surgical debulking should be performed, in contrast to other pancreatic malignancies. Metastases can be removed with enucleations or lobectomies and some patients with unresectable SPN may also have a long term survival [<xref ref-type="bibr" rid="B14">14</xref>]. The overall five-year survival rate of patients with SPN is about 95% [<xref ref-type="bibr" rid="B8">8</xref>].</p><p>Malignant SPN, designated as a solid-pseudopapillary carcinoma, occurs in 15% of adult patients. According to the WHO classification system, these are: 1) solid-pseudopapillary neoplasms with borderline malignancy potential; and 2) solid-pseudopapillary carcinomas. Criteria which distinguish potentially malignant tumors and which are classified as ‘SP carcinoma’ are: 1) angioinvasion; 2) perineural invasion; and 3) deep invasion of the surrounding pancreatic parenchyma. A recent study showed that some histological features, such as extensive necrosis, nuclear atypia, high mitotic rate, immunohistochemistry findings of expression of Ki-67 and sarcomatoid areas may be associated with aggressive behavior [<xref ref-type="bibr" rid="B16">16</xref>].</p><p>Adjuvant therapy is used only in a small number of patients because of the high resectability of SPN. The role of chemotherapy or chemoradiotherapy in the treatment of SPN is also unclear. In some studies, adjuvant chemotherapy and radiotherapy are reported in some unresectable cases with good results [<xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B18">18</xref>]. Neoadjuvant chemotherapy or chemoradiotherapy is also reported to have been successful in a few cases [<xref ref-type="bibr" rid="B19">19</xref>-<xref ref-type="bibr" rid="B22">22</xref>].</p><p>In the light of previous studies, our two patients (patients number 6 and 9) had capsular invasion besides significant mitosis (20/10 per HPF), nuclear pleomorphism and necrosis at the time of diagnosis and Ki-67 index was 10% (+). These two patients were accepted as having malignant SPN. They were given gemcitabine + cis-platinum chemotherapy. Multiple liver and omentum metastases developed in case number 2 at the seventh postoperative month; she died at the ninth postoperative month. Multiple liver and omentum metastases developed in case number 9 at the 20th postoperative month and she died at the 24th postoperative month.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>SPN is a rare neoplasm that primarily affects young women. The prognosis is favorable even in the presence of distant metastasis. Although surgical resection is generally curative, a close follow-up is advised in order to diagnose a local recurrence or distant metastasis and choose the proper therapeutic option for the patient.</p></sec><sec><title>Abbreviations</title><p>SPN: Solid pseudopapillary neoplasia; HPF: High-power fields; CT: Abdominal computed tomography; MRI: Magnetic resonance imaging.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>AY and SY participated in the data acquisition, data analysis, literature review and drafted the manuscript of this article. AC, NE and MY planned the analysis, participated in data acquisition, data analysis, literature review, patient treatment, and drafting and critical revision of the manuscript. EY and HP participated in immunohistochemisty and data analysis. All authors read and approved the final manuscript.</p></sec>
Turn-over rate of academic faculty at the College of Health Sciences, Addis Ababa University: a 20-year analysis (1991 to 2011)	<sec><title>Background</title><p>Faculty turn-over affects both workers and organizations. Turnover of faculty and researchers is increasing alarmingly and costing the universities and the country at large. Fast turnover of health professionals from the health system and from academic institutions has recently received substantial attention from both academia and health sector managers. This paper calculates the faculty turnover rate at the College of Health Sciences of Addis Ababa University during the period of September 1991 to August 2011.</p></sec><sec><title>Methods</title><p>The study was conducted at the College of Health Sciences, Addis Ababa University. Retrospective analysis of employee records was done. All records of the faculty that were working in the College during the 20-year period, starting from September 1991 to August 2011 were retrospectively reviewed. Data were collected from the employee records accessed from the College’s human resources database and supplemented by payroll sheets and different reports. A structured checklist was used to extract the required data from the database. The crude turnover rate for academic faculty was calculated.</p></sec><sec><title>Results</title><p>Within the 20-year period of September 1991 to August 2011, a total of 120 faculty members left. The overall turn-over rate was 92.8 %. The rate in the most recent five years (172 %) is 8.5 times higher than the rate for the first five years (20 %). The average retention period before the termination of an employment contract was 4.9 years. The top five departments where employment contracts were relatively higher include: Nursing 15 (15.6 %), Internal Medicine 12 (12.5%), Public Health 10 (10.4%), Pediatrics 9 (9.4%) and Surgery 9 (9.4%). About two thirds (66.6%) of the faculty who were leaving were at the ranks of assistant professorship and above.</p></sec><sec><title>Conclusion</title><p>This study revealed that outflow of faculty has been continuously increasing in the period reviewed. This implies that the College had been losing highly skilled professionals with considerably higher costs in monetary terms. In this regard, an urgent response is required to retain or significantly decrease the outflow of faculty. Different motivation and retention mechanisms should be identified and implemented. Various modalities of faculty development programs should also be initiated.</p></sec>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Hailu</surname><given-names>Alemayehu</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I3">3</xref><email>alemayehu4all@gmail.com</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Mariam</surname><given-names>Damen Haile</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I3">3</xref><email>damen_h@hotmail.com</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Fekade</surname><given-names>Daniel</given-names></name><xref ref-type="aff" rid="I2">2</xref><xref ref-type="aff" rid="I3">3</xref><email>danielfekade127@gmail.com</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Derbew</surname><given-names>Miliard</given-names></name><xref ref-type="aff" rid="I2">2</xref><xref ref-type="aff" rid="I3">3</xref><email>milliardderbew@gmail.com</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Mekasha</surname><given-names>Amha</given-names></name><xref ref-type="aff" rid="I2">2</xref><xref ref-type="aff" rid="I3">3</xref><email>amekashaw@yahoo.com</email></contrib>	Human Resources for Health	<sec><title>Background</title><p>Employees are not “owned” by organizations like any other asset and, as such, staff turnover is a reality for many organizations. It is natural and healthy for people to leave the organization from time to time as this allows for the introduction of fresh ideas and innovations, flexible career opportunities, and enhances satisfaction in the workplace [<xref ref-type="bibr" rid="B1">1</xref>]. On the other hand, unless organizations retain workers for a reasonable period, they are unlikely to be able to provide the quality services required to remain competitive [<xref ref-type="bibr" rid="B2">2</xref>].</p><p>The fate of a university depends on its ability to recruit and retain talented faculty members [<xref ref-type="bibr" rid="B2">2</xref>]. Faculty turnover is a pervasive feature of the employment market. Fast turnover of academicians affects both the faculties and the university. For faculty members that leave their employment, it can not be easy to learn new job-specific skills and find different career prospects. Universities will also lose job-specific skills, which will be disruptive to their teaching/learning, as well as to their service rendering processes [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B3">3</xref>]. Subsequently, fast turn-over of faculty increases the cost that the universities incur in their human resources development activities [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B5">5</xref>].</p><p>Currently, turnover of faculty and researchers is increasing alarmingly within the universities in Ethiopia. At the country level, the situation is aggravated by the high rate of brain drain. For example, in the 1960s and 1970s, faculty sent for study leave abroad came back immediately after completing their studies to return to their appointments. In recent times, however, the opposite is the case. A sizable number of Ethiopian academics have migrated abroad in search of better conditions [<xref ref-type="bibr" rid="B6">6</xref>-<xref ref-type="bibr" rid="B9">9</xref>].</p><p>Moreover, this fast turn-over of faculty as well as different professionals from academic institutions has not yet received due attention from both academics and health sector managers. Minimal, if any, effort has been given to investigate and understand the causes of such high turn-over in developing countries. Faculty turn-over is driven by certain identifiable characteristics such as types of workers, tasks, organizations and markets. Evidence elsewhere has revealed that it is possible to significantly reduce the occurrence of turn-over of clinicians and other faculty in schools of health and medical sciences. However, actions to prevent such occurrences are rarely seen [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B10">10</xref>].</p><p>The current analysis is aimed at investigating the magnitude and trend of turn-over of the faculty of the College of Health Sciences (CHS), Addis Ababa University (AAU). The result could also help to proactively act in and facilitate decisions in struggling to achieve quality medical education in Ethiopia.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Study design and data analysis</title><p>This study is a retrospective analysis of the official records of employees. There are a number of ways to measure the rate of employee turn-over. For this study, the crude turn-over rate (CTR) is used. Crude turn-over rate is calculated based on a formula that is the number of those leaving in a given period divided by the average number of faculty members during the same period multiplied by 100. The rate was calculated separately for each consecutive five years using the same formula. The numerator includes all people leaving, even people who left involuntarily due to dismissal, redundancy or retirement.</p><p>Frequency and percentage was calculated to characterize the study subjects based on their socio-demographics, academic rank and their respective departments. The faculty turn-over rate was calculated for each five years starting from September 1991 to August 2011. Length of stay was computed to identify the magnitude of difference in longevity within the College. Early leavers were those who stayed for less than four years; middle leavers were those who stayed for four to eight years, and later leavers were those with longevity of more than eight years.</p></sec><sec><title>Study subjects and data sources</title><p>The study was conducted in the CHS, AAU. The CHS was established in 1964 as a Faculty of Medicine with the goal of producing medical doctors to handle the country’s health problems. The Faculty of Medicine is the oldest and the largest among the health training institutions in the country, staffed with the most senior specialists in the country.</p><p>A retrospective cohort of employed faculty working at the CHS was followed for 20 years starting from September 1991 to August 2011 (<italic>Meskerem</italic> 1984 to <italic>Pagume</italic> 2003 in the Ethiopian Calendar). The term ‘faculty’ means the staff members with academic positions of lecturer and above academic rank. Data were collected from the records of the employees accessed from the College’s human resources database, payroll sheets and other relevant reports. The checklist was prepared to extract the required data from the database. Responsible people in the college were informed by a formal letter about this study and the required data. Any identification of the employee was not recorded anywhere on the checklist and appropriate measures were also taken to ensure confidentiality of information.</p></sec></sec><sec sec-type="results"><title>Results</title><sec><title>Currently working faculty</title><p>Table  <xref ref-type="table" rid="T1">1</xref> displays the distribution of currently working male and female faculty across different departments within the College. Out of a total of 253 faculty members in the College, the vast majority (85.4%) were males. More than 50% of the total faculty members were employed by 5 of the 18 departments within the College: Departments of Surgery 42 (16.2%), Internal Medicine 34 (13.4%), School of Nursing 25 (9.9%), Public Health 21 (9.1%) and Pediatrics and Child Health 16 (6.3%). Departments of Obstetrics/Gynecology, Microbiology and Psychiatry each had 12 (4.7%) academic faculty members.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Distribution of faculties by sex and their departments, September 2011, Addis Ababa</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left" valign="bottom"> <hr/></th><th align="left" valign="bottom"> <hr/></th><th colspan="4" align="center" valign="bottom"><bold>Academic rank</bold><hr/></th><th colspan="2" align="center" valign="bottom"><bold>Total</bold><hr/></th></tr><tr><th align="left"> </th><th align="left"> </th><th align="left"><bold>Profs.</bold></th><th align="left"><bold>Asso. Profs</bold></th><th align="left"><bold>Assi. Profs</bold></th><th align="left"><bold>Lec. s</bold></th><th align="left"><bold>N</bold></th><th align="left"><bold>%</bold></th></tr></thead><tbody valign="top"><tr><td rowspan="2" align="left" valign="bottom"><bold> <italic>Sex</italic> </bold><hr/></td><td align="left" valign="bottom">Female<hr/></td><td align="left" valign="bottom">1<hr/></td><td align="left" valign="bottom">4<hr/></td><td align="left" valign="bottom">18<hr/></td><td align="left" valign="bottom">14<hr/></td><td align="left" valign="bottom">37<hr/></td><td align="left" valign="bottom">14.6<hr/></td></tr><tr><td align="left" valign="bottom">Male<hr/></td><td align="left" valign="bottom">17<hr/></td><td align="left" valign="bottom">38<hr/></td><td align="left" valign="bottom">127<hr/></td><td align="left" valign="bottom">34<hr/></td><td align="left" valign="bottom">216<hr/></td><td align="left" valign="bottom">85.4<hr/></td></tr><tr><td align="left" valign="bottom"><bold> <italic>Department</italic> </bold><hr/></td><td align="left" valign="bottom">Surgery<hr/></td><td align="left" valign="bottom">2<hr/></td><td align="left" valign="bottom">5<hr/></td><td align="left" valign="bottom">29<hr/></td><td align="left" valign="bottom">5<hr/></td><td align="left" valign="bottom">41<hr/></td><td align="left" valign="bottom">16.2<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Internal medicine<hr/></td><td align="left" valign="bottom">1<hr/></td><td align="left" valign="bottom">4<hr/></td><td align="left" valign="bottom">24<hr/></td><td align="left" valign="bottom">5<hr/></td><td align="left" valign="bottom">34<hr/></td><td align="left" valign="bottom">13.4<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Nursing<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">4<hr/></td><td align="left" valign="bottom">21<hr/></td><td align="left" valign="bottom">25<hr/></td><td align="left" valign="bottom">9.9<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Public health<hr/></td><td align="left" valign="bottom">3<hr/></td><td align="left" valign="bottom">4<hr/></td><td align="left" valign="bottom">13<hr/></td><td align="left" valign="bottom">3<hr/></td><td align="left" valign="bottom">23<hr/></td><td align="left" valign="bottom">9.1<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Pediatrics<hr/></td><td align="left" valign="bottom">1<hr/></td><td align="left" valign="bottom">2<hr/></td><td align="left" valign="bottom">13<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">16<hr/></td><td align="left" valign="bottom">6.3<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">GYN-OBS<hr/></td><td align="left" valign="bottom">1<hr/></td><td align="left" valign="bottom">2<hr/></td><td align="left" valign="bottom">9<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">12<hr/></td><td align="left" valign="bottom">4.7<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">DMIP<hr/></td><td align="left" valign="bottom">4<hr/></td><td align="left" valign="bottom">3<hr/></td><td align="left" valign="bottom">1<hr/></td><td align="left" valign="bottom">4<hr/></td><td align="left" valign="bottom">12<hr/></td><td align="left" valign="bottom">4.7<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Psychiatry<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">6<hr/></td><td align="left" valign="bottom">6<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">12<hr/></td><td align="left" valign="bottom">4.7<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Radiology<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">3<hr/></td><td align="left" valign="bottom">7<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">10<hr/></td><td align="left" valign="bottom">4.0<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Ophthalmology<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">3<hr/></td><td align="left" valign="bottom">7<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">10<hr/></td><td align="left" valign="bottom">4.0<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Physiology<hr/></td><td align="left" valign="bottom">3<hr/></td><td align="left" valign="bottom">1<hr/></td><td align="left" valign="bottom">3<hr/></td><td align="left" valign="bottom">2<hr/></td><td align="left" valign="bottom">9<hr/></td><td align="left" valign="bottom">3.6<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Dermatology<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">1<hr/></td><td align="left" valign="bottom">7<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">8<hr/></td><td align="left" valign="bottom">3.2<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Pharmacology<hr/></td><td align="left" valign="bottom">1<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">3<hr/></td><td align="left" valign="bottom">4<hr/></td><td align="left" valign="bottom">8<hr/></td><td align="left" valign="bottom">3.2<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Anatomy<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">3<hr/></td><td align="left" valign="bottom">1<hr/></td><td align="left" valign="bottom">3<hr/></td><td align="left" valign="bottom">7<hr/></td><td align="left" valign="bottom">2.8<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Biochemistry<hr/></td><td align="left" valign="bottom">1<hr/></td><td align="left" valign="bottom">1<hr/></td><td align="left" valign="bottom">4<hr/></td><td align="left" valign="bottom">1<hr/></td><td align="left" valign="bottom">7<hr/></td><td align="left" valign="bottom">2.8<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Orthopedics<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">2<hr/></td><td align="left" valign="bottom">5<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">7<hr/></td><td align="left" valign="bottom">2.8<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Anesthesiology<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">6<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">6<hr/></td><td align="left" valign="bottom">2.4<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Pathology<hr/></td><td align="left" valign="bottom">1<hr/></td><td align="left" valign="bottom">2<hr/></td><td align="left" valign="bottom">3<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">6<hr/></td><td align="left" valign="bottom">2.4<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Total <italic>N</italic><hr/></td><td align="left" valign="bottom">18<hr/></td><td align="left" valign="bottom">42<hr/></td><td align="left" valign="bottom">145<hr/></td><td align="left" valign="bottom">48<hr/></td><td align="left" valign="bottom">253<hr/></td><td align="left" valign="bottom">100<hr/></td></tr><tr><td align="left"> </td><td align="left">%</td><td align="left">7.1</td><td align="left">16.6</td><td align="left">57.3</td><td align="left">19.0</td><td align="left">100</td><td align="left"> </td></tr></tbody></table></table-wrap><p>In terms of academic rank, more than half (57.3%) of the faculty members were assistant professors, one sixth (16.6%) were associate professors, 7.1% were professors, while the remaining 19% were lecturers. Out of a total of 18 professors in the College, only one is a female.</p><p>The Department of Microbiology had the largest number of professors [<xref ref-type="bibr" rid="B4">4</xref>] compared with other departments in the College, followed by the Department of Physiology [<xref ref-type="bibr" rid="B3">3</xref>] and the School of Public Health [<xref ref-type="bibr" rid="B3">3</xref>]. Of the four major clinical departments, surgery had two professors while the rest had one each. Departments of Biochemistry, Pathology and Pharmacology had one professor each (Table  <xref ref-type="table" rid="T1">1</xref>).</p></sec><sec><title>Faculty turnover</title><p>Within the 20 year period (September 1991 to August 2011), a total of 120 faculty members left for various reasons. Out of these, data were obtained only for 96. The top five departments where higher numbers of employees left include: Nursing, 15 (15.6%), Internal Medicine, 12 (12.5%), Public Health, 10 (10.4%), Pediatrics, 9 (9.4%) and Surgery, 9 (9.4%). On the other hand, the Departments of Anatomy, Dermatology, Physiology and Psychiatry each lost only one faculty member within the specified period (Figure  <xref ref-type="fig" rid="F1">1</xref>).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p>Frequency distribution of academic staff leaving from different departments (1991 to 2011).</p></caption><graphic xlink:href="1478-4491-11-61-1"/></fig><p>Out of the 120 faculty members who were leaving, data on academic rank were obtained for 106 of them. Accordingly, the last academic rank achieved before the termination of the employment contract was assistant professor for more than half (58.5%), while 13.2% of them were associate professors during the termination of their employment contracts (Figure  <xref ref-type="fig" rid="F2">2</xref>).</p><fig id="F2" position="float"><label>Figure 2</label><caption><p>Distribution of leaving faculties across different academic ranks (1991 to 2011).</p></caption><graphic xlink:href="1478-4491-11-61-2"/></fig><p>Among leavers, the mean longevity at the College was 4.9 years (with a standard deviation of 3.8 years and a range of 1 month to 18 years) (Table  <xref ref-type="table" rid="T2">2</xref>).</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Sex and years of service of the staff members leaving (1991 to 2011)</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"> </th><th align="left"><bold>N</bold></th><th align="left"><bold>%</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom"><bold>Sex</bold><hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Male<hr/></td><td align="left" valign="bottom">107<hr/></td><td align="left" valign="bottom">89.2<hr/></td></tr><tr><td align="left" valign="bottom">Female<hr/></td><td align="left" valign="bottom">13<hr/></td><td align="left" valign="bottom">10.8<hr/></td></tr><tr><td align="left" valign="bottom"><bold>Service year (n = 80, 21 = missing)</bold><hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">>10<hr/></td><td align="left" valign="bottom">6<hr/></td><td align="left" valign="bottom">5.9<hr/></td></tr><tr><td align="left" valign="bottom">(9 to 10)<hr/></td><td align="left" valign="bottom">6<hr/></td><td align="left" valign="bottom">5.9<hr/></td></tr><tr><td align="left" valign="bottom">(7 to 8)<hr/></td><td align="left" valign="bottom">12<hr/></td><td align="left" valign="bottom">11.9<hr/></td></tr><tr><td align="left" valign="bottom">(5 to 6)<hr/></td><td align="left" valign="bottom">16<hr/></td><td align="left" valign="bottom">15.8<hr/></td></tr><tr><td align="left" valign="bottom">(3 to 4)<hr/></td><td align="left" valign="bottom">11<hr/></td><td align="left" valign="bottom">10.9<hr/></td></tr><tr><td align="left" valign="bottom"><3<hr/></td><td align="left" valign="bottom">29<hr/></td><td align="left" valign="bottom">28.7<hr/></td></tr><tr><td align="left">Total</td><td align="left">120</td><td align="left">100</td></tr></tbody></table></table-wrap><p>Among faculties employed during the overall period of analysis, a total of 120 faculty members were leaving due to different reasons, with an overall 20-year turn-over rate of 92.8%. The turn-over rate for the last five years (172%) was two times higher when compared with that of the preceding five years (Figure  <xref ref-type="fig" rid="F3">3</xref>).</p><fig id="F3" position="float"><label>Figure 3</label><caption><p>Staff turnover rate in the last 20 years: September 1991 to August 2011.</p></caption><graphic xlink:href="1478-4491-11-61-3"/></fig></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>The College of Health Sciences at Addis Ababa University comprises the Schools of Medicine, Pharmacy, Public Health and Allied Health Sciences. The core functions of the College and the institutions under it include teaching, research and community service. According to the findings of the present study, the turn-over rate in the College was very high (92.8%), indicating that nearly all faculty members were replaced with new staff during the 20 years of analysis. The figure for the overall, last 20 years, turnover rates (92.8%) seems that the turnover is stable. A lot of faculty members had been assigned in the early two to five years of the study period due to the expansion of the university. But the reality is quite alarming in the most recent five years (172%). For developing countries like Ethiopia where human resources are limited, and in a situation where the number of senior faculty members is few, the depicted magnitude is quite alarming.</p><p>In terms of academic rank, the majority of the faculty members leaving were assistant professors. Most of the faculty members are medical doctors with specialty training in various fields, indicating that the College is losing highly skilled professionals with cost implications that are devastating. This shows the need for an urgent response to retain faculty or to significantly decrease their outflow.</p><p>Turn-over of faculty in the health and medical sciences is mostly due to a combination of attrition (through long-term illness and death, resignation, retirement, dismissal) and transfers (lateral, promotion, study leave). The main issue in most countries is the high rate of transfers of health professionals from public institutions to private and non-governmental organizations - seeking higher pay and incentives [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B11">11</xref>]. Departments with high turn-over may require more attention and positions with high turn-over may need to be restructured to be more interesting [<xref ref-type="bibr" rid="B11">11</xref>].</p><p>A high level of faculty turn-over could be caused by many other factors. Recruiting and seeking the wrong employees in the first place is the main factor causing fast turn-over in most organizations, and this may apply in the case of the CHS, AAU as well. Poor morale and a low level of motivation within the workforce, and a mismatch among the employee’s personal values, career, goals and plans with the larger corporate culture are also other factors for fast turn-over. A buoyant local labor market offering more attractive opportunities to employees also increases the rate.</p><p>Different studies indicate that salary, gender, age, position/title or academic rank, absences and average number of previous jobs as being significantly associated with whether employees remain within or leaving the organization. Study findings also revealed that higher employee turnover is also affecting the stability of the organization through eliminating skilled professionals from the university [<xref ref-type="bibr" rid="B3">3</xref>].</p><p>This study is only limited to measuring the faculty turnover rate using retrospective human resource data. Our study could not investigate the reasons why the faculty members were terminating their employment contract with the university due to the data limitation we faced. Detailed analysis of turn-over, motivating and demotivation factors, investigation of employee behavior to pinpoint why they leave the University and specifically their departments and what can be done to retain them are required. We also recommend that the employee record should be well organized, documented and digitalized so that it could serve for further analysis.</p></sec><sec sec-type="conclusions"><title>Conclusion</title><p>As has been found in other studies, our findings revealed that retaining high caliber academic staff was a serious challenge to the college of Health Sciences at Addis Ababa University due to different reasons. The average retention period before the termination of the employment contract of faculty members was 4.9 years.</p><p>As this study revealed, the outflow of academic faculty is continuously increasing. Therefore, an urgent response is required to fully retain or balance the outflow, including the consideration of different motivations and retention mechanisms (such as salary increases and some other non-monetary incentives, promotion prospects, provision of job opportunities for spouses, a conducive and friendly atmosphere, overseas attendance at conferences, participating in the university’s community outreach and administrative services and academic promotional opportunities).</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>All authors (AH and DH, MD, DF and AM) participated in the conception of the research idea, design of the study, analysis of the data, interpretation, write-up and manuscript preparation equally. All the authors read and approved the final manuscript.</p></sec>
Friendship networks and physical activity and sedentary behavior among youth: a systematized review	<sec><title>Background</title><p>Low levels of physical activity and increased participation in sedentary leisure-time activities are two important obesity-risk behaviors that impact the health of today’s youth. Friend’s health behaviors have been shown to influence individual health behaviors; however, current evidence on the specific role of friendship networks in relation to levels of physical activity and sedentary behavior is limited. The purpose of this review was to summarize evidence on friendship networks and both physical activity and sedentary behavior among children and adolescents.</p></sec><sec><title>Method</title><p>After a search of seven scientific databases and reference scans, a total of thirteen articles were eligible for inclusion. All assessed the association between friendship networks and physical activity, while three also assessed sedentary behavior.</p></sec><sec><title>Results</title><p>Overall, higher levels of physical activity among friends are associated with higher levels of physical activity of the individual. Longitudinal studies reveal that an individual’s level of physical activity changes to reflect his/her friends’ higher level of physical activity. Boys tend to be influenced by their friendship network to a greater extent than girls. There is mixed evidence surrounding a friend’s sedentary behavior and individual sedentary behavior.</p></sec><sec><title>Conclusion</title><p>Friends’ physical activity level appears to have a significant influence on individual’s physical activity level. Evidence surrounding sedentary behavior is limited and mixed. Results from this review could inform effective public health interventions that harness the influence of friends to increase physical activity levels among children and adolescents.</p></sec>	<contrib contrib-type="author" id="A1"><name><surname>Sawka</surname><given-names>Keri Jo</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>kj.sawka@ucalgary.ca</email></contrib><contrib contrib-type="author" corresp="yes" id="A2"><name><surname>McCormack</surname><given-names>Gavin R</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>gmccorma@ucalgary.ca</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Nettel-Aguirre</surname><given-names>Alberto</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>alberto.nettel-aguirre@albertahealthservices.ca</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Hawe</surname><given-names>Penelope</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>phawe@ucalgary.ca</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Doyle-Baker</surname><given-names>Patricia K</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>pdoyleba@ucalgary.ca</email></contrib>	The International Journal of Behavioral Nutrition and Physical Activity	<sec><title>Background</title><p>Physical activity plays a vital role in the health of children and adolescents [<xref ref-type="bibr" rid="B1">1</xref>]. Along with a high caloric diet, low levels of physical activity and increased participation in sedentary leisure-time activity are two important lifestyle behaviors that have contributed to the increased prevalence of overweight and obesity among youth and adults [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B3">3</xref>]. In children and adolescents, overweight and obesity are associated with an increased risk of high blood pressure, dyslipidemia, impaired glucose tolerance, cardiovascular disease, and type II diabetes [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B5">5</xref>]. Furthermore, overweight children are highly likely to become overweight adults, which may reflect the tracking of obesity-risk behaviors (i.e., physical activity and diet) from childhood into adulthood [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B6">6</xref>].</p><p>The social environment comprises the physical surroundings, social relationships and cultural milieu within which people function and interact [<xref ref-type="bibr" rid="B7">7</xref>]. It has been shown to influence obesity-risk behaviors in adults [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B9">9</xref>]; those reporting low social support from family and friends are more likely to be insufficiently active for health benefits compared to those with high levels of social support [<xref ref-type="bibr" rid="B8">8</xref>]. The social environment also plays an important role in relation to children’s physical activity and sedentary behavior. The social environment of children includes the influence of parents, siblings, friends, neighbors, teachers, and coaches [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B11">11</xref>]. While parents are the most important source of influence in early-life, parental influence on their child’s day-to-day behavior becomes less evident as the child matures [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B13">13</xref>]. Children and adolescents spend a significant portion of their time at school with friends and peers. Evidence suggests that the dietary behavior of a friend or group of friends influences the dietary behavior of the individual [<xref ref-type="bibr" rid="B14">14</xref>], with similar results observed for sports participation [<xref ref-type="bibr" rid="B14">14</xref>] and sedentary behavior [<xref ref-type="bibr" rid="B15">15</xref>].</p><p>The pathways by which behaviors may be similar among groups of friends during childhood, however, are complex. Similar behaviors among friends likely reflect the processes of homophily or selection (i.e., an individual with certain behaviors seeking out others who also share similar behaviors) and peer influence or peer contagion (i.e., the influence of friends’ behaviors causing changes in an individual’s behavior) [<xref ref-type="bibr" rid="B16">16</xref>]. Several mechanisms may explain the processes of peer influence and contagion on physical activity and sedentary behavior including: behavioral modeling (i.e., observing a peer perform a behavior leading to increased motivation to perform a behavior); peer pressure (i.e., direct attempts to impose a certain behavior on a peer); group norms (i.e., the underlying attitudes and behaviors shared among a group of peers), and; co-participation (i.e., undertaking a behavior with a peer potentially contributing to behavioral reinforcement) [<xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B18">18</xref>].</p><p>Social network analysis or sociometry [<xref ref-type="bibr" rid="B19">19</xref>] provides a means of studying the inter-relationships among friends themselves and does not rely on an individual recalling or reporting the behavior of his/her friends or peers. Social network analysis is a quantitative method for assessing the structure and patterns of the ties or relationships among a set of entities (e.g., people or organizations) [<xref ref-type="bibr" rid="B20">20</xref>]. It can provide information about an individual’s local relations (e.g., who he or she is friends with) and network position (e.g., whether he or she is centralized within a given network) as well as measures of the entire network itself (e.g., number of connections between people, and degrees of separation [<xref ref-type="bibr" rid="B16">16</xref>]). In child and adolescent health, social network analysis has been used extensively to investigate behaviors such as smoking, substance use, and delinquency in relation to individual-level network measures [<xref ref-type="bibr" rid="B21">21</xref>-<xref ref-type="bibr" rid="B24">24</xref>]. For example, popularity, or being nominated as a friend by many others, is associated with higher odds of drinking alcohol among thirteen and fifteen year olds [<xref ref-type="bibr" rid="B21">21</xref>], while substance use is associated with receiving fewer friendship nominations [<xref ref-type="bibr" rid="B25">25</xref>]. Smoking [<xref ref-type="bibr" rid="B26">26</xref>], delinquency [<xref ref-type="bibr" rid="B24">24</xref>], substance abuse [<xref ref-type="bibr" rid="B21">21</xref>], and depression [<xref ref-type="bibr" rid="B27">27</xref>] studies that have used social network analyses suggest that the attitudes and behaviors of adolescents influence the attitudes and behaviors of others in their friendship networks (i.e., peer contagion). Moreover, the influence of peer contagion might also be gender-specific. Mercken et al. [<xref ref-type="bibr" rid="B28">28</xref>] found that teenage girls, but not boys, were influenced by their peer group to initiate smoking, while delinquent behavior in friends may be more influential in boys than girls [<xref ref-type="bibr" rid="B29">29</xref>].</p><p>Regarding physical activity, some evidence derived from social network analysis suggests that higher physical activity levels within friendship groups could be associated with higher levels of participation among individual group members [<xref ref-type="bibr" rid="B30">30</xref>]. Much of this evidence is based on individual-level or ego-network measures (i.e., a direct link between individuals) rather than an individual’s position in the network of a class or school or the characteristics of the networks themselves. Furthermore, similar to other behaviors, there is preliminary support for gender-specific relationships between individual measures of friendship networks and physical activity. Jago et al. [<xref ref-type="bibr" rid="B31">31</xref>] found that moderate-to-vigorous physical activity of boys’ best friends, but not girls’ best friends, was positively associated with an individual’s moderate-to-vigorous physical activity.</p><p>Little is known about how specific network ties (i.e., local relations) and specific network roles (i.e., positions within the network) might influence physical activity and sedentary behaviors among children and adolescents. For example, a non-reciprocated friendship nomination (i.e., person ‘A’ says ‘B’ is my friend, but person ‘B’ does not say ‘A’ is my friend) may have a different influence on behavior compared to a reciprocated nomination. The concept of reciprocation in a friendship network can indicate the presence of strong ties (reciprocated nomination) and weak ties (non-reciprocated nomination) between individuals. Strength of ties may also be related to degree of friendship separation (i.e., friend of a friend) [<xref ref-type="bibr" rid="B32">32</xref>], or intimacy of friendship (i.e., first nominated friend, second nominated friend) [<xref ref-type="bibr" rid="B33">33</xref>]. Specific roles within a network may also influence behavior, such as being an isolate (i.e., no ties to other individuals) or liaison (i.e., providing ties between groups within a network) [<xref ref-type="bibr" rid="B26">26</xref>]. While studies have identified relationships between specific network roles (e.g., isolates) and smoking [<xref ref-type="bibr" rid="B21">21</xref>], as well as network characteristics (e.g., density) and delinquency [<xref ref-type="bibr" rid="B24">24</xref>], these relationships in the physical activity and sedentary behavior literature are still poorly understood. Knowledge of the dynamics of friendship networks in relation to physical activity and sedentary behavior could be useful for informing health promotion interventions within social settings (i.e., schools).</p><p>A recent systematic review found strong similarities between a child or adolescent’s level of physical activity and that of his/her close friends and wider peer group, but limited evidence on the role of social networks in influencing sedentary behavior [<xref ref-type="bibr" rid="B30">30</xref>]. These authors, along with others [<xref ref-type="bibr" rid="B34">34</xref>], suggest that better interventions may come from better understanding of friendship networks and behavior. To do so, however, requires a deeper understanding of the psychology and sociology of networks, such as who should be recruited to interventions and how experiences and messages can be amplified (or diluted) across the group [<xref ref-type="bibr" rid="B35">35</xref>]. School-based, peer-group interventions in drug use lacked this sophistication, with consequent modest or negligible effects [<xref ref-type="bibr" rid="B36">36</xref>].</p><p>The purpose of this review was to expand and reassess the conclusions of a previous synthesis [<xref ref-type="bibr" rid="B30">30</xref>] by undertaking a systematized literature review of studies examining the association between friendship networks and both physical activity and sedentary behavior. A systematized review encompasses several, but not all aspects of a full systematic review [<xref ref-type="bibr" rid="B37">37</xref>]. The objectives of this review were to: 1) examine the association between a friend’s level of physical activity and sedentary behavior and an individual’s levels of physical activity and sedentary behavior; 2) determine if the number of friends a child or adolescent has influences his/her own physical activity or sedentary behavior, and; 3) identify and differentiate the effects of different types of social network measures, for example, network ties and positions, that are potentially associated with physical activity and sedentary behavior, especially as they operate at gender-specific levels.</p></sec><sec sec-type="methods"><title>Method</title><sec><title>Database search and study inclusion</title><p>To identify studies for possible inclusion in our review, seven scientific online databases covering the medical, (MEDLINE, PubMed, CINAHL), kinesiology (SPORTDiscus), education (ERIC), sociology (SocINDEX), and psychology (PsycINFO) fields were searched. Search terms and phrases were combined and reflected the population of interest (i.e. child, preteen, adolescent, student, teen, boy, or girl), the exposure (i.e. social network, friend, peer, or social group), and the outcomes (i.e., physical activity, play, sport, exercise, sedentary, inactivity, or leisure). Searches within each database were restricted to English language, peer-reviewed, and primary studies. No restrictions were placed on year of publication. Databases were searched in June, 2012. Our broad search strategy resulted in 21,354 articles. KJS initially reviewed these titles and removed duplicates, non-journal articles and irrelevant titles. The remaining abstracts (n = 1,676) were reviewed in detail by KJS and a random sub-sample (n = 300) were reviewed by GRM to ensure scientific rigor (88.3% overall agreement).</p><p>Seventy-one articles were identified to undergo a full paper review and were read in detail by KJS and GRM. Studies eligible for this review must have included: children or adolescents aged six to eighteen years of age; a measure of a participant’s friendship network through either friendship nominations (i.e., participant nominating friends from a class list) or friendship rating (i.e., participant indicating whom they prefer to play with most), and; a measure of physical activity or sedentary leisure-time activity (i.e., direct observation, motion monitors, direct or indirect calorimetry, doubly-labeled water, parent proxy, or self-report) for both the participant and the participant’s nominated friends. Studies that utilized a general social support measure (i.e., how often does your best friend encourage you to exercise?) were excluded. We also excluded studies that used participant’s proxy measure of friend’s physical activity or sedentary behavior. This was to ensure that each participant identified his or her friends (whom also participated in the study), and that each participant recorded his or her own level of physical activity and sedentary behavior. Final inclusion of each study was based on consensus of two authors (KJS and GRM). To broaden our search, reference lists from included studies were scanned to further identify potential studies.</p></sec><sec><title>Data extraction and analysis</title><p>From each included study, information regarding study design, sample size, participant characteristics, description of friendship network or friendship rating measure, physical activity and/or sedentary behavior, confounders, and study findings were extracted and tabled. The most robust results from each study were included (e.g., findings based on adjusted estimates would be presented instead of findings based on unadjusted estimates if both were presented within a single study). Factors affecting study validity including sample design, sample size, response rate, control for confounders, and method of physical activity or sedentary behavior measurement were appraised and synthesized, along with study results of the relationships between friendship networks and physical activity and sedentary behavior. Information regarding the use of a theoretical framework or model, where reported, was also extracted from each article.</p></sec></sec><sec sec-type="results"><title>Results</title><p>A total of thirteen studies were included in this review, four [<xref ref-type="bibr" rid="B38">38</xref>-<xref ref-type="bibr" rid="B41">41</xref>] of which were not included in the previous review [<xref ref-type="bibr" rid="B30">30</xref>] (Figure <xref ref-type="fig" rid="F1">1</xref>).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p>Flow diagram of article search and selection.</p></caption><graphic xlink:href="1479-5868-10-130-1"/></fig><sec><title>Characteristics of studies reviewed</title><p>The reviewed studies included children and adolescents ranging from six to eighteen years of age (Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Table S1). One study [<xref ref-type="bibr" rid="B42">42</xref>] included girls only, while the other studies had approximately equal proportions of boys and girls. Eleven studies reported response rates ranging from 58.6% to 93% [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B39">39</xref>,<xref ref-type="bibr" rid="B41">41</xref>-<xref ref-type="bibr" rid="B46">46</xref>]. Of those, six had response rates of 80% or lower [<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B39">39</xref>,<xref ref-type="bibr" rid="B44">44</xref>-<xref ref-type="bibr" rid="B47">47</xref>]. The geographical location of studies included Australia (n = 4), the United States (n = 3), the United Kingdom (n = 2), Canada (n = 1), Estonia (n = 1), Finland (n = 1) and Norway (n = 1). All of the studies occurred within a school or after-school setting.</p><p>Nine studies were cross-sectional [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B41">41</xref>,<xref ref-type="bibr" rid="B42">42</xref>,<xref ref-type="bibr" rid="B44">44</xref>,<xref ref-type="bibr" rid="B46">46</xref>],[<xref ref-type="bibr" rid="B47">47</xref>], while the remaining four were longitudinal [<xref ref-type="bibr" rid="B39">39</xref>,<xref ref-type="bibr" rid="B40">40</xref>,<xref ref-type="bibr" rid="B43">43</xref>,<xref ref-type="bibr" rid="B45">45</xref>]. Length of follow-up time for the longitudinal studies ranged from one to five years. Seven studies measured physical activity using self-administered questionnaire [<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B39">39</xref>,<xref ref-type="bibr" rid="B41">41</xref>,<xref ref-type="bibr" rid="B43">43</xref>,<xref ref-type="bibr" rid="B46">46</xref>,<xref ref-type="bibr" rid="B47">47</xref>], four via accelerometer [<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B40">40</xref>,<xref ref-type="bibr" rid="B44">44</xref>,<xref ref-type="bibr" rid="B45">45</xref>], one via pedometer [<xref ref-type="bibr" rid="B42">42</xref>], and one via face-to-face interview [<xref ref-type="bibr" rid="B14">14</xref>]. Three studies [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B46">46</xref>] also assessed the amount of sedentary leisure-time activities, which included hours per day of watching television and videos, playing video or computer games, or using the Internet. For the participant’s friendship network measure, all but two studies [<xref ref-type="bibr" rid="B45">45</xref>,<xref ref-type="bibr" rid="B47">47</xref>] used participant nominated friends and best friends in their class, grade, school, or after school program. Livesey et al. [<xref ref-type="bibr" rid="B47">47</xref>] asked children to rate how much they liked to interact during play with other children included in the sample, while Ommundsen et al. [<xref ref-type="bibr" rid="B45">45</xref>] used children’s preferences to play and work with other children in the study to create a socio-metric status score for each participant. Further, Strauss and Pollack [<xref ref-type="bibr" rid="B46">46</xref>] measured participant’s five best male and five best female friends, and determined the relationship between this measure of popularity and both sports participation per week and hours of television or video watching per day.</p><p>Twelve studies statistically controlled for at least one confounding variable, while Schofield et al. [<xref ref-type="bibr" rid="B42">42</xref>] did not report controlling for confounders. Across these twelve studies, demographic variables were controlled for, including age and gender. Six studies adjusted for weight status [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B40">40</xref>,<xref ref-type="bibr" rid="B45">45</xref>,<xref ref-type="bibr" rid="B46">46</xref>]. Several studies also adjusted for socioeconomic factors including parent socio-economic status, parent education level, and/or participant pocket money [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B39">39</xref>,<xref ref-type="bibr" rid="B41">41</xref>,<xref ref-type="bibr" rid="B44">44</xref>-<xref ref-type="bibr" rid="B46">46</xref>]. Only three studies [<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B41">41</xref>,<xref ref-type="bibr" rid="B43">43</xref>] explicitly stated the use or application of a theoretical framework or model with regard to their study design or interpretation of findings. De la Haye et al. [<xref ref-type="bibr" rid="B43">43</xref>] used the Theory of Planned Behavior, with particular focus on perceptions of peer (subjective) norms as a key mechanism of peer influence. These authors however, noted that Self-Perception Theory, where an individual becomes aware of their own psychological and emotional states based on the individual’s observation of their own behaviors, might have provided a better explanation of their results. Raudsepp and Viira [<xref ref-type="bibr" rid="B41">41</xref>] used Social Learning Theory, with particular focus on the concept of behavioral modeling to explain their significant findings whereby best friend’s physical activity was positively associated with an individual’s physical activity. Yli–Piipari et al. [<xref ref-type="bibr" rid="B38">38</xref>] applied the expectancy-value model, which emphasizes personal values and expectancies, as a means to help define socialization and friendship interactions and further explain similarities in physical activity behavior among groups of friends.</p><p>In terms of friendship nominations, one study used only reciprocated nominations [<xref ref-type="bibr" rid="B39">39</xref>], while others used both reciprocated and non-reciprocated nominations [<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B40">40</xref>,<xref ref-type="bibr" rid="B42">42</xref>-<xref ref-type="bibr" rid="B44">44</xref>]. Two studies [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B41">41</xref>] did not indicate whether they used reciprocated and or non-reciprocated nominations. For studies that specifically examined popularity (e.g., the number of times a participant was nominated as a friend) or a socio-metric measure (e.g., preference to play with particular individual), reciprocation of a friendship nomination was not needed as this measure is based on how many times a participant was nominated [<xref ref-type="bibr" rid="B45">45</xref>-<xref ref-type="bibr" rid="B47">47</xref>].</p></sec><sec><title>Associations between friendship networks and physical activity</title><p>Of the ten studies [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B38">38</xref>-<xref ref-type="bibr" rid="B44">44</xref>] that measured close friends’ or friendship groups’ physical activity levels, all found some evidence that levels of physical activity among friends was associated with the level of physical activity of the individual (Table <xref ref-type="table" rid="T1">1</xref>)<italic>.</italic></p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Summary of the associations between friendship networks and physical activity and sedentary behavior across reviewed studies</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="center" valign="bottom"> <hr/></th><th align="center" valign="bottom"> <hr/></th><th colspan="3" align="center" valign="bottom"><bold> <italic>Associations with physical activity</italic> </bold><hr/></th><th colspan="3" align="center" valign="bottom"><bold> <italic>Associations with sedentary behavior</italic> </bold><hr/></th></tr><tr><th align="center"> </th><th align="center"> </th><th align="center"><bold>Positive</bold></th><th align="center"><bold>Null</bold></th><th align="center"><bold>Negative</bold></th><th align="center"><bold>Positive</bold></th><th align="center"><bold>Null</bold></th><th align="center"><bold>Negative</bold></th></tr></thead><tbody><tr><td align="center" valign="bottom"><bold>Boys</bold><hr/></td><td align="center" valign="bottom">Close friends<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B15">15</xref>], [<xref ref-type="bibr" rid="B39">39</xref>]<sup>a, b</sup>, [<xref ref-type="bibr" rid="B31">31</xref>], [<xref ref-type="bibr" rid="B41">41</xref>]<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B15">15</xref>], [<xref ref-type="bibr" rid="B31">31</xref>]<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B15">15</xref>]<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B15">15</xref>]<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">Friendship group<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B38">38</xref>]<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">Popularity<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B15">15</xref>]<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B15">15</xref>], [<xref ref-type="bibr" rid="B47">47</xref>]<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B45">45</xref>]<sup>b</sup><hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B15">15</xref>]<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">Friendship selection<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="center" valign="bottom"><bold>Girls</bold><hr/></td><td align="center" valign="bottom">Close friends<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B15">15</xref>], [<xref ref-type="bibr" rid="B39">39</xref>]<sup>a, b</sup>, [<xref ref-type="bibr" rid="B41">41</xref>], [<xref ref-type="bibr" rid="B42">42</xref>]<sup>a</sup><hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B15">15</xref>], [<xref ref-type="bibr" rid="B39">39</xref>]<sup>a</sup> , [<xref ref-type="bibr" rid="B31">31</xref>], [<xref ref-type="bibr" rid="B41">41</xref>], [<xref ref-type="bibr" rid="B42">42</xref>]<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B15">15</xref>]<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B15">15</xref>]<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">Friendship group<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B42">42</xref>], [<xref ref-type="bibr" rid="B38">38</xref>]<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">Popularity<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B45">45</xref>]<sup>b</sup><hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B15">15</xref>], [<xref ref-type="bibr" rid="B47">47</xref>]<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B15">15</xref>]<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B15">15</xref>]<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">Friendship selection<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="center" valign="bottom"><bold>Boys and girls</bold><hr/></td><td align="center" valign="bottom">Close friends<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B43">43</xref>]<sup>b</sup>, [<xref ref-type="bibr" rid="B40">40</xref>]<sup>b</sup><hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">Friendship group<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B14">14</xref>], [<xref ref-type="bibr" rid="B44">44</xref>]<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B44">44</xref>]<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B14">14</xref>]<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">Popularity<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B45">45</xref>]<sup>b</sup>, [<xref ref-type="bibr" rid="B46">46</xref>]<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B40">40</xref>]<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B45">45</xref>]<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B46">46</xref>]<hr/></td></tr><tr><td align="center"> </td><td align="center">Friendship selection</td><td align="center">[<xref ref-type="bibr" rid="B43">43</xref>]<sup>b</sup></td><td align="center">[<xref ref-type="bibr" rid="B40">40</xref>]<sup>b</sup></td><td align="center"> </td><td align="center"> </td><td align="center"> </td><td align="center"> </td></tr></tbody></table><table-wrap-foot><p><italic>Note.</italic> Associations significant at p <.10. All other associations significant at p <.05.</p><p><sup>a</sup> = reciprocated nominations only.</p><p><sup>b</sup> = longitudinal analysis.</p><p><italic>Close friends</italic>: Physical activity or sedentary behavior of nominated best friend or close friends. <italic>Friendship group</italic>: Average physical activity or sedentary behavior of nominated friends. <italic>Popularity</italic>: Higher number of received friendship nominations or a higher measure of friendship rating/status (i.e., number of nominations received for preference to play with). <italic>Friendship selection</italic>: Individual choosing a friend based on similarities with his or her own physical activity or sedentary behavior.</p></table-wrap-foot></table-wrap><sec><title>Popularity, socio-metric status, and physical activity</title><p>Five studies [<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B40">40</xref>,<xref ref-type="bibr" rid="B45">45</xref>-<xref ref-type="bibr" rid="B47">47</xref>] assessed popularity level or socio-metric status, and physical activity level of the individual and found differing results. Strauss and Pollack [<xref ref-type="bibr" rid="B46">46</xref>] found that a higher count of friendship nominations was associated with higher sports participation. This supported De la Haye et al.’s [<xref ref-type="bibr" rid="B15">15</xref>] finding that boys who played more organized physical activity tended also to be the most popular among school friends. In contrast, Gesell et al. [<xref ref-type="bibr" rid="B40">40</xref>] and Livesey et al. [<xref ref-type="bibr" rid="B47">47</xref>] did not find any significant association between popularity level and physical activity among boys and girls. Ommundsen et al. [<xref ref-type="bibr" rid="B45">45</xref>] found that higher total accelerometer counts were correlated with lower socio-metric status in grade one children. Furthermore, in a longitudinal analysis, Ommundsen et al. [<xref ref-type="bibr" rid="B45">45</xref>] found that, for girls, higher total accelerometer counts in grade one were associated with a higher socio-metric status in grade four, while for boys, higher total accelerometer counts in grade one were associated with a lower socio-metric status in grade four.</p><p>Three longitudinal studies [<xref ref-type="bibr" rid="B39">39</xref>,<xref ref-type="bibr" rid="B40">40</xref>,<xref ref-type="bibr" rid="B43">43</xref>] assessed the change in participant’s physical activity level over time, and all found that participants’ level of physical activity significantly changed over time to emulate friends’ higher levels of physical activity. Two longitudinal studies [<xref ref-type="bibr" rid="B40">40</xref>,<xref ref-type="bibr" rid="B43">43</xref>] also examined whether participant’s friendship selection was based on physical activity levels; De la Haye et al. [<xref ref-type="bibr" rid="B43">43</xref>] found that friendship selection was significantly influenced by similarities in physical activity levels, whereas Gesell et al. [<xref ref-type="bibr" rid="B40">40</xref>] did not.</p></sec><sec><title>Network position and physical activity</title><p>Schofield et al. [<xref ref-type="bibr" rid="B42">42</xref>], although not adjusting for other factors, found that a higher pedometer step count for girls’ first nominated <italic>reciprocated</italic> friends was moderately correlated with a high pedometer step count for the individual; however, first <italic>non-reciprocated</italic> friend’s step count was not correlated with an individual’s step count. Moreover, this study also found that the correlation between step count and nominated friends attenuated as friend’s intimacy (i.e., second and third nominated friend) decreased regardless of whether or not the nomination was reciprocated [<xref ref-type="bibr" rid="B42">42</xref>]. Macdonald-Wallis et al. [<xref ref-type="bibr" rid="B44">44</xref>] measured degree of friendship separation, and found that the correlation of moderate-to-vigorous physical activity and counts per minute among friends was strongest with more immediate friendships (i.e., no separation via another person). Beyond nomination reciprocation and degrees of separation, studies did not include measures of local network roles (e.g., isolate, liaison), nor did they examine network-level measures (e.g., density, centrality).</p></sec><sec><title>Gender differences between friendship networks and physical activity</title><p>Six studies [<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B39">39</xref>,<xref ref-type="bibr" rid="B41">41</xref>,<xref ref-type="bibr" rid="B45">45</xref>] reviewed found differences between the influence of friends on physical activity and sedentary behaviors of boys and girls. Boys tended to be more active, and were more likely to be influenced by the physical activity behaviors of their friends compared to girls. For example, Jago et al. [<xref ref-type="bibr" rid="B31">31</xref>] and Raudsepp and Viira [<xref ref-type="bibr" rid="B41">41</xref>] found that boys’ friend’s moderate-to-vigorous physical activity was associated with individual’s moderate-to-vigorous physical activity, but this association was not statistically significant for girls. Denault and Poulin [<xref ref-type="bibr" rid="B39">39</xref>] found that, for boys, a higher level of friend’s sports participation was associated with a higher level of individual sports participation.</p></sec></sec><sec><title>Associations between friendship networks and sedentary behavior</title><p>Three studies [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B46">46</xref>] examined the association between friendship networks and sedentary behavior and found contradicting results (Table <xref ref-type="table" rid="T1">1</xref>). Ali et al. [<xref ref-type="bibr" rid="B14">14</xref>] found no association between the weekly hours of television and video viewing of nominated close friends’ and an individual’s television and video viewing. In contrast, De la Haye et al [<xref ref-type="bibr" rid="B15">15</xref>] found significant positive associations between friends’ video/computer gaming and Internet use and individual’s (girls only) video/computer gaming and internet use in three separate age-based networks (school 1/grade 8; school 2/grade 8; school 2/grade 9). A positive association was also found for boys for the school 2/grade 8 network [<xref ref-type="bibr" rid="B15">15</xref>].</p><sec><title>Popularity, socio-metric status, and sedentary behavior</title><p>Strauss and Pollack [<xref ref-type="bibr" rid="B46">46</xref>] found that as an adolescent’s (boys and girls combined) popularity increased, they spent less time per day watching television.</p></sec><sec><title>Network position and sedentary behavior</title><p>There were no studies that examined differences in reciprocated or non-reciprocated friendships, degree of separation, specific network positions or network characteristics.</p></sec><sec><title>Gender differences between friendship networks and sedentary behavior</title><p>One study stratified their results by gender [<xref ref-type="bibr" rid="B15">15</xref>]. De la Haye et al. [<xref ref-type="bibr" rid="B15">15</xref>] found an association between higher levels of girls’ friends’ video/computer gaming and Internet use and higher levels of individual video/computer gaming and Internet use in all three networks examined. Boys associations were only present in one network [<xref ref-type="bibr" rid="B15">15</xref>]. Contrary to Strauss and Pollack [<xref ref-type="bibr" rid="B46">46</xref>], De la Haye et al. [<xref ref-type="bibr" rid="B15">15</xref>] also identified a small but significant association between a girl’s popularity (i.e., greater count of friendship nominations) and increased level of participation in video/computer gaming and Internet use.</p></sec></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>Friendship networks are associated with physical activity among children and adolescents, with some, albeit less, evidence suggesting that friendship networks might also be associated with sedentary behavior. Our findings confirm evidence from a previous review [<xref ref-type="bibr" rid="B30">30</xref>] which showed that peer networks have a greater influence on physical activity and sedentary behavior for boys compared with girls. This observation is strengthened by more longitudinal evidence, lending weight to the peer contagion models of physical activity (i.e., after becoming friends, behavior become similar) as opposed to the peer selection model (i.e., adolescents choosing friends who have similar behavior to themselves at the outset). This review identified a lack of explicit use of theoretical frameworks in studies to date.</p><p>The differential influence of friendship on physical activity for boys and girls may reflect differences in attitudes towards physical activity and differences in peer social norms [<xref ref-type="bibr" rid="B48">48</xref>]. Moreover, boys generally have higher levels of fitness and physical activity participation compared with girls [<xref ref-type="bibr" rid="B49">49</xref>,<xref ref-type="bibr" rid="B50">50</xref>]. Higher levels of physical activity in and of itself might provide more opportunities for co-participation and modeling (i.e., an individual witnessing another individual being active and may be therefore motivated to participate in the same activity). Another, albeit weaker, explanation could be that the faster rate of maturity among girls, on average, might result in girls developing a more concrete set of values sooner and therefore less likely to conform to group norms [<xref ref-type="bibr" rid="B51">51</xref>]. Gender differences have also been identified for diet, with boys’ friends being more alike in their consumption of high caloric foods than girls’ friends [<xref ref-type="bibr" rid="B15">15</xref>]. This could suggest that gender-specific approaches to promoting healthy weight might be needed, especially if the primary vehicle for the intervention is the friendship network. However, more research is needed to identify which social mechanisms might be more influential in determining physical activity and sedentary behavior for boys and girls.</p><p>Similarities in friendship network behaviors can be both the result of social influence, where children or adolescents adopt behaviors based on the attitudes and behaviors of friends within a network, or a result of friendship selection, whereby individuals select friends that share similar interests, attitudes, and behaviors [<xref ref-type="bibr" rid="B15">15</xref>]. The processes of peer influence and peer selection are found to be associated in other health behaviors in the adolescent population including smoking [<xref ref-type="bibr" rid="B52">52</xref>] and delinquency [<xref ref-type="bibr" rid="B53">53</xref>]. Disentangling these pathways is difficult based on cross-sectional study design, which includes the majority of studies reviewed here. While cross-sectional studies are able to tell us whether a relationship exists between a friendship network and an individual’s behavior, the direction of causality cannot be ascertained. The longitudinal studies in this review offer key information in terms of the influence of friendship networks on physical activity as they allow potential causal pathways to be extricated. Three of these studies [<xref ref-type="bibr" rid="B39">39</xref>,<xref ref-type="bibr" rid="B40">40</xref>,<xref ref-type="bibr" rid="B43">43</xref>] found that an individual’s physical activity level changed over time to become more similar to a friend’s higher level of physical activity, while the fourth longitudinal study [<xref ref-type="bibr" rid="B45">45</xref>] found a positive relationship for girls’ socio-metric status in grade four and accelerometer counts in grade one. These results provide evidence to support a causal pathway, where friends influence an individual’s physical activity level (i.e., peer contagion). This friendship influence could be a result of social norms. Pressure from peers to conform to group norms is a strong motivator for behavior adoption or maintenance, and is often combined with negative consequences, such as social isolation, if behaviors are not adopted [<xref ref-type="bibr" rid="B18">18</xref>]. Future research that assesses reasons for choosing friends will assist in understanding the factors (i.e., friendship selection versus friendship influence) that influence similarities in health behaviors across friendship networks.</p><p>Studies included in this review used mainly ego-based networks, where participants were asked to self identify and nominate their best or close friends; this compared to using complete friendship networks, where participants are given a full class or school list and asked to nominate their friends, thereby allowing the identification of each participant’s role within a friendship network. Previous research has recognized the importance of friendship network roles and characteristics (e.g., density, centrality) in relation to health behaviors in youth [<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B24">24</xref>,<xref ref-type="bibr" rid="B26">26</xref>,<xref ref-type="bibr" rid="B54">54</xref>]. A review by Seo and Huang [<xref ref-type="bibr" rid="B54">54</xref>] found that isolates (i.e., no ties to other individuals [<xref ref-type="bibr" rid="B19">19</xref>]) were more likely to be smokers compared to clique members (i.e., members of a group of at least three individuals, where all three individuals are linked through friendship nominations [<xref ref-type="bibr" rid="B19">19</xref>]), and further identified that non-smoking adolescents were more likely to become smokers if they belong to a smoking clique. There were no studies in our review that investigated the specific roles within a complete friendship network, such as liaisons (i.e., providing ties between groups within a network [<xref ref-type="bibr" rid="B19">19</xref>]) or isolates. Examining the relationship between isolates and physical activity and sedentary behavior may have important health implications, as one study [<xref ref-type="bibr" rid="B46">46</xref>] found that decreased friendship nominations was associated with greater television and video viewing. Furthermore, liaisons are characterized as having a strong degree of interaction among several cliques, and therefore may be a useful mechanism to promote physical activity to a greater number of individuals.</p><p>Studies included in this review did not measure the length of friendship, frequency of friend contact, or context in which friends normally interacted (e.g., playing at recess or after school). The former measures can indicate the strength of bond between two individuals, while the latter measure may have a specific impact on a friend’s influence on sedentary behavior, as sedentary leisure-time activities generally occur outside of the school setting. As well, the level of influence friends have on one another’s behavior might depend on whether the context and activities are organized or non-organized (e.g., sports vs. unstructured play). Stronger bonds, as seen through reciprocated friendship nominations, have a greater impact on physical activity levels as compared to non-reciprocated friends [<xref ref-type="bibr" rid="B42">42</xref>]. Accounting for the quality or strength of friendship bonds in addition to friendship ties may provide greater insight into the mechanisms explaining peer influences on physical activity and sedentary behavior.</p><p>As with any review, the issue of publication bias should be considered when interpreting our findings. This review did not objectively-assess the scientific quality of each included study nor weigh findings based on their validity (i.e., using a validity assessment). Noteworthy, was that only three studies explicitly mentioned the use of a specific theoretical framework or model. Integration of the mechanism of peer selection or contagion within existing social cognitive models of behavior may provide greater understanding regarding peer influence on physical activity and sedentary behavior. At a minimum, future studies should describe the theoretical frameworks informing their methodologies and interpretation of results.</p><p>Despite undertaking a broader search of literature to identify studies, we found only four additional studies not included in a review completed approximately two years ago [<xref ref-type="bibr" rid="B30">30</xref>]. Nevertheless, these additional studies contributed to current knowledge – for example, one study provided additional support for gender differences with regard to peer influence as well as the association between peer influence and physical activity intensity [<xref ref-type="bibr" rid="B41">41</xref>], and two studies provided longitudinal evidence showing emulation of friends physical activity behavior over time [<xref ref-type="bibr" rid="B39">39</xref>,<xref ref-type="bibr" rid="B40">40</xref>]. However, our review identified several gaps in current knowledge, not previously identified, including the lack of evidence regarding the association between specific social network ties, roles, positions, and characteristics and physical activity and sedentary behavior, the dearth of studies incorporating measures strength or quality of peer relationships, the lack of details regarding theoretical frameworks and models, and the need for more longitudinal study designs. Given that there are only thirteen published studies on this topic suggests that our understanding of the role of social networks on physical activity and sedentary behavior among youth is in its early stages and that this topic demands more research attention.</p><p>Findings from this review provide support for a relationship between friend’s physical activity and an individual’s physical activity in children and adolescents, but findings for sedentary behavior are mixed. Harnessing the influence of friendship to increase physical activity levels and decrease sedentary leisure-time activity would have a beneficial impact on reducing the current prevalence of overweight and obese youth through an increase in energy expenditure. More research examining sedentary behavior among children is needed, including investigation of virtual peer networks that result from on-line gaming, as well as the influence of networks outside of the school setting (e.g., family, sports teams, camps, social clubs) on obesity-risk behaviors.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contribution</title><p>KJS/GRM/ANA conceived the study. KJS lead the database search, article selection, synthesis, and drafting of the manuscript. GRM/ANA assisted in article selection and synthesis. All authors contributed to the interpretation of findings and writing of the manuscript. All authors read and approved the final manuscript.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional file 1: Table S1</title><p>Characteristics of reviewed studies.</p></caption><media xlink:href="1479-5868-10-130-S1.docx"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
Brain-State-Dependent Non-Invasive Brain Stimulation and Functional Priming: A Hypothesis	Could not extract abstract	<contrib contrib-type="author"><name><surname>Sergeeva</surname><given-names>Elena G.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="corresp" rid="cor1">*</xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/180201"/></contrib><contrib contrib-type="author"><name><surname>Henrich-Noack</surname><given-names>Petra</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/164984"/></contrib><contrib contrib-type="author"><name><surname>Bola</surname><given-names>Michał</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/84785"/></contrib><contrib contrib-type="author"><name><surname>Sabel</surname><given-names>Bernhard A.</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><uri xlink:type="simple" xlink:href="http://frontiersin.org/people/u/98826"/></contrib>	Frontiers in Human Neuroscience	<p>The aim of using non-invasive brain stimulation techniques in neurorehabilitation is to improve neurological function by modulating brain plasticity in the specific areas of the brain.</p><p>The fundamental idea of current stimulation treatment is that it alters cortical excitability so as to enhance plasticity in subsequent perceptual or motor training. Another goal is to achieve an entrainment of brain oscillations with external currents, which are delivered at certain frequencies to improve the functions by altering physiological activity that outlasts the stimulation period [see for review, Nitsche and Paulus, <xref rid="B7" ref-type="bibr">2011</xref>; Antal and Paulus, <xref rid="B1" ref-type="bibr">2013</xref>].</p><p>However, such approaches often do not consider that the brain is a dynamical system with activity levels and connectivity patterns constantly changing in a highly variable, and so far non-predictable, manner. But it has been known that processing of external (e.g., visual) stimuli depends to some extent on the instantaneous state of brain networks at stimulus onset. Similarly, the effects of the stimulation depend not only on the predefined parameters but also on the state of the brain before and during the stimulation (Silvanto et al., <xref rid="B11" ref-type="bibr">2007</xref>, <xref rid="B10" ref-type="bibr">2008</xref>; Herrmann et al., <xref rid="B5" ref-type="bibr">2013</xref>; Neuling et al., <xref rid="B6" ref-type="bibr">2013</xref>). However, the translation of these findings into clinical practice was so far not realized.</p><p>In this respect a recent study by Gharabaghi et al. (<xref rid="B4" ref-type="bibr">2014</xref>) is of particular interest. These authors explored the possibility of using a brain-state-dependent stimulation (BSDS) approach in post-stroke patients. Here, the subjects were instructed to imagine opening a hand (without actually doing so, moreover, the patient was not capable of hand opening) in order to achieve desynchronization of beta band oscillations within the motor neural circuits. To facilitate the execution of EEG desynchronization, a contingent haptic biofeedback to the hand was provided. Transcranial magnetic pulses were then applied to the motor cortex but only if such desynchronization was achieved, as shown by concurrently recorded EEG.</p><p>Both in a healthy control volunteer and in a patient with severe hemiparesis, BSDS induced a significant increase in excitability of the motor cortex as measured by motor evoked potentials (MEP). Notably, that only BSDS evoked substantial increase of MEP amplitude, while the stimulation pulses applied without the motor-related EEG desynchronization evoked MEP amplitude decrease, though different TMS stimulation paradigms applied independent of the brain state are currently explored to improve motor function after stroke.</p><p>An important aspect of the Gharabaghi et al. study (Gharabaghi et al., <xref rid="B4" ref-type="bibr">2014</xref>) is the fact that the brain stimulation was not applied <italic>prior to</italic> or <italic>alternating with</italic> motor exercise, but <italic>during</italic> the neurohabilitation training. This suggests that not “simple” excitability changes were involved here (when excitability is modified by TMS through the whole stimulated area independent on specific functional activity), but that additional mechanisms were involved that altered the brain’s response to the external manipulation. The authors propose that volitional modulation of brain activity with motor imagery improved susceptibility of inherent motor circuits to TMS pulses, perhaps due to voluntary depolarization of intracortical connections targeting pyramidal tract neurons and decrease of the motor cortical excitability as did motor imagery with haptic feedback alone.</p><p>Though this experiment involved only one healthy subject and one stroke patient, this finding nevertheless is novel because it may lead to new concepts of how brain stimulation may act: in order for plastic changes to emerge in a certain brain area, the central network, and external stimulation drive should be temporally and spatially related.</p><p>In line with the study from Gharabaghi et al. (<xref rid="B4" ref-type="bibr">2014</xref>) is the finding that the endogenous power of brain oscillations (changing with anesthesia stages) has a huge impact on the “aftereffects” of alternating current stimulation (ACS) (Sergeeva et al., <xref rid="B8" ref-type="bibr">2012</xref>). Moreover, Neuling et al. (<xref rid="B6" ref-type="bibr">2013</xref>) demonstrated that when the timing was just right, the phase alignment of intrinsic oscillators with the external stimulation lead to an increased amplitude of the response.</p><p>The importance of the actual brain state to determine behavioral and perceptual effects of TMS and TDS was already addressed by Silvanto et al. (<xref rid="B10" ref-type="bibr">2008</xref>). They showed that prior manipulation of neural activation enabled TMS to selectively target populations of neurons to increase functional resolution and achieve a selective excitation of task-related areas (Silvanto et al., <xref rid="B11" ref-type="bibr">2007</xref>).</p><p>Therefore, the BSDS as described by Gharabaghi et al. (<xref rid="B4" ref-type="bibr">2014</xref>) may permit to accurately stimulate the brain, thereby improving task performance as a function of altered excitability in the areas, which were functionally primed.</p><p>This consideration of state-dependent stimulation is novel in the context of brain current stimulation, though it follows the early theory by Sherrington (<xref rid="B9" ref-type="bibr">1965</xref>), and implies that the arousal of brain structures by natural tasks leads to a certain neural constellations of excitation and inhibition, which may serve as an immanent substrate for external stimuli. Since the processing of these stimuli is dependent not only on their physical properties but also on the intrinsic constitution of the stimulated system, we hypothesize that a pre-set task-primed system may show greater responsiveness in terms of better functional output to neuromodulation by brain stimulation.</p><p>The modulation of brain activity with non-invasive current stimulation has become tremendously popular. But the major concern is how to improve their precision and effectiveness. We therefore expect that future neuromodulation approaches use increasingly more fine-tuned BSDS similar to those we have witnessed most recently with optogenetic approaches (Zemelman et al., <xref rid="B12" ref-type="bibr">2002</xref>): just like the light can specifically activate cells that have optogenetic sensors, the current injection patterns could be controlled to just activate or inhibit particular (primed) neuronal populations. In this manner, functional priming of certain brain areas and even groups of neurons prior or during the current/magnetic stimulation would be a possible solution to better control efficacy and safety of non-invasive brain current stimulation.</p><p>Because the state of brain networks in patients is likely to be altered, as it was observed in our laboratory in patients with visual system damage (Bola et al., <xref rid="B2" ref-type="bibr">2014</xref>), the stimulation protocols known to exert certain effects in healthy subjects might not work in the same way in patients. Therefore, while it is desirable to discover general principles of priming-dependent stimulation effects in normal subjects, it might be difficult to define protocols optimal for all patients suffering from a certain condition. Rather, stimulation methods should be used in combination with neuroimaging (Fox et al., <xref rid="B3" ref-type="bibr">2012</xref>), e.g., EEG or fMRI, to probe the brain state. These efforts should ultimately lead to closed-loop devices adjusting stimulation parameters automatically based on patient’s brain activity patterns.</p><p>The BSDS approach provides a basis for a novel restoration strategy. Further exploration of the mechanisms underlying BSDS, i.e., Hebbian plasticity or homeostatic metaplasticity and gating (Ziemann and Siebner, <xref rid="B13" ref-type="bibr">2008</xref>) and how to prime different modes of stimulation in functional domains beyond the motor system will help to advance the field and help us pinpoint the most effective non-invasive brain stimulation protocols for neurorehabilitation and restoration.</p><sec id="S1"><title>Conflict of Interest Statement</title><p>The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.</p></sec>
The chromatin remodelling component SMARCB1/INI1 influences the metastatic behavior of colorectal cancer through a gene signature mapping to chromosome 22	<sec><title>Background</title><p>INI1 (Integrase interactor 1), also known as SMARCB1, is the most studied subunit of chromatin remodelling complexes. Its role in colorectal tumorigenesis is not known.</p></sec><sec><title>Methods</title><p>We examined SMARCB1/INI1 protein expression in 134 cases of colorectal cancer (CRC) and 60 matched normal mucosa by using tissue microarrays and western blot and categorized the results according to mismatch repair status (MMR), CpG island methylator phenotype, biomarkers of tumor differentiation CDX2, CK20, vimentin and p53. We validated results in two independent data sets and in cultured CRC cell lines.</p></sec><sec><title>Results</title><p>Herein, we show that negative SMARCB1/INI1 expression (11% of CRCs) associates with loss of CDX2, poor differentiation, liver metastasis and shorter patients’ survival regardless of the MMR status or tumor stage. Unexpectedly, even CRCs displaying diffuse nuclear INI1 staining (33%) show an adverse prognosis and vimentin over-expression, in comparison with the low expressing group (56%). The negative association of SMARCB1/INI1-lack of expression with a metastatic behavior is enhanced by the <italic>TP53</italic> status. By interrogating global gene expression from two independent cohorts of 226 and 146 patients, we confirm the prognostic results and identify a gene signature characterized by SMARCB1/INI1 deregulation. Notably, the top genes of the signature (<italic>BCR, COMT, MIF)</italic> map on the long arm of chromosome 22 and are closely associated with SMARCB1/INI1.</p></sec><sec><title>Conclusion</title><p>Our findings suggest that <italic>SMARCB1/INI1</italic>-dysregulation and genetic hot-spots on the long arm of chromosome 22 might play an important role in the CRC metastatic behavior and be clinically relevant as novel biomarkers.</p></sec>	<contrib contrib-type="author" corresp="yes" equal-contrib="yes" id="A1"><name><surname>Pancione</surname><given-names>Massimo</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>massimo.pancione@unisannio.it</email></contrib><contrib contrib-type="author" equal-contrib="yes" id="A2"><name><surname>Remo</surname><given-names>Andrea</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>remino76@yahoo.it</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Zanella</surname><given-names>Caterina</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>caterina.zanella@aulsslegnago.it</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Sabatino</surname><given-names>Lina</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>sabatino@unisannio.it</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Blasi</surname><given-names>Arturo Di</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>arturo.diblasi@tin.it</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Laudanna</surname><given-names>Carmelo</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>laudanna.carmelo@gmail.com</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Astati</surname><given-names>Laura</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>laura852005@libero.it</email></contrib><contrib contrib-type="author" id="A8"><name><surname>Rocco</surname><given-names>Michele</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>michelerocco@virgilio.it</email></contrib><contrib contrib-type="author" id="A9"><name><surname>Bifano</surname><given-names>Delfina</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>delfinabifano@virgilio.it</email></contrib><contrib contrib-type="author" id="A10"><name><surname>Piacentini</surname><given-names>Paolo</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>paolo.piacentini@aulsslegnago.it</email></contrib><contrib contrib-type="author" id="A11"><name><surname>Pavan</surname><given-names>Laura</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>laura.pavan09@gmail.com</email></contrib><contrib contrib-type="author" id="A12"><name><surname>Purgato</surname><given-names>Alberto</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>albertopurgato@gmail.com</email></contrib><contrib contrib-type="author" id="A13"><name><surname>Greco</surname><given-names>Filippo</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>filippo.greco@aulsslegnago.it</email></contrib><contrib contrib-type="author" id="A14"><name><surname>Talamini</surname><given-names>Alberto</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>alberto.talamini@aulsslegnago.it</email></contrib><contrib contrib-type="author" id="A15"><name><surname>Bonetti</surname><given-names>Andrea</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>andrea.bonetti@aulsslegnago.it</email></contrib><contrib contrib-type="author" id="A16"><name><surname>Ceccarelli</surname><given-names>Michele</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>m.ceccarelli@gmail.com</email></contrib><contrib contrib-type="author" id="A17"><name><surname>Vendraminelli</surname><given-names>Roberto</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>roberto.vendraminelli@aulsslegnago.it</email></contrib><contrib contrib-type="author" id="A18"><name><surname>Manfrin</surname><given-names>Erminia</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>erminia.manfrin@univr.it</email></contrib><contrib contrib-type="author" corresp="yes" id="A19"><name><surname>Colantuoni</surname><given-names>Vittorio</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>colantuoni@unisannio.it</email></contrib>	Journal of Translational Medicine	<sec><title>Background</title><p>The chromatin remodelling (CR) complexes dynamically regulate transcription by using the energy from ATP hydrolysis to reposition nucleosomes and modulate accessibility of specific genes to the transcriptional machinery [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>]. Recently, inactivating mutations in the CR complexes have been identified at high frequency in a variety of tumors, highlighting the widespread role of epigenome alterations in tumor suppression or oncogenic activation [<xref ref-type="bibr" rid="B1">1</xref>]. Integrase interactor 1 (INI1, also known as SMARCB1) is a core subunit of the SWI/SNF ATP-dependent CR complex encoded by the corresponding gene at chromosomal position 22q11.2 [<xref ref-type="bibr" rid="B3">3</xref>-<xref ref-type="bibr" rid="B5">5</xref>]. SMARCB1/INI1 is ubiquitously expressed in normal cells and can be readily identified by immunohistochemistry. <italic>SMARCB1/INI1</italic> germ-line mutations were first described in the malignant rhabdoid tumors (MRT) of infancy and atypical theratoid/rhabdoid tumors of the central nervous system and define a hereditary condition known as “Rhabdoid predisposition syndrome” [<xref ref-type="bibr" rid="B3">3</xref>-<xref ref-type="bibr" rid="B6">6</xref>]. Deletions at chromosome 22 or loss of <italic>SMARCB1/INI1</italic> expression have also been implicated in the pathogenesis of additional tumor types: renal medullary carcinomas, epithelioid sarcomas, myoepithelial carcinomas and extraskeletal myxoid chondrosarcomas [<xref ref-type="bibr" rid="B7">7</xref>]. Although SMARCB1/INI1 is the most extensively studied subunit of the SWI/SNF complex, very little is known about its role in the pathogenesis of colorectal cancer (CRC) [<xref ref-type="bibr" rid="B8">8</xref>]. Recently, we reported that <italic>SMARCB1/INI1</italic> inactivation or, alternatively, a genomic rearrangement at the chromosome region 22q12 are involved in Rhabdoid Colorectal Tumor (RCT), a rare and highly aggressive neoplasm of the gastrointestinal tract [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. <italic>SMARCB1/INI1</italic>-deficient mice develop rapidly aggressive undifferentiated sarcomas, implying a cancer-related function [<xref ref-type="bibr" rid="B11">11</xref>]. Notably, in the same mouse model, the conditional inactivation of <italic>TP53</italic> leads to a dramatic acceleration of tumor formation and a wider spectrum of cancers than those seen in <italic>TP53</italic> deficient mice alone [<xref ref-type="bibr" rid="B12">12</xref>]. These results suggest a cooperative effect of both genes to prevent oncogenic transformation and a dominant role of <italic>SMARCB1/INI1</italic> to hamper cancer aggressiveness. Despite the evidence in mouse models, the link between <italic>SMARCB1/INI1</italic> alterations and the molecular changes underlying CRC progression remains still poorly understood. In order to shed light on the biological role of <italic>SMARCB1/INI1</italic>, in this study we investigated its expression profile and evaluated the relationship between molecular alterations and clinico-histological markers of dedifferentiated and aggressive colorectal carcinomas. We hypothesize that its assessment might be clinically relevant to predict CRC prognosis.</p></sec><sec sec-type="materials\|methods"><title>Materials and methods</title><sec><title>Tumor samples and TMA construction</title><p>Colorectal cancer specimens and matched normal mucosa were collected at two institutions, Fatebenefratelli Hospital, Benevento, and Legnago Hospital, Verona, Italy. This study was carried out according to the principles of the Declaration of Helsinki with appropriate patient’s informed consent and approved by the Institutional Review Board of both hospitals. Altogether, a total of 134 patients, 85 men and 49 women with mean age of 70.5 ± 11.8 were analyzed. The tumors were classified and graded according to the criteria of the TNM and tumor stages I-IV classification systems, (Table <xref ref-type="table" rid="T1">1</xref>). None of the patients had a familial history of intestinal dysfunction or CRC, had received chemotherapy or radiation prior to resection nor had taken non-steroidal anti-inflammatory drugs on a regular basis. For each patient, the date of colon cancer diagnosis, date of last follow-up, and vital status at last follow-up (<italic>i.e.,</italic> living or deceased) were recorded. TMAs were constructed from archival tissue blocks of normal and colorectal cancer using a Beecher tissue microarray instrument (Beecher Instruments, Hacken-sack, NJ, USA). Tissue cylinders, with a diameter of 0.6 mm, were punched from paraffin blocks in demarcated areas on parallel haematoxylin&eosin-stained sections. Three separate cores were sampled from each block deposited into a recipient master paraffin block. Each core was placed 1 mm apart on the x-axis and 1.5 mm apart on the y-axis of the master block. In total, 12 microarrays paraffin block were prepared, 4 μ thick sections were cut from each TMA block and stained with haematoxylin&eosin. Microarray sections were then reviewed to ensure that the sections from each case were morphologically similar to those of the corresponding whole tissue section and represented cancerous or normal epithelial cells. Further 4 μ thick sections were then cut from each of the master blocks for immunohistochemical (IHC) analyses, the cores containing too little tumor sample were not included in the study. Due to technical problems and/or tissue exhaustion, the number of lesions that were available for evaluation by immunohistochemistry included 134 carcinomas (Table <xref ref-type="table" rid="T1">1</xref>).</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p><bold>Correlation between</bold> SMARCB1/INI1 <bold>expression pattern and patients’ clinico-pathological parameters</bold></p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th rowspan="2" align="left" valign="top"><bold>Parameters</bold></th><th rowspan="2" align="left"> </th><th rowspan="2" align="center" valign="top"><bold>n</bold></th><th colspan="3" align="center" valign="bottom"><bold>INI1</bold><hr/></th><th rowspan="2" align="center" valign="top"><bold> <italic>P </italic> </bold><bold>value</bold></th></tr><tr><th align="center"><bold>Neg (%)</bold></th><th align="center"><bold>Low (%)</bold></th><th align="center"><bold>High (%)</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom"><bold>Age</bold><hr/></td><td align="left" valign="bottom">≤60<hr/></td><td align="center" valign="bottom">22<hr/></td><td align="center" valign="bottom">1 (4.5)<hr/></td><td align="center" valign="bottom">12 (54.5)<hr/></td><td align="center" valign="bottom">9 (41)<hr/></td><td align="center" valign="bottom">0.456<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">>60<hr/></td><td align="center" valign="bottom">112<hr/></td><td align="center" valign="bottom">14 (12.5)<hr/></td><td align="center" valign="bottom">63 (56.2)<hr/></td><td align="center" valign="bottom">35 (31.3)<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"><bold>Sex</bold><hr/></td><td align="left" valign="bottom">F<hr/></td><td align="center" valign="bottom">49<hr/></td><td align="center" valign="bottom">9 (18.4)<hr/></td><td align="center" valign="bottom">21 (42.6)<hr/></td><td align="center" valign="bottom">19 (39)<hr/></td><td align="center" valign="bottom">0.056<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">M<hr/></td><td align="center" valign="bottom">85<hr/></td><td align="center" valign="bottom">6 (7.1)<hr/></td><td align="center" valign="bottom">54 (63.5)<hr/></td><td align="center" valign="bottom">25 (29.4)<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"><bold>Location</bold><hr/></td><td align="left" valign="bottom">Proximal<hr/></td><td align="center" valign="bottom">51<hr/></td><td align="center" valign="bottom">5 (9.8)<hr/></td><td align="center" valign="bottom">34 (66.6)<hr/></td><td align="center" valign="bottom">12 (23.6)<hr/></td><td align="center" valign="bottom">0.136<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Distal<hr/></td><td align="center" valign="bottom">83<hr/></td><td align="center" valign="bottom">10 (12)<hr/></td><td align="center" valign="bottom">41 (49.4)<hr/></td><td align="center" valign="bottom">32 (38.6)<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"><bold>Histology</bold><hr/></td><td align="left" valign="bottom">ADC<hr/></td><td align="center" valign="bottom">108<hr/></td><td align="center" valign="bottom">10 (9.2)<hr/></td><td align="center" valign="bottom">62 (57.4)<hr/></td><td align="center" valign="bottom">36 (33.4)<hr/></td><td align="center" valign="bottom">0.667<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">A-Muc<hr/></td><td align="center" valign="bottom">17<hr/></td><td align="center" valign="bottom">3 (17.6)<hr/></td><td align="center" valign="bottom">9 (52.9)<hr/></td><td align="center" valign="bottom">5 (29.5)<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Other<hr/></td><td align="center" valign="bottom">9<hr/></td><td align="center" valign="bottom">2 (22.2)<hr/></td><td align="center" valign="bottom">4 (44.4)<hr/></td><td align="center" valign="bottom">3 (33.3)<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"><bold>Grade</bold><hr/></td><td align="left" valign="bottom">Well/mod<hr/></td><td align="center" valign="bottom">109<hr/></td><td align="center" valign="bottom">7 (6.4)<hr/></td><td align="center" valign="bottom">67 (61.5)<hr/></td><td align="center" valign="bottom">35 (32.1)<hr/></td><td align="center" valign="bottom">0.003<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Poor<hr/></td><td align="center" valign="bottom">25<hr/></td><td align="center" valign="bottom">8 (32)<hr/></td><td align="center" valign="bottom">8 (32)<hr/></td><td align="center" valign="bottom">9 (36)<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"><bold>N stage</bold><hr/></td><td align="left" valign="bottom">N0<hr/></td><td align="center" valign="bottom">88<hr/></td><td align="center" valign="bottom">7 (7.9)<hr/></td><td align="center" valign="bottom">51 (62.5)<hr/></td><td align="center" valign="bottom">30 (29.6)<hr/></td><td align="center" valign="bottom">0.387<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">N1<hr/></td><td align="center" valign="bottom">24<hr/></td><td align="center" valign="bottom">3 (12.5)<hr/></td><td align="center" valign="bottom">14 (58.3)<hr/></td><td align="center" valign="bottom">7 (29.2)<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">N2<hr/></td><td align="center" valign="bottom">22<hr/></td><td align="center" valign="bottom">5 (22.7)<hr/></td><td align="center" valign="bottom">10 (45.4)<hr/></td><td align="center" valign="bottom">7 (31.9)<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"><bold>LiverMet</bold><hr/></td><td align="left" valign="bottom">Negative<hr/></td><td align="center" valign="bottom">93<hr/></td><td align="center" valign="bottom">4 (4.3)<hr/></td><td align="center" valign="bottom">58 (62.4)<hr/></td><td align="center" valign="bottom">31 (33.3)<hr/></td><td align="center" valign="bottom">0.001<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Positive<hr/></td><td align="center" valign="bottom">41<hr/></td><td align="center" valign="bottom">11 (26.8)<hr/></td><td align="center" valign="bottom">17 (41.4)<hr/></td><td align="center" valign="bottom">13 (31.8)<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"><bold>Stage</bold><hr/></td><td align="left" valign="bottom">I<hr/></td><td align="center" valign="bottom">12<hr/></td><td align="center" valign="bottom">0<hr/></td><td align="center" valign="bottom">5 (41.6)<hr/></td><td align="center" valign="bottom">7 (58.4)<hr/></td><td align="center" valign="bottom">0.002<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">II<hr/></td><td align="center" valign="bottom">63<hr/></td><td align="center" valign="bottom">3 (4.8)<hr/></td><td align="center" valign="bottom">42 (66.6)<hr/></td><td align="center" valign="bottom">18 (28.6)<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">III<hr/></td><td align="center" valign="bottom">20<hr/></td><td align="center" valign="bottom">1 (5)<hr/></td><td align="center" valign="bottom">12 (60)<hr/></td><td align="center" valign="bottom">7 (35)<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">IV<hr/></td><td align="center" valign="bottom">39<hr/></td><td align="center" valign="bottom">11 (28.2)<hr/></td><td align="center" valign="bottom">16 (41)<hr/></td><td align="center" valign="bottom">12 (30.8)<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left"><bold>Total</bold></td><td align="left"> </td><td align="center">134</td><td align="center">15 (11)</td><td align="center">75 (56)</td><td align="center">44 (33)</td><td align="center"> </td></tr></tbody></table><table-wrap-foot><p>Tumor classification was based on the TNM (Tumor-Node-Metastasis) system, according to the criteria of the International Union Against Cancer (UICC).</p><p><italic>Abbreviations:</italic><italic>Proximal</italic> caecum, ascending and transverse colon, <italic>Distal</italic> descending and sigmoid colon, rectum, <italic>Adc</italic> adenocarcinoma, <italic>AD-Muc</italic> adenocarcinoma with a mucinous component below 50%, <italic>Other</italic> squamous or rhabdoid, <italic>Well/mod</italic> well and moderately differentiated, <italic>Poor</italic> poorly differentiated adenocarcinoma, <italic>Liver Met</italic> Liver metastasis.Chi-square statistic significant at the 0.01 level.</p></table-wrap-foot></table-wrap></sec><sec><title>Immunohistochemistry</title><p>The TMAs were serially sectioned at 4 μ, dewaxed in xylene and rehydrated through graded alcohol to water. Slides were subjected to microwave antigen retrieval in 10 mM Citrate buffer (pH of 6.0) before incubation with the primary antibodies. The following antibodies, at 1:100 dilution, were employed: SMARCB1/INI1 clone 25/BAF47; (DAKO Cytomation, Glostrup, Denmark). CK20 clone Ks 20.8; vimentin clone VIM 3B4; p53 clone Bp53-11; (Novocastra Laboratories, Newcastle, UK); CDX2 clone EPR2764Y (Ventana Medical Systems, Tucson, AZ, USA). Automated immunohistochemistry system (Ventana Medical Systems, Tucson, AZ, USA) was employed to detect immunostaining as previously reported [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B13">13</xref>]. Finally, the sections were counterstained with hematoxylin, dehydrated, and cover-slipped. In each run, primary antibodies were omitted in negative controls.</p></sec><sec><title>Evaluation of immunohistochemistry</title><p>All immunohistochemical results were interpreted by 2 independent observers (A. R, and M. P.) blinded to clinical data and laboratory results. For SMARCB1/INI1, p53 and vimentin the immunostaining was recorded regardless of intensity, according to the proportion of positive neoplastic cells. According to the number of positive tumor cells, we stratified the carcinomas into three groups: 1) “Low expression”, in which the positivity was observed in a limited number of tumor cells, scattered in a background of either negative or weakly positive tumor cells; this subgroup was also defined as Partly positive; 2) “High expression” or strongly diffuse expression, corresponding to an homogeneous staining in virtually all tumor cells 3) “Negative expression” when less than 5% of tumor cells were positive. Positivity in normal colonic mucosa, inflammatory and stromal cells adjacent to neoplastic cells served as positive internal controls. For CK20 and CDX2, the staining in less than 5% of tumor cells was scored as negative. For each marker, normal colonic mucosal tissue was used as positive control.</p></sec><sec><title>Mismatch repair, MSI and CIMP analysis</title><p>To evaluate mismatch repair, the following antibodies at a dilution 1:100, were used: anti–MLH-1 clone (M1); anti-MSH-2 clone (G219-1129); anti-MSH6 clone 44; anti-PMS2 clone EPR394; (Ventana Medical Systems, Tucson, AZ, USA). The tumors were defined as mismatch repair-deficient when they showed an absence of nuclear staining in at least one of following marker: MLH1 or MSH2 or MSH6 or PMS2. Inflammatory and stromal cells adjacent to neoplastic cells served as positive internal controls. Microsatellite instability (MSI) assessment in both mismatch repair-deficient or proficient cases was performed comparing tumor DNA and matched normal mucosa through a panel of highly-specific five mononucleotide repeats, as described [<xref ref-type="bibr" rid="B14">14</xref>]. An agreement of the 95% between MSI and MMR status was obtained, supporting the use of MMR profile for subsequent analyses. Genomic DNA isolation and sodium bisulphite modification were carried out as reported. The converted DNA was subjected to quantitative methylation specific PCR as reported [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B13">13</xref>]. The following genes (<italic>RUNX3, IGF2, SOCS1, NEUROG1, CDKN2A</italic> (p16) and h<italic>MLH1</italic>) with methylation levels greater than 15% were considered positive. Tumors with at least three methylated loci were classified as CpG island methylator phenotype (CIMP)-positive and the remaining cases as CIMP-negative [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B13">13</xref>]. The primers for promoter methylation analysis have already been reported.</p></sec><sec><title>Cell culture, migration, western blot and qRT-PCR analysis</title><p>Human CRC derived cell lines DLD1, HCT116, LoVo, RKO and SW480 were purchased from ATCC and cultured as recommended. Cell migration was evaluated by the wound-healing as previously described [<xref ref-type="bibr" rid="B10">10</xref>] and ref. therein. Western blot analysis and qRT-PCR were performed as already reported [<xref ref-type="bibr" rid="B10">10</xref>] and ref. therein. Expression levels were normalized to β-actin or to GAPDH mRNA, respectively. A detailed description of the primer sets will be provided upon request.</p></sec><sec><title>Independent CRC data sets and statistical analysis</title><p>The following independent, publically available CRC datasets, deposited in the Gene Expression Omnibus (GEO) GSE17536, GSE17537 and GSE41258 series (<ext-link ext-link-type="uri" xlink:href="http://www.ncbi.nlm.nih.gov/geo">http://www.ncbi.nlm.nih.gov/geo</ext-link>) were analyzed to validate <italic>SMARCB1/INI1</italic> expression and its prognostic significance [<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B16">16</xref>]<italic>.</italic> The GSE17536 and GSE17537 pooled series (cohort I) consists of 226 patients; while the GSE41258 series (cohort II) consists of 146 patients [<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B16">16</xref>]. Disease-specific survival was considered as a prognostic variable, whereas, the data on <italic>TP53</italic> mutations status were available only for cohort II. A fold-change of at least 1.5 (<italic>p</italic> value <0.05) was used to identify up- and down-regulated genes, respectively. Volcano plot analysis was employed to visualize differential expression. In order to find differentially expressed genes (DEGs) co-regulated with <italic>SMARCB1/INI1,</italic> a heat map with hierarchical clustering analysis was performed. The DEGs were separated in two clusters using a random-variance t test. Subsequently, they were selected for Gene Ontology (GO) terms and pathway analysis. Ingenuity Pathways Analysis (IPA; Ingenuity Systems, <ext-link ext-link-type="uri" xlink:href="http://www.ingenuity.com">http://www.ingenuity.com</ext-link>) was used for gene set enrichment analysis and gene network analysis. Statistical analyses were performed using GeneSpring R/bioconductor v.12.5. Data are reported as median or mean and standard deviation (SD), and the mean values compared using the Student’s t test, as indicated. The χ<sup>2</sup> or Spearman tests were employed to assess the association of markers and clinico-pathological parameters. Univariate analyses were performed by using Kaplan-Meier estimates and log-rank tests, with raw score data obtained for each individual biomarker. A Cox regression model stepwise selection procedure for was used to identify those markers that independently predict disease outcome whereby hazard ratios (HR), 95% confidence interval (95% CI) and significance levels were estimated. Statistical analyses were carried out with the SPSS (version 15.0) for Windows (SPSS Inc., Chicago, Ill., USA). Results were considered statistically significant when a <italic>p ≤</italic> 0.05 was obtained.</p></sec></sec><sec sec-type="results"><title>Results</title><sec><title>SMARCB1/INI1 expression profile in colorectal cancer and matched normal mucosa</title><p>In the normal mucosa, SMARCB1/INI1 nuclear positivity was evenly distributed between proliferative and differentiated colonic cells (Figure <xref ref-type="fig" rid="F1">1</xref>A). In few cases (5/60, 8%), we observed a stronger positivity in the proliferative compartment of the crypts. To identify cancer-specific alterations, we first investigated the differences in SMARCB1/INI1 expression in CRCs and paired normal mucosa (Figure <xref ref-type="fig" rid="F1">1</xref>A). CRC samples exhibited a higher percentage of SMARCB1/INI1-positive cells than matched normal colonic mucosa (Figure <xref ref-type="fig" rid="F1">1</xref>B). We further analyzed CRC samples and classified SMARCB1/INI1 expression pattern into three groups, according to the proportion and distribution of positive neoplastic cells. By applying this criterion, we detected a moderate expression in 56% (75/134) of tumors, classified as Low or Partly positive; 33% (44/134) had a diffuse and stronger positivity and classified as High, while 11% (15/134) did not show any significant SMARCB1/INI1 immunoreactivity, classified as Negative (Figure <xref ref-type="fig" rid="F1">1</xref>C). The expression profile was validated on twenty representative cases (5 negative and 15 positive), by evaluating the consensus between each core of the TMAs and the corresponding whole tissue section. We found no discordance, supporting the value of the TMA method to screen SMARCB1/INI1 expression in our CRC dataset. To further corroborate the IHC expression profile and have more quantitative data, twenty selected frozen CRC specimens and matched normal mucosas from the same cohort of patients were analyzed by western blot. The bands were quantitated by densitometry after normalization to β-actin for protein loading. SMARCB1/INI1 showed variable expression levels in CRC specimens as compared to case-matched normal tissue. Five tumors, defined SMARCB1/INI1-negative, showed lack of SMARCB1/INI1 protein as compared to normal mucosa, confirming the IHC results (Figure <xref ref-type="fig" rid="F1">1</xref>D). In contrast, five tumors were SMARCB1/INI1-positive as the expression was significantly higher than controls. The remaining ten cases showed no significant changes versus the normal mucosa. Although the data referred only to twenty cases, they confirmed the specificity of the results and reinforced the differences between normal and tumor samples detected by IHC on TMAs (Figure <xref ref-type="fig" rid="F1">1</xref>D).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>SMARCB1/INI1 expression analysis in Tissue Microarray of CRC and matched normal mucosa. (A)</bold> Examples of TMA cores representative of the normal mucosa and CRC specimens stained with SMARCB1/INI1; the immunostaining pattern allows to divide tumors into three categories, Neg, Low, High<bold>. (B)</bold> The box-plot shows the SMARCB1/INI1score (number of positive cells in 10-high power fields) in paired normal mucosa and tumor samples (60 cases). <bold>(C)</bold> SMARCB1/INI1 expression pattern expressed as percentage of cases for each category in all 134 CRCs. <bold>(D)</bold> Four representative frozen CRC specimens (T) and matched normal mucosa (N) from the same cohort of patients are analyzed by immunoblot. The β-actin is used as loading control to normalize SMARCB1/INI1 band intensities. The histogram reports quantitative expression levels of SMARCB1/INI1 after normalization. The box-plot shows that SMARCB1/INI1 expression detected by western blot in normal mucosa is lower than tumor samples (20 cases). The <italic>p</italic> value is reported in each graph.</p></caption><graphic xlink:href="1479-5876-11-297-1"/></fig></sec><sec><title>SMARCB1/INI1 expression profiles correlate with poorly differentiated tumors and liver metastasis</title><p>We then associated the SMARCB1/INI1 expression patterns with patients’ clinico-pathological features, immunohistochemical and molecular markers of tumor differentiation. SMARCB1/INI1-negative immunostaining showed a significant relation with poor differentiation, presence of liver metastasis and advanced tumor stage IV (Table <xref ref-type="table" rid="T1">1</xref>). No statistically significant difference was found taking into account other clinico-pathological features such as: age, gender, histology, tumor location and lymph node metastasis (Table <xref ref-type="table" rid="T1">1</xref>). Next, we examined whether its expression could correlate with multiple biomarkers such as: CDX2, CK20, vimentin, p53, CIMP and MMR status (Figure <xref ref-type="fig" rid="F2">2</xref>A, B and C and data not shown). SMARCB1/INI1-negative tumors showed lower CDX2 positivity than any other group (<italic>p =</italic> 0.049). The same group exhibited a diffuse pattern of vimentin overexpression. Unexpectedly, also SMARCB1/INI1-high tumors were markedly vimentin positive (Figure <xref ref-type="fig" rid="F2">2</xref>B). We did not detect any significant correlation with either p53 expression or CIMP-positive tumors (Figure <xref ref-type="fig" rid="F2">2</xref>C).</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Correlation between SMARCB1/INI1 expression and different molecular markers. (A)</bold> CDX2, vimentin, p53 and SMARCB1/INI1 immunostaining profile in a poorly differentiated tumor <bold>(a-d). (B)</bold> The SMARCB1/INI1 expression profile is divided into three categories Neg, Low and High and correlated with CDX2 and vimentin expression pattern in tumor cells. <bold>(C)</bold> The same analysis takes into account p53 expression and the CpG island methylator phenotype (CIMP) status. Tumors with at least three methylated markers (<italic>RUNX3, IGF2, SOCS1, NEUROG1, CDKN2A</italic> and h<italic>MLH1</italic>) were classified as CIMP-positive, the remaining as CIMP-negative. The <italic>p</italic> value is reported in each graph.</p></caption><graphic xlink:href="1479-5876-11-297-2"/></fig><p>We sought to investigate the relationship of each of the markers analyzed with the MMR status, by dividing the CRCs in two groups according to the proficient or deficient condition. 23% (23/134) of the cases were MMR deficient (MMR negative); as expected, they occurred more frequently at the right colon and were poorly differentiated tumors. Consistent with previous studies, this subgroup exhibited lack of CDX2, CK20 and p53 expression (Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Table S1) and, interestingly, higher vimentin positivity than the MMR proficient CRCs that showed instead low vimentin levels (94 vs 6%). Finally, we detected no correlation between the different SMARCB1/INI1 expression profiles and the MMR or MSI status (Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Table S1, data not shown). These results indicate that loss of SMARCB1/INI1 expression is associated with poorly differentiated tumors and presence of liver metastasis. Furthermore, even a significant proportion of CRCs with high SMARCB1/INI1 expression exhibit a marked vimentin positivity.</p></sec><sec><title>Altered SMARCB1/INI1 expression correlates with patients’ prognosis in our CRC dataset</title><p>In our cohort, cancer related death occurred in 35.8% of the cases (48/134 patients). We stratified patients’ overall survival (OS) into three categories according to the SMARCB1/INI1 expression patterns. Low SMARCB1/INI1-expressing tumors had the best prognosis as compared to those with High or Negative expression (Figure <xref ref-type="fig" rid="F3">3</xref>A). The SMARCB1/INI1-negative subgroup preserved the worst impact on patients’ survival also in tumor stages I-II or when adjusted for all tumor stages in a multivariate analysis (Figure <xref ref-type="fig" rid="F3">3</xref>B and data not shown). To investigate whether the prognostic impact of SMARCB1/INI1 was dependent upon the MMR, we stratified the tumors according to the MMR deficient or proficient status. Two groups of patients, the SMARCB1/INI1-negative or -high expressing ones, were associated with a shorter survival time than the low expressing ones in both MMR-proficient and deficient CRCs (Figure <xref ref-type="fig" rid="F3">3</xref>C). A multivariate model showed that SMARCB1/INI1 expression preserves a prognostic significance when adjusted for MMR status (data not shown). Since SMARCB1/INI1 and p53 co-inactivation can accelerate the rate of tumorigenesis, we explored the effects of such a combination on patients’ outcome. To this end, we classified tumors into 4 groups according to positive or negative expression (Figure <xref ref-type="fig" rid="F4">4</xref>A). Interestingly, the SMARCB1/INI1<sup>-</sup>/p53<sup>+</sup> group (8% of cases, 11/134 patients) showed a very short OS in all tumor stages I-IV or stages I-II alone, when compared to any other SMARCB1/INI1/p53 combination (Figure <xref ref-type="fig" rid="F4">4</xref>B,C). In agreement with these data, almost the entire SMARCB1/INI<sup>-</sup>/p53<sup>+</sup> subgroup (90%) developed liver metastases with respect to any other group (Figure <xref ref-type="fig" rid="F4">4</xref>D). To further support our findings, we interrogated a CRC independent dataset, validation cohort II, from which the transcriptional profile of <italic>SMARCB1/INI1</italic>, <italic>TP53</italic> mutation status and clinical follow-up are publicly available [<xref ref-type="bibr" rid="B16">16</xref>]. We confirmed that the <italic>SMARCB1/INI1</italic> down-regulation, combined with <italic>TP53</italic> mutations, correlated with a poorer patients’ prognosis than any other group (Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Figure S1A). Altogether, these results indicate that SMARCB1/INI1-negative or -high expression is associated with an adverse CRC prognosis regardless of the MMR status and is influenced at least in part by the <italic>TP53</italic> status.</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>SMARCB1/INI1 expression in CRC and its impact on patients’ survival. (A)</bold> Overall survival (OS) referred to all tumor stages (I-IV) is estimated using the Kaplan–Meier method and stratifying the patients according to three categories of SMARCB1/INI1 expression (Neg, Low, High). <bold>(B)</bold> The same Kaplan-Meier survival analysis is carried out taking into account only tumor stages I-II. <bold>(C)</bold> Mean survival time referred to the three categories of SMARCB1/INI1 expression and stratified according to the mismatch repair (MMR) status. MMR (+) and MMR (−) indicate MMR proficient and deficient tumors, respectively. <italic>p <0.01.</italic> The <italic>p</italic> value is reported in each graph.</p></caption><graphic xlink:href="1479-5876-11-297-3"/></fig><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>SMARCB1/INI1 and p53 expression impacts aggressive features of tumors. (A)</bold> Frequency of the four CRC categories according to the various combinations of p53 and SMARCB1/INI1 expression. <bold>(B)</bold> OS analysis for each category referred to all tumor stages (I-IV) <bold>(C)</bold> The same analysis is carried out taking into account only tumor stages I-II. <bold>(D)</bold> Frequency of liver metastasis according to the various combinations of p53/ SMARCB1/INI1 expression. M0 and M1 indicate absence or presence of liver metastases, respectively. The <italic>p</italic> value is reported in each graph.</p></caption><graphic xlink:href="1479-5876-11-297-4"/></fig></sec><sec><title><italic>SMARCB1/INI1</italic> expression is validated in two independent cohorts of patients and reveals a gene signature mapping to chromosome 22</title><p>We further validated the <italic>SMARCB1/INI1</italic> expression profiles and its association with patients’ outcome by interrogating two CRC independent datasets, validation cohorts I and II, respectively [<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B16">16</xref>]. We computed the differences in gene expression by applying a fold-change of at least 1.5, and divided the microarray data into three quartiles (Figure <xref ref-type="fig" rid="F5">5</xref>A). In validation cohort I, 42 out of 226 patients (19%) were included in the 1st quartile and classified as <italic>SMARCB1/INI1</italic>-Negative; 132 (59%), in the 2nd and 3rd quartiles and classified as Low; 50 (22%) in the 4th quartile and classified as High (Figure <xref ref-type="fig" rid="F5">5</xref>B). Notably, these expression profiles were comparable with those observed in our CRC cohort, suggesting that <italic>SMARCB1/INI1</italic> expression at mRNA and protein level is stably maintained also in an independent dataset.</p><fig id="F5" position="float"><label>Figure 5</label><caption><p><bold>SMARCB1/INI1 expression profile is validated in an independent CRC microarray dataset, cohort I. (A)</bold> Volcano plot shows the graphical breakdown of the statistical analysis of SMARCB1/INI1 microarray data; The (x-axis) is the base 2 logarithm of the fold change, and the (Y-axis) is the negative base 2 logarithm of the q value (or adjusted <italic>p</italic>-value). Thresholds for both the statistical (q < 0.05) and the biological significance are highlighted and assembled in the top left and top right corner of the graph. The fold-changes with significant <italic>p</italic> values corresponding to SMARCB1/INI1-Neg and SMARCB1/INI1-High tumors (on the vertical axis) show that a 1.5-fold up- or down-regulation in gene expression is equivalent to log-ratios of +0.5 and −0.5; <bold>(B)</bold> Frequency of the identified subgroups displaying differential <italic>SMARCB1/INI1</italic> transcription in the validation series, cohort I, expressed in percentage; <bold>(C)</bold> Kaplan-Meier survival analysis is carried out taking into account each category; <bold>(D)</bold> Heat-map of differentially expressed genes. A hierarchical clustering method was used to construct the gene tree as described in Materials and Methods. The lists of differentially expressed genes with a t-test <italic>p-</italic>value <italic><0.05</italic> including multiple testing corrections were generated (for details, see also Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Table S1). Data are shown in a matrix format: each row represents a single gene and each column represents a group. Red indicates overexpressed genes (expression levels over the median) and green indicates underexpressed genes (expression levels below the median; see legend). The pattern and length of the branches in the dendrograms reflect the relatedness of the samples or the genes. The <italic>p</italic> value is reported in each graph.</p></caption><graphic xlink:href="1479-5876-11-297-5"/></fig><p>We next examined the association of <italic>SMARCB1/INI1</italic> expression with disease specific survival for the 226 patients of validation cohort I, whose follow-up data were available. Disease specific survival referred to the three categories showed that <italic>SMARCB1/INI1-</italic>Negative- or -High patients had a shorter survival time than Low expressing ones (Figure <xref ref-type="fig" rid="F5">5</xref>C). Remarkably, similar results were obtained taking into account cohort II, an independent series of 146 patients (Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Figure S1B). On the basis of these observations, we focused on two main groups, <italic>SMARCB1/INI1</italic>-Negative (down-regulated) and <italic>SMARCB1/INI1-</italic>High (up-regulated) tumors that significantly correlated with patients’ survival. The separation in two clusters was further confirmed by generating a two-dimensional hierarchical clustering heatmap. By this approach, we identified a robust set of genes, about 50, which significantly discriminated between <italic>SMARCB1/INI1</italic>-up and -down regulated tumors (Figure <xref ref-type="fig" rid="F5">5</xref>D and Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Table S2). Overall, the differentially expressed genes were significantly enriched in GO biological processes including: gastrointestinal cancer, cell cycle control, chromosomal replication and epithelial cell differentiation (Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Figure S2A, B).</p><p>Most notably, a cluster of loci, which represents the top differentially expressed genes (<italic>BCR, COMT, MCM5</italic> and <italic>MIF</italic>) mapped to the same long (q) arm of chromosome 22 where <italic>SMARCB1/INI1</italic> resides (Figure <xref ref-type="fig" rid="F6">6</xref>A,B). In particular, two of the most coregulated genes (<italic>BCR</italic> and <italic>MIF</italic>) were located on the same cytogenetic band 22q11.23, about 60 kb apart from <italic>SMARCB1/INI1</italic> (Figure <xref ref-type="fig" rid="F6">6</xref>A,B). The results obtained from cohort I were confirmed by interrogating cohort II. Also in this case, the top differentially expressed genes were localized close to <italic>SMARCB1/INI1,</italic> expanding the list of deregulated genes that are mapped to chromosome 22 (Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Figure S1C).</p><fig id="F6" position="float"><label>Figure 6</label><caption><p><bold>The top ranked genes, identified in Cohort I, map to the long arm of chromosome 22 and are closely linked to</bold><bold><italic>SMARCB1/INI1.</italic></bold><bold>(A)</bold> Top genes and biological functions that varied most in terms of differential expression across tumor samples displaying SMARCB1/INI1 down (Neg) or up-regulation (High). <bold>(B)</bold> The top deregulated genes (<italic>BCR, MIF, COMT</italic> and <italic>MCM5</italic>) with the highest statistically significant <italic>p</italic> values are located on the long (q) arm of chromosome 22 close to <italic>SMARCB1/INI1</italic>.</p></caption><graphic xlink:href="1479-5876-11-297-6"/></fig><p>Finally, to verify whether the changes in <italic>SMARCB1/INI1</italic> expression were associated with variations of the selected genes, we investigated a panel of four representative human CRC cell lines (Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Figure S3A-D). Interestingly, in poorly differentiated and more invasive RKO and DLD1 cell lines, we did confirm that the top deregulated genes were significantly associated with molecular features of enhanced vimentin and reduced CDX2 expression (Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Figure S3C-D). The association between co-regulation and gene co-localization was confirmed by performing interphase FISH at 22q12 locus in a subgroup of CRC specimens (data not shown).</p></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>The chromatin remodelling complexes mobilize nucleosomes to expose DNA to the transcriptional machinery. Alterations of these complexes are emerging as a critical step in carcinogenesis; in fact, high-frequency mutations in SWI/SNF members have been found in a variety of cancers by whole genome sequencing [<xref ref-type="bibr" rid="B2">2</xref>]. SMARCB1/INI1 is a core subunit of the SWI/SNF complex and a recognized hallmark for the diagnosis of MRT and other mesenchymal cancers [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B7">7</xref>]. Negative SMARCB1/INI1 expression is quite rare in epithelial tumors and none of the studies published so far has addressed its role in colorectal cancer [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B17">17</xref>]. Only few SWI/SNF components (<italic>BRM, BRG</italic> and <italic>ARID1A</italic>) have been reported mutated or deregulated in colon cancer; limited functional insights into the mechanisms of oncogenesis promoted by chromatin remodelling complexes are available so far. Even more, the prognostic significance of a large number of SWI/SNF subunits remains unknown [<xref ref-type="bibr" rid="B17">17</xref>-<xref ref-type="bibr" rid="B21">21</xref>]. Recently, we found that SMARCB1/INI1 expression was either negative or high in rhabdoid colorectal tumors and in a small group of sporadic CRCs [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B10">10</xref>].</p><p>In the present study, we investigated the SMARCB1/INI1 expression profiles in a larger CRC series and found that the majority (89%) express SMARCB1/INI1 with two distinct patterns of nuclear positivity, low (56%) and high (33%), respectively. The SMARCB1/INI1 nuclear positivity observed in the low expressing group resembled that detected in 60 normal colon tissues. A small group that accounts for 11% of our CRC series displayed a negative SMARCB1/INI1 immunostaining. Notably, negative SMARCB1/INI1 expression was related to poorly differentiated tumors and high frequency of liver metastases disclosing an association between its altered expression and the CRC subgroups more prone to metastatic spreading. SMARCB1/INI1 negative tumors frequently showed loss of CDX2 and high expression of vimentin, two key markers involved in colonic differentiation and mesenchymal phenotype, respectively. Unexpectedly, enhanced vimentin positivity was also found in the group displaying diffuse SMARCB1/INI1 expression.</p><p><italic>SMARCB1/INI1</italic> loss-of-function mutations or haploinsufficiency are recurrent in a variety of tumors, especially with rhabdoid features [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B17">17</xref>]. The molecular mechanisms underlying SMARCB1/INI1 protein inactivation in CRC were not explored in the present study; however, in agreement with recent data, we ruled out hypermethylation of the <italic>SMARCB1/INI1</italic> promoter region in our CRC cohort (our unpublished data) [<xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B22">22</xref>]. A recent comprehensive genome-wide analysis on 276 CRCs has identified <italic>SMARCB1/INI1</italic> mutations in less than 1% of cases [<xref ref-type="bibr" rid="B21">21</xref>]. These results suggest that epigenetic events might be responsible for <italic>SMARCB1/INI1</italic> inactivation because mutations alone do not fully explain the frequent variations in expression detected in CRCs. Further investigations are needed to answer this question.</p><p>The morphological revision of the slides from the 15 SMARCB1/INI1-negative tumors (7%, 1/15) revealed that only one showed a composite rhabdoid histology. The patient had a very short survival time (1 month), confirming the aggressive nature of this subgroup [<xref ref-type="bibr" rid="B8">8</xref>-<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B13">13</xref>]. Unlike others RCTs, we found a <italic>KRAS</italic> mutation, no <italic>BRAF</italic> mutations nor microsatellite instability. These findings reinforce our previous data, implying that <italic>SMARCB1/INI1</italic> plays a crucial role in later stages of colon carcinogenesis [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B10">10</xref>].</p><p>The most striking finding of our study is the association between loss of <italic>SMARCB1/INI1</italic> expression and a worse clinical outcome, regardless of the tumor stage and MMR status. Unexpectedly, even SMARCB1/INI1-high expression is an adverse prognostic indicator in comparison with SMARCB1/INI1-low expressing tumors. The reasons for this apparent contradiction are not clear: they might be linked to the specific deregulated cross-talks between chromatin remodelling components, acquisition of mesenchymal markers and genomic alterations such as chromosomal instability (CIN). Although still debated, it has been suggested that <italic>SMARCB1/INI1</italic> could have a critical function in determining aneuploidy [<xref ref-type="bibr" rid="B23">23</xref>]. Indeed, a subgroup of CRCs with enhanced SMARCB1/INI1 expression has a consistent proportion of aneuploid cells, even exhibiting MMR deficiency (our unpublished data); these latter tumors, in fact, typically show a near-diploid karyotype [<xref ref-type="bibr" rid="B8">8</xref>]. Whether and how SMARCB1/INI1 dysfunctions are causally implicated in genomic instability remains controversial. We further investigated the SMARCB1/INI1 prognostic significance by exploring its effect in combination with the <italic>TP53</italic> status. Interestingly, the SMARCB1/INI1<sup>-</sup>/p53<sup>+</sup> tumor group is closely correlated with very short survival and liver metastases when compared with other SMARCB1/INI1/p53 combinations, demonstrating a cooperative effect of both genes in restraining cancer aggressiveness in CRC advanced stages [<xref ref-type="bibr" rid="B12">12</xref>]. These results evoke the dramatic increase in tumor formation and metastasis obtained by inactivating <italic>TP53</italic> in <italic>SMARCB1/INI1</italic>-heterozygous mice. The clinical relevance of deregulated <italic>SMARCB1/INI1</italic> expression is confirmed in two independent CRC datasets of 226 and 146 patients, respectively, providing support to our findings. By interrogating genome-wide expression data, we identified several genes that were coordinately down- or up-regulated and separated in two distinct clusters. Notably, the top genes of the signature (<italic>BCR, COMT, MIF)</italic> map to the long arm of chromosome 22 at the cytogenetic band 22q11.23, closely associated with SMARCB1/INI1. The gene expression signature was confirmed also in CRC cell lines displaying molecular features of enhanced vimentin expression, reduced CDX2 and more mesenchymal phenotype. A chromosomal rearrangement (translocation/deletion) at 22q12 has recently been identified in a RCT and correlated with high SMARCB1/INI1 expression [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. A further translocation involving TTC28 at 22q12.1 or focal amplification of multiple genes mapped at 22q12.3 has been reported by the Cancer Genome Atlas Network and correlated with tumor aggressiveness [<xref ref-type="bibr" rid="B21">21</xref>]. Based on these evidences, is tempting to speculate that a number of alterations, such as translocations or amplifications, involving a specific region on the long arm of chromosome 22 might be associated with clinical aggressiveness and a more mesenchymal phenotype.</p><p>In conclusion, we demonstrate that SMARCB1/INI1 deficiency, alone or in combination with <italic>TP53</italic> mutations, influences the CRC aggressive behavior, regardless of the MMR status. Surprisingly, even SMARCB1/INI1 diffuse expression is associated with poor survival, as confirmed in two independent cohorts of patients. We identify several over-expressed or repressed genes located on chromosome 22, close to <italic>SMARCB1/INI1</italic> and coordinately deregulated. Our findings suggest that <italic>SMARCB1/INI1</italic> and genetic hot spots mapping to the long arm of chromosome 22 play an important role in tumor metastatic spreading. SMARCB1/INI1 might then be a useful clinical prognostic marker to complement the histological examination and grading and to select patients for adjuvant medical treatments. Mechanistic and larger clinical studies are needed to define how chromatin remodelling components and which specific genomic rearrangements influence the CRC metastatic behavior.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>Conceived the ideas for this study MP, AR; pathology analysis AR, EM, ADB; MR, DB; Acquisition, analysis and interpretation of data (acquired and managed patients, provided facilities, carried out the immunohistochemistry studies etc.), MP, AR, CZ, LS, CL, LA, PP, LP, AP, FG, AT, AB, RV; Development of methodology (e.g., statistical analysis, biostatistics, computational analysis) MP; CL and MC; Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): MP, EM, LA, PP, LP, AP, FG, AT, AB, RV, CL, AR; Wrote the manuscript MP; AR and VC; All authors read and approved the final manuscript.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional file 1: Table S1</title><p> SMARCB1/INI1 expression profiles and molecular markers of tumor differentiation stratified according to the MMR status. <bold>Table S2.</bold> List of the 45 genes whose expression changes significantly correlate with <italic>INI1</italic> deregulation (SMARCB1 in the list) relative to Cohort I that comprises 226 patients. The negative value indicates under-expressed genes. <bold>Figure S1.</bold> SMARCB1/INI1 expression and <italic>TP53</italic> mutation status correlate with prognosis in a CRC independent data set, cohort II. <bold>Figure S2.</bold> Top gene networks identified through integrative pathways analysis. <bold>Figure S3.</bold> The gene signature mapping to chromosome 22 close to <italic>SMARCB1/INI1</italic> is maintained in a panel of CRC cell lines.</p></caption><media xlink:href="1479-5876-11-297-S1.doc"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
Epithelium percentage estimation facilitates epithelial quantitative protein measurement in tissue specimens	<sec><title>Background</title><p>The rapid advancement of high-throughput tools for quantitative measurement of proteins has demonstrated the potential for the identification of proteins associated with cancer. However, the quantitative results on cancer tissue specimens are usually confounded by tissue heterogeneity, e.g. regions with cancer usually have significantly higher epithelium content yet lower stromal content.</p></sec><sec><title>Objective</title><p>It is therefore necessary to develop a tool to facilitate the interpretation of the results of protein measurements in tissue specimens.</p></sec><sec><title>Methods</title><p>Epithelial cell adhesion molecule (EpCAM) and cathepsin L (CTSL) are two epithelial proteins whose expressions in normal and tumorous prostate tissues were confirmed by measuring staining intensity with immunohistochemical staining (IHC). The expressions of these proteins were measured by ELISA in protein extracts from OCT embedded frozen prostate tissues. To eliminate the influence of tissue heterogeneity on epithelial protein quantification measured by ELISA, a color-based segmentation method was developed in-house for estimation of epithelium content using H&E histology slides from the same prostate tissues and the estimated epithelium percentage was used to normalize the ELISA results. The epithelium contents of the same slides were also estimated by a pathologist and used to normalize the ELISA results. The computer based results were compared with the pathologist’s reading.</p></sec><sec><title>Results</title><p>We found that both EpCAM and CTSL levels, measured by ELISA assays itself, were greatly affected by epithelium content in the tissue specimens. Without adjusting for epithelium percentage, both EpCAM and CTSL levels appeared significantly higher in tumor tissues than normal tissues with a p value less than 0.001. However, after normalization by the epithelium percentage, ELISA measurements of both EpCAM and CTSL were in agreement with IHC staining results, showing a significant increase only in EpCAM with no difference in CTSL expression in cancer tissues. These results were obtained with normalization by both the computer estimated and pathologist estimated epithelium percentage.</p></sec><sec><title>Conclusions</title><p>Our results show that estimation of tissue epithelium percentage using our color-based segmentation method correlates well with pathologists' estimation of tissue epithelium percentages. The epithelium contents estimated by color-based segmentation may be useful in immuno-based analysis or clinical proteomic analysis of tumor proteins. The codes used for epithelium estimation as well as the micrographs with estimated epithelium content are available online.</p></sec>	<contrib contrib-type="author" id="A1"><name><surname>Chen</surname><given-names>Jing</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>chenjing@jhmi.edu</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Toghi Eshghi</surname><given-names>Shadi</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>shadi@jhu.edu</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Bova</surname><given-names>George Steven</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>gsb@telamon1.us</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Li</surname><given-names>Qing Kay</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>qli23@jhmi.edu</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Li</surname><given-names>Xingde</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>xingde@jhu.edu</email></contrib><contrib contrib-type="author" corresp="yes" id="A6"><name><surname>Zhang</surname><given-names>Hui</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>hzhang32@jhmi.edu</email></contrib>	Clinical Proteomics	<sec sec-type="intro"><title>Introduction</title><p>The rapid advancement of high-throughput tools for measurement of proteins from cancer tissues or body fluids has demonstrated the potential for the identification of proteins associated with diseases in all areas of medicine. Most of these high-throughput tools utilize either mass spectrometry (MS)-microarray-, or immunosorbent assays for quantitative analysis of proteins [<xref ref-type="bibr" rid="B1">1</xref>]. With the advantage of quantitative measurement, currently, many protein assays with good sensitivity and specificity have been developed for research and clinical use in serum, urine and other body fluids. However, the analysis of proteins in tissue specimens is limited to the semi-quantitative immunohistochemistry (IHC) assay that are required to obtain the tissue spatial information and cell type-specific staining patterns. The usage of quantitative protein assays such as MS, microarray, or enzyme linked immunosorbent assay (ELISA) on tissue specimens, however, has its limitations. Due to the loss of spatial information, the measurements acquired are usually confounded by tissue heterogeneity. Since tissue specimens contain various types of cells, where the expressions of target proteins differ, protein assay results become hard to interpret and may even be misleading.</p><p>With respect to cancer research, assessment of the expression of epithelial proteins is of great interest, since over 90% of the carcinoma is of epithelial origin [<xref ref-type="bibr" rid="B2">2</xref>]. Compared to regions with normal tissue, regions with cancer usually have significantly higher epithelium content yet lower stromal content. Depending on tumor density, the epithelium to stroma ratio may vary considerably and may influence protein quantitation readings significantly when an epithelial protein is concerned, e.g. a higher epithelial protein reading in tumor tissues might be solely due to the increased epithelial content of the epithelium rather than the biological overexpression of that protein. Therefore, it would be important to consider the epithelium content when we analyze the protein levels using quantitative protein assays.</p><p>There are a number of approaches to identify and quantify epithelium content from histology slides. Traditionally, the epithelium contents are read based on nuclei counts from a hematoxylin and eosin (H&E) stained histology slide by a pathologist. Another approach is to stain the histology slide with anti-cytokeratin antibody CAM 5.2 (staining for epithelia) and Masson trichrome (staining for collagenous stromal structures) [<xref ref-type="bibr" rid="B3">3</xref>]. More recently, with the digitization of whole slide imaging, a number of algorithms have been developed for computer-assisted readings. These methods rely on image features such as morphology, texture, color and intensity to segment images and classify them into various pathologically different regions. Automated histopathological image analysis reduces the inter-and intra-observer errors and provides additional quantitative information to aid diagnosis [<xref ref-type="bibr" rid="B4">4</xref>]. However, to our knowledge, these measurements have never been utilized for protein measurements.</p><p>Epithelial cell adhesion molecule (EpCAM) and cathepsin L (CTSL) are epithelial proteins that have been found abundantly expressed in prostate adenocarcinomas [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B6">6</xref>]. EpCAM is a well known tumor associated antigen and is expressed in various adenocarcinomas and squamous cell carcinomas (e.g. prostate, lung, colon, gastric carcinomas) [<xref ref-type="bibr" rid="B7">7</xref>-<xref ref-type="bibr" rid="B9">9</xref>]. Its expression on normal epithelia, on the other hand, is rather variable yet much lower than the carcinoma cells [<xref ref-type="bibr" rid="B10">10</xref>]. CTSL is a lysosomal cystein proteinase that plays a major role in the catabolism of intracellular and extracellular proteins [<xref ref-type="bibr" rid="B6">6</xref>]. Studies on prostate cancer cell lines suggested that CTSL was associated with the motility of prostate tumor cells and therefore might be involved in tumor metastasis [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B12">12</xref>]. Although previous studies suggested an increase in CTSL mRNA expression in prostate adenocarcinomas [<xref ref-type="bibr" rid="B13">13</xref>], a recent study showed that CTSL staining in prostate tissues is comparable between prostate adenocarcinomas and normal tissues [<xref ref-type="bibr" rid="B14">14</xref>].</p><p>In this study, we assessed EpCAM and CTSL levels with ELISA in prostate cancer tissues and determined the effect of epithelium content on tissue protein quantitation. To determine the effect of tissue heterogeneity on the interpretation of the ELISA result, we developed an in-house color-based segmentation method for estimation of epithelium content and applied the method on ELISA results. A pathologist estimation of the epithelium content was also applied on ELISA results. We found that both EpCAM and CTSL levels, measured by ELISA itself, were greatly affected by epithelium content in the tissue specimens. However, after normalization by epithelium percentage, ELISA measurements of both EpCAM and CTSL were in agreement with IHC staining results, demonstrating the need of normalization using epithelium content in quantitative measurement of epithelial proteins in tissue specimens.</p></sec><sec sec-type="materials\|methods"><title>Materials and methods</title><sec><title>Materials</title><p>LSAB + Kits, biotin blocking system, antibody dilution buffer were from Dako, Carpinteria, CA. Goat-anti-CTSL antibody, antigen retrieval buffer, recombinant protein, capture and detection antibody of human CTSL, EpCAM, streptavidin-HRP conjugates and ELISA plates were from R&D Systems, Minneapolis, MN. All other chemicals were from Sigma-Aldrich (St. Louis, MO).</p></sec><sec><title>Clinical specimens</title><p>Samples and clinical information were obtained with informed consent and performed with the approval of the Institutional Review Board of the Johns Hopkins University. Formalin fixed paraffin embedded (FFPE) prostate tissue slides were acquired for 6 individuals with primary prostate tumors. Additional thirty-six OCT-embedded prostate tumors were collected from radical prostatectomy at Johns Hopkins Hospital and Johns Hopkins Bayview Medical Center under the NCI-funded Johns Hopkins prostate cancer SPORE project. These tumors includes nineteen specimens with a Gleason score of 6, seven specimens with a Gleason score of 7, five specimens with a Gleason score of 8 and five specimens with a Gleason score of 9 (Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Table S1). Eight OCT-embedded normal prostate tissues were collected from healthy transplant donors. All specimens were snap-frozen, embedded in OCT and stored at −80°C till use.</p></sec><sec><title>Immunohistochemical staining and tissue microarrays</title><p>IHC staining was performed on FFPE prostate tissue slides from 6 individuals with primary prostate tumors. Sections of tissue were deparaffinized and rehydrated. Tissues were incubated in antigen retrieval buffer at 92-95°C for 10 min. CTSL was stained with Universal LSAB™ + Kits per manufacture’s protocol. Briefly, tissues were blocked by peroxidase block and 3% BSA/PBS for 30 min each followed by avidin and biotin block with Biotin Blocking System for 15 min each at room temperature. The tissues were then incubated with goat anti-CTSL primary antibody in antibody dilution buffer at 4 μg/mL followed by incubation with anti-goat biotin labeled secondary antibody and high sensitivity streptavidin-HRP for 30 min each. The CTSL staining was detected with DAB chromogen.</p></sec><sec><title>Measurements of proteins from clinical specimens using ELISA</title><p>The protein samples were collected by sectioning the OCT-embedded frozen prostate tissues. The adjacent sections of about every 15 tissue sections (6 μm each) were stained with H&E for use in the computer-aided and pathologist estimation of epithelium content. For tumor specimens, the adjacent H&E slides were also used for cryostat micro-dissection to enrich the tumor tissue in the collected sample for immunoassay analysis. Places where tissues were trimmed were marked in the H&E slides and excluded for epithelium percentage estimation. An estimated number of ten to twenty 6 μm-thick tissue sections were collected in sterile screw-cap bullet tubes for each sample for protein assays. Proteins were then extracted from tissue sections using cell lysis buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 0.1% SDS, 0.5% Na Deoxycholate, 1% Triton × 100). BCA assay was performed to determine and adjust the protein concentration for each tissue sample to 1 μg/mL with PBS. Tissue EpCAM and CTSL levels were then measured with ELISA assay as described before [<xref ref-type="bibr" rid="B15">15</xref>]. Briefly, CTSL (1 μg/mL) or EpCAM (4 μg/mL) capture antibody were coated overnight in a 96-well plate. The wells were then blocked with 3% BSA, incubated with 100 μL diluted sample, with CTSL (0.5 μg/mL) or EpCAM (0.2 μg/mL) biotinylated detection antibody for 1 h each, and with streptavidin-HRP conjugates (1:200) for 30 min. The assays were then developed with TMB substrate, stopped with H<sub>2</sub>SO<sub>4</sub> and measured by reading the plate at 450 nm with a spectrophotometer.</p></sec><sec><title>Estimation of epithelium ratio in prostate tissue specimens</title><p>An in-house color-based segmentation method was developed for estimation of epithelial areas in prostate tissue specimens using H&E stained face sections of prostate tissues. For each of the 44 cases (36 tumors and 8 normal prostate tissues), digital slides were acquired by scanning the H&E stained slide with AT Turbo (Aperio technologies, Vista, CA.). Every 2.1 × 1.3 mm<sup>2</sup> area of the micrograph was then saved into a .tiff file at a resolution of 72 pixels per inch using the ImageScope software (Aperio technologies, Vista, CA.) and an estimated number of 13 ± 10 image files were generated for each one of the 44 cases. One image from each case was randomly selected to serve as the training image for the classification algorithm and the rest of the images were used as the test image set. The training images were used as the input to the classification training code. All computer simulations were implemented in MATLAB (Mathworks, Natick, MA). Each training image was segmented into four regions based on the pixel colors using a k-means clustering algorithm. K-means clustering algorithm is a clustering analysis tool for grouping a number of observations into k clusters based on the similarities between the observations. Briefly, the observations were randomly assigned to clusters for initialization and the centroid of each cluster was calculated. In an iterative manner, the cluster of each observation was updated to its nearest centroid and the centroids of the clusters were re-calculated to reflect the changes to the clusters, until the centroids converged to the optimal values. In other words, each color cluster was formed by minimizing the squared euclidean distance of the cluster members to its centroid. This will group pixels with similar colors together in a color cluster of white, bright pink, dark pink or purple in an H&E stained slide where white, pink and purple correspond to lumen, stroma and epithelium respectively. Each micrograph was arranged into 20 × 20 pixel grids such that each grid covered a 0.04 × 0.04 mm<sup>2</sup> area of the tissue section. The ratio of the area of the four colors to the total area of the cell was calculated for each grid cell. Clearly, the resultant four color ratios would sum to 1 in each grid cell; thus only three of the color ratios were linearly independent. The epithelial regions of the original training image were manually marked by an experienced researcher to serve as the benchmark for the training of the classification algorithm. The marked epithelial regions are shown by a green shade. The grid cells were then divided into two groups based on whether they were marked as epithelium or not and illustrated on a scatterplot in the space of the three base colors of the H&E micrographs. Any grid cell with white content greater than 70% was marked as luminal. Knowing the class of each of the grid cells based on the marked epithelial regions, K-means clustering was again used to divide the space of the three base colors into three clusters representing the epithelial, stromal and luminal regions. These clusters, established in the space of three optionally selected color ratios, were then used to segment the images in the test dataset into epithelium, stroma and lumen area, thus estimating the percentage of each in the whole face section of prostate tissue. Segmentation code worked similar to the training code as the test image was segmented into four colors and the color ratios were calculated for each grid cell on the image. Each grid cell was classified into epithelial, stromal or luminal depending on its distance from the established clusters in the space of the three color ratios. The epithelium percentage was calculated by the following formula: Epithelium area/(Epithelium area + Stroma area) × 100%. The Matlab codes and H&E micrographs used for epithelium estimation, as well as the micrographs with estimated epithelium content are available for download at <ext-link ext-link-type="uri" xlink:href="http://sdrv.ms/17tWsqd">http://sdrv.ms/17tWsqd</ext-link>.</p></sec><sec><title>Statistical analysis</title><p>Wilcoxon signed rank order test (unpaired, two-sided) was used for determination of statistical significance of EpCAM and CTSL immunoassay measurements.</p></sec></sec><sec sec-type="results"><title>Results</title><sec><title>IHC staining of CTSL in prostate tissue specimens</title><p>To assess CTSL expression level in prostate tissue specimens, IHC staining was performed on 6 FFPE prostate tumor cases. Figure <xref ref-type="fig" rid="F1">1</xref> shows representative fields of CTSL stained and H&E stained slides from 3 representative cases of the 6 FFPE cases. Similar to that reported before [<xref ref-type="bibr" rid="B14">14</xref>], we found that CTSL stained epithelium with no staining in the stroma compartment. Medium to strong staining of CTSL was observed in all 6 prostate tumor cases. Heterogeneity in staining intensity was observed in both tumor and adjacent normal epithelium. No differences in staining intensity were observed between tumors and the adjacent normal tissues. These results suggested that the expression of CTSL in the prostate epithelium was similar between normal and tumor tissues.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Immunohistochemical staining of prostate tumor tissues.</bold> Tumor tissue slides from 6 individuals were stained with anti-cathepsin L antibody and then counterstained with hematoxylin. Representative images from 3 individuals were shown in <bold>(A)</bold>, <bold>(B)</bold>, and <bold>(C)</bold> with corresponding hematoxylin and eosin stained reference images as shown in <bold>(D)</bold>, <bold>(E)</bold> and <bold>(F)</bold>.</p></caption><graphic xlink:href="1559-0275-10-18-1"/></fig></sec><sec><title>ELISA measurements of EpCAM and CTSL in prostate tissue specimens</title><p>To measure EpCAM and CTSL level in prostate tissue specimens, ELISA assays were developed for EpCAM and CTSL. The limits of detection (LOD) calculated as background OD ± 3SD were 188 and 27 pg/mL for EpCAM and CTSL respectively; the dose responsive ranges were from 0 to 2 ng/mL for both CTSL and EpCAM (Figure <xref ref-type="fig" rid="F2">2</xref>A and B). At the lowest standard point, the coefficients of variation (CV) were 0.78% and 6.3% for EpCAM and CTSL respectively. The intra-assay CVs were 3.2% and 9.9% for EpCAM and CTSL respectively. These assays were then used to analyze 8 normal prostate tissues and 36 prostate tumors. The average concentration of EpCAM was 9.39 ± 4.22 ng/mg total protein for normal tissues and 44.61 ± 23.40 ng/mg for prostate tumors (Figure <xref ref-type="fig" rid="F2">2</xref>C). The average concentration of CTSL measured was 6.30 ± 2.06 ng/mg total protein for normal tissues and 11.83 ± 4.56 ng/mg total protein for prostate tumors (Figure <xref ref-type="fig" rid="F2">2</xref>D). Compared to normal tissues, immunoassay data showed that both EpCAM and CTSL expression were significantly increased in prostate tumors with a 4.75 fold increase for EpCAM and a 1.88 fold increase for CTSL. However, as discussed earlier, since tissue protein quantitation using ELISA assay is influenced by both biological expression of the target protein and the percentage of epithelial cells expressing the proteins, further analysis needs to be done to determine the percentage of epithelial cells in order to elucidate the biological expression of EpCAM and CTSL between normal prostate tissues and prostate tumors.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Immunoassay measurements of epithelial cell adhesion molecule (EpCAM) and cathepsin L (CTSL) protein expression. (A)</bold> Standard curve of EpCAM. <bold>(B)</bold> Standard curve of CTSL. <bold>(C)</bold> EpCAM measurement of tissue lysate from 8 normal prostate tissues and 36 prostate tumors. <bold>(D)</bold> CTSL measurement of tissue lysate from 8 normal prostate tissues and 36 prostate tumors. *, p < 0.001, compared to normal prostate tissues.</p></caption><graphic xlink:href="1559-0275-10-18-2"/></fig></sec><sec><title>Estimation of epithelium percentage in prostate tissue</title><p>To account for the epithelium content of the tissue specimens, we used two methods: 1) estimation of epithelium content by a board certified pathologist; 2) estimation of epithelium content by a computer-aided method. The H&E stained adjacent sections of all 44 cases where CTSL and EpCAM levels were measured were analyzed for epithelium content with both methods. For the computer aided method, we developed an in-house color-based segmentation method. The 44 training H&E images, 1 from each case (Figure <xref ref-type="fig" rid="F3">3</xref>A), were segmented into four colors representing the luminal (white color, Figure <xref ref-type="fig" rid="F3">3</xref>B), stromal (light and dark pink, Figure <xref ref-type="fig" rid="F3">3</xref>C and D) and epithelial (purple, Figure <xref ref-type="fig" rid="F3">3</xref>E) regions of the tissues. Each image was then divided into grid cells each covering 0.04 × 0.04 mm<sup>2</sup> of the tissue section (Figure <xref ref-type="fig" rid="F3">3</xref>F). The grid cells were already marked by a researcher as epithelial or non-epithelial. Figure <xref ref-type="fig" rid="F3">3</xref>G shows a scatter plot of the grid cells of Figure <xref ref-type="fig" rid="F3">3</xref>F in the space of the three of the four H&E base colors (white, light pink, dark pink and purple). Each dot represents one of the grid cells, which depending on its color was marked as epithelial (red) or non-epithelial (blue). The three axes of the image stand for the proportion of the pixels in each grid cell that fall into the corresponding color cluster. The non-epithelial and epithelial cloud of the dots in the defined three-dimensional space show low overlap, suggesting that a clustering algorithm can build clusters for classification of grid cells into epithelial and non-epithelial according to the color ratios in the cells. Therefore, the marked and unmarked grid cells on the training image along with their color ratios were used as inputs to the K-means clustering algorithm for generating the epithelial, stromal and luminal clusters in the space of the three color ratios. These clusters, established in the space of three optionally selected color ratios, were then used to segment the images in the test image set (13 ± 10 micrographs per case, 44 cases) into epithelium, stroma and lumen area for calculation of epithelium percentage.</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Color based segmentation and k-means clustering of grid cells into epithelial and non-epithelial regions based on color area ratios.</bold> A representative micrograph of prostate tissue section at 20× <bold>(A)</bold> is segmented into four regions based on the pixel colors of <bold>(B)</bold> white, <bold>(C)</bold> light pink, <bold>(D)</bold> dark pink and <bold>(E)</bold> purple, using a k-means clustering algorithm. <bold>(F)</bold> The epithelial areas of the training image were marked by an experienced prostate cancer researcher and were arranged into 20 × 20 pixel grid cells. <bold>(G)</bold> The four color ratios were calculated in each grid cell. Knowing the epithelial and non-epithelial regions in training sets, we classify the grid cells into two clusters. A scatter plot shows these clusters in the space of three colors, which have small overlap.</p></caption><graphic xlink:href="1559-0275-10-18-3"/></fig><p>Figure <xref ref-type="fig" rid="F4">4</xref>A and C shows a representative micrograph of normal tissue and cancer specimen where the predicted epithelium is highlighted in green. Figure <xref ref-type="fig" rid="F4">4</xref>B and D shows the false positive and false negative regions, marked in blue and magenta respectively. Cross-validation was performed on 44 micrographs from the training image set (44 cases) to assess the accuracy of this epithelium prediction method. For this analysis, 50% of the grid cells of each of the training images were randomly selected to form the training dataset for k-means clustering, while the remaining 50% of the cells were saved to form the validation dataset for cross-validation to determine the accuracy of the method. According to cross-validation results, the method predicted epithelium ratio with an accuracy of 84.41%; with the 44 H&E histology micrographs tested, the false positive rate was 8.79 ± 5.06% and the false negative rate was 8.12 ± 5.35%. Such proximity of false positive and false negative rates improves the estimation of total epithelial area. The determination coefficient (R<sup>2</sup>) between the estimated epithelium area and marked epithelial area in the validation set was 0.965 (Figure <xref ref-type="fig" rid="F4">4</xref>E).</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>Method output and cross-validation.</bold> The in-house color-based segmentation algorithm was implemented on micrographs of H&E stained prostate tissue sections of 8 normal prostate tissues and 36 prostate tumors to segment the testing images into epithelial and non-epithelial regions. The method output is depicted for a normal <bold>(A, B)</bold> and a cancerous <bold>(C, D)</bold> tissue sections. In the left column, the estimated epithelial regions are highlighted in green. In the right column, the false positive and false negative regions are highlighted in blue and magenta, respectively. The set of 44 H&E histology micrographs representing 44 cases was divided into training and validation datasets. The epithelial, lumenal and stromal clusters were formed by analyzing the training set for each subject. The performance of the classifier was then evaluated by examining the training and validation dataset. <bold>(E)</bold> The estimated and actual epithelial areas of the validation datasets were well correlated (R<sup>2</sup> = 0.965).</p></caption><graphic xlink:href="1559-0275-10-18-4"/></fig></sec><sec><title>Normalization of EpCAM and CTSL measurements with estimated epithelium percentage</title><p>Subsequently, using the in-house developed method, we estimated the epithelium percentage of the 8 normal prostate tissues and 36 prostate tumors whose EpCAM and CTSL levels were measured. The average epithelium percentage in normal prostate tissues was 24.14 ± 5.58% (Figure <xref ref-type="fig" rid="F5">5</xref>A). This ratio was similar to what was reported before [<xref ref-type="bibr" rid="B3">3</xref>]. The average epithelium percentage in prostate tumors was 57.98 ± 19.75%. As expected, the epithelium percentage was significantly higher in tumors compared to that in normal tissues. Similarly, the results from pathologist estimation showed that epithelium percentage in normal prostate tissues was 33.75 ± 11.88% with 62.36 ± 15.51% estimated for prostate tumors. The computer aided and pathologist estimated epithelium percentages are statistically positively related (p < 0.001), with a Pearson correlation coefficient of 0.72 (Figure <xref ref-type="fig" rid="F5">5</xref>B).</p><fig id="F5" position="float"><label>Figure 5</label><caption><p><bold>Immunoassay measurements of epithelial cell adhesion molecule (EpCAM) and cathepsin L (CTSL) protein expression after normalization by epithelium percentage. (A)</bold> Epithelium percentage estimated for 8 normal and 36 cancer tissue slides with computer-aided classification and pathologist estimation. <bold>(B)</bold> Scatter plot of computer-aided epithelium percentage estimated vs. pathologist estimation. <bold>(C)</bold> EpCAM ELISA measurements were adjusted with epithelium percentage estimated with the computer-aided method. <bold>(D)</bold> CTSL ELISA measurements were adjusted with epithelium percentage estimated with the computer aided method. <bold>(E)</bold> EpCAM ELISA measurements were adjusted with epithelium percentage estimated by a pathologist. <bold>(F)</bold> CTSL ELISA measurements were adjusted with epithelium percentage estimated by a pathologist. Corr: Pearson’s correlation coefficient. , p < 0.01, compared to normal prostate tissue. ***, p < 0.001, compared to normal prostate tissue.</p></caption><graphic xlink:href="1559-0275-10-18-5"/></fig><p>As EpCAM and CTSL are expressed in the prostatic epithelium from IHC staining results, we used the epithelium percentage to normalize the EpCAM and CTSL ELISA results. After computer-aided normalization, the average measured EpCAM was 39.11 ± 16.69 and 82.70 ± 39.56 ng/mg total protein/epithelium percentage for normal prostate tissues and prostate tumors respectively (Figure <xref ref-type="fig" rid="F5">5</xref>C); the average measured CTSL was 27.27 ± 10.38 and 23.44 ± 12.80 ng/mg total protein/epithelium percentage for normal prostate tissues and prostate tumors (Figure <xref ref-type="fig" rid="F5">5</xref>D). After normalization by the pathologist estimated epithelium percentage, the average measured EpCAM was 32.02 ± 21.32 and 73.27 ± 34.12 ng/mg total protein/epithelium percentage for normal prostate tissues and prostate tumors respectively (Figure <xref ref-type="fig" rid="F5">5</xref>E); The average measured CTSL was 19.99 ± 7.78 and 20.74 ± 10.53 ng/mg total protein/epithelium percentage for normal prostate tissues and prostate tumors (Figure <xref ref-type="fig" rid="F5">5</xref>F). With both epithelium estimations, EpCAM expression was significantly increased in prostate tumors by about 2 fold, compared to the 4.75 fold increase without normalization (Figure <xref ref-type="fig" rid="F3">3</xref>C). In contrast to the significant elevated CTSL expression in prostate tumors from ELISA results without normalization by epithelial content (Figure <xref ref-type="fig" rid="F3">3</xref>D), CTSL expression was comparable between normal tissues and prostate tumors after epithelium normalization. These results are in agreement with previous reports where IHC staining was used to assess the protein expression of EpCAM and CTSL [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B14">14</xref>], demonstrating the positive impact of epithelium normalization in analyzing immunoassay results.</p></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>Tissue specimen is a great source for identification of disease related molecules, e.g. cancer related molecules/markers. To evaluate protein expression in tissue specimens, IHC staining is one of the most common techniques utilized. IHC staining provides insight into tissue heterogeneity, disease relevance of protein markers, and the expression pattern of protein markers in different cell types. However, IHC staining is subject to inter-observer error and is at best semi-quantitative. Direct measurements of proteins by mass spectrometry, microarray, or immunosorbent assay, on the other hand, are quantitative and can be standardized for quality assurance. Consequently, use of immunosorbent assays to measure protein expression in tissue specimens is highly desirable if the result can be properly interpreted.</p><p>In this study, we introduced epithelial percentage normalization as a tool in interpreting immunoassay results for epithelial proteins. We developed a tool for automated segmentation of micrograph slides into epithelial and non-epithelial regions using k-means clustering. K-means clustering is one the fastest and simplest clustering analysis tools that can reduce large datasets into smaller, more manageable subspaces based on the similarities observed in the dataset. The proposed epithelial percentage estimation method segments the image into four colors that are most dominant in typical H&E micrograph slides. The four colors are determined by the k-means clustering of the single pixels on each micrograph. Therefore, although these colors are categorized as white, light and dark pink and purple, they might slightly vary from one slide to the other depending on the strength of the staining on each slide. Subsequently, this method automatically accounts for the variations between the color staining on different tissue sections, which is a significant challenge in universal image processing of histology slides. In addition, this algorithm achieved an accuracy of 84% on a database of normal and prostate cancer tissue sections. The false classification of 16% was almost equally divided between the false positive and false negative results. Although the false positive and negative results are not desirable and must be minimized, in the context of epithelium percentage estimation, the inaccuracy introduced by false classifications is cancelled out to a great degree. Therefore, the equity of false positive and negative results explains the high correlation of the estimated epithelium percentage with the marked epithelial percentage, which is desirable for normalizing and interpreting the immunoassay for determining the biological changes in protein expression.</p><p>It needs to be noted that to use epithelium percentage estimation in epithelial protein measurement in tissue specimens, the H&E stain needs to be representative of the entire tissue that is studied for a given protein, and the protein needs to be measured on the same exact piece of tissue. This is because the range of epithelial to stromal ratios on a given mass of tissue varies greatly depending on the size of the tissue and the homogeneity of the tissue structures (e.g. a given prostate cancer could be 80% epithelial in some area and 5% in others). In this study, to ensure the accuracy of epithelium percentage estimated, adjacent H&E slide of approximately 15 sections of prostate tissues (6 μm each) was used for epithelium estimation.</p><p>With EpCAM and CTSL expressions in prostate tumor tissues, we demonstrated that normalization by epithelial percentage is useful in analyzing ELISA results for epithelial proteins. With IHC staining carried out in this study and in a previously published study [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B14">14</xref>], we showed that CTSL expression was not significantly different between normal tissue and prostate tumors after epithelium normalization. While ELISA itself misleadingly showed a significant increase of CTSL in prostate tumors, with normalization by epithelium percentage, ELISA analysis also showed that CTSL expression was comparable between these two groups. EpCAM was shown to be up-regulated in prostate carcinomas in a number of studies with IHC staining [<xref ref-type="bibr" rid="B7">7</xref>-<xref ref-type="bibr" rid="B9">9</xref>]. In this study, we also found a significant increase in EpCAM with immunoassay analysis both with and without normalization by epithelium percentage. However, the difference of EpCAM expression between tumor and normal tissues dropped significantly from 4.68 to around 2, which better depicted the biological differences of EpCAM between tumor and normal cells.</p><p>In addition to facilitate analysisof ELISA results of epithelial proteins, epithelium percentage estimation can also be used in other quantitative assays (e.g. mRNA expression, protein activity assay, clinical proteomic analysis etc.) where spatial information is lost due to sample homogenization to account for epithelium heterogeneity. With epithelium percentage normalization, improvement in the accuracy of biochemical measurements in homogenized tissue specimens may be achieved. In addition, by accounting for tissue heterogeneity, interesting new protein markers may be identified. However, further studies needs to be carried out in testing the accuracy of epithelium percentage estimation.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>In summary, we developed an in-house color-based segmentation method for estimation of epithelium content and demonstrated the accuracy of the method in epithelium estimation. Using EpCAM and CTSL as examples, we demonstrated that protein expressions measured by immunoassays correlate well with that measured IHC staining, suggesting that normalization by epithelium percentage is helpful in interpreting ELISA and similarly other biochemical or proteomics based assay results.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>Dr. JC carried out the ELISA measurements and normalization of ELISA results by epithelium percentage and participated in the study design, manuscript writing, data analysis and development of the color segmentation based method for epithelium percentage estimation. STE developed the color segmentation based method for epithelium percentage estimation, preformed the immunohistochemical staining of cathepsin L and participated in the manuscript writing. Dr. GSB provided the clinical samples and participated in the pathologist estimation of epithelium percentages of the samples. Dr. QKL participated in the pathologist estimation of epithelium percentages of the samples. Dr. XL participated in the development of the color segmentation based method for epithelium percentage estimation. Dr. HZ participated in the study design, manuscript writing and data analysis. All authors read and approved the final manuscript.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional file 1: Table S1</title><p>Clinical information of normal prostate and prostate tumor tissues employed in this study.</p></caption><media xlink:href="1559-0275-10-18-S1.xlsx"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
Chitotriosidase - a putative biomarker for sporadic amyotrophic lateral sclerosis	<sec><title>Background</title><p>Potential biomarkers to aid diagnosis and therapy need to be identified for Amyotrophic Lateral Sclerosis, a progressive motor neuronal degenerative disorder. The present study was designed to identify the factor(s) which are differentially expressed in the cerebrospinal fluid (CSF) of patients with sporadic amyotrophic lateral sclerosis (SALS; ALS-CSF), and could be associated with the pathogenesis of this disease.</p></sec><sec><title>Results</title><p>Quantitative mass spectrometry of ALS-CSF and control-CSF (from orthopaedic surgical patients undergoing spinal anaesthesia) samples showed upregulation of 31 proteins in the ALS-CSF, amongst which a ten-fold increase in the levels of chitotriosidase-1 (CHIT-1) was seen compared to the controls. A seventeen-fold increase in the CHIT-1 levels was detected by ELISA, while a ten-fold elevated enzyme activity was also observed. Both these results confirmed the finding of LC-MS/MS. CHIT-1 was found to be expressed by the Iba-1 immunopositive microglia.</p></sec><sec><title>Conclusion</title><p>Elevated CHIT-1 levels in the ALS-CSF suggest a definitive role for the enzyme in the disease pathogenesis. Its synthesis and release from microglia into the CSF may be an aligned event of neurodegeneration. Thus, high levels of CHIT-1 signify enhanced microglial activity which may exacerbate the process of neurodegeneration. In view of the multifold increase observed in ALS-CSF, it can serve as a potential CSF biomarker for the diagnosis of SALS.</p></sec>	<contrib contrib-type="author" equal-contrib="yes" id="A1"><name><surname>Varghese</surname><given-names>Anu Mary</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>v.anumary@gmail.com</email></contrib><contrib contrib-type="author" equal-contrib="yes" id="A2"><name><surname>Sharma</surname><given-names>Aparna</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>aparnasharma12@gmail.com</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Mishra</surname><given-names>Poojashree</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>poojashri.mishra@gmail.com</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Vijayalakshmi</surname><given-names>Kalyan</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>vijayalakshmi.kb@gmail.com</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Harsha</surname><given-names>Hindalahalli Chandregowda</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>harsha@ibioinformatics.org</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Sathyaprabha</surname><given-names>Talakad N</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>drsathyaprabha@gmail.com</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Bharath</surname><given-names>Srinivas MM</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>thathachar@rediffmail.com</email></contrib><contrib contrib-type="author" id="A8"><name><surname>Nalini</surname><given-names>Atchayaram</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>atchayaramnalini@yahoo.co.in</email></contrib><contrib contrib-type="author" id="A9"><name><surname>Alladi</surname><given-names>Phalguni Anand</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>alladiphalguni@gmail.com</email></contrib><contrib contrib-type="author" corresp="yes" id="A10"><name><surname>Raju</surname><given-names>Trichur R</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>trraju.nimhans@gmail.com</email></contrib>	Clinical Proteomics	<sec><title>Background</title><p>Selective loss of cortical and spinal motor neurons is the characteristic feature of Amyotrophic Lateral Sclerosis (ALS), an adult onset progressive fatal neurodegenerative disorder. Factors predisposing the most prominent form of this multifactorial disease viz. sporadic ALS (SALS) remain obscure due to the difficulties in developing animal models. Therefore, development of novel therapeutics is also severely hampered. This is also largely attributed to the lack of a ‘biomarker’ which can be objectively measured as an indicator of pathogenic processes and/or pharmacologic response to therapeutic interventions [<xref ref-type="bibr" rid="B1">1</xref>]. The discovery of ideal biomarkers may offer tools for rapid diagnosis, monitoring disease progression and provide insights into the pathophysiology of the disease; thereby broadening therapeutic options. Proximity to the central nervous system (CNS) renders Cerebrospinal Fluid (CSF) to be the ideal biofluid for detection of biomarkers in CNS pathologies. It is speculated that toxic agents which propagate the disease are synthesized in the affected areas, injure the neighboring cells, and are released into the extracellular space and CSF [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B3">3</xref>].</p><p>We have earlier shown that exposure of embryonic rat spinal cord cultures to ALS-CSF (<italic>in-vitro</italic>) and intrathecal injection of the same into neonatal rats (<italic>in-vivo</italic>) induced degenerative changes in motor neurons and showed the involvement of astrocytes [<xref ref-type="bibr" rid="B4">4</xref>-<xref ref-type="bibr" rid="B11">11</xref>]. Intracerebroventricular infusion of ALS-CSF in adult rats perturbed the cortical motor neuronal activity and was associated with poor motor performance [<xref ref-type="bibr" rid="B12">12</xref>]. Thus several studies, including ours, support the presence of toxic factor(s) in ALS-CSF and attribute them a role in eliciting its pathophysiology [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B13">13</xref>-<xref ref-type="bibr" rid="B15">15</xref>].</p><p>We undertook a study to identify the toxic factor(s) in ALS-CSF through proteomic analysis. Quantitative mass spectrometric analysis of ALS-CSF compared to age-matched controls showed upregulation of 31 proteins, amongst which Chitotriosidase-1 (CHIT-1) showed more than 10 fold increase. The biological significance of CHIT-1 expression is intriguing in view of the absence of its natural substrate chitin in human brain. Earlier studies have shown an increase in CSF CHIT-1 levels in multiple sclerosis (MS) and Alzheimer’s disease (AD) [<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B17">17</xref>].</p></sec><sec sec-type="results"><title>Results</title><sec><title>Confirmation of toxicity of ALS-CSF samples</title><p>The toxicity of the ALS-CSF samples was confirmed prior to proteomic analysis. Exposure of NSC-34 cells to the CSF obtained from ALS patients (ALS-CSF) resulted in a dramatic decrease in their viability when compared to the control groups; i.e. the cells treated with CSF from normal individuals (N-CSF) or without CSF (NC) (*p < 0.001 vs NC; ###p < 0.001 vs N-CSF; Figure <xref ref-type="fig" rid="F1">1</xref>A). It also induced enhancement of LDH activity in ALS-CSF group (p < 0.001 vs. NC and $$$p < 0.001 vs. N-CSF; Figure <xref ref-type="fig" rid="F1">1</xref>B).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Mass spectrometric analysis of ALS-CSF samples: Toxicity assays (A & B) were performed.</bold> Histogram of MTT assay showing 40% reduction in the viability of NSC-34 cells upon exposure to 10%(v/v) ALS-CSF compared to the cells exposed to normal-CSF (###p < 0.001 vs. N-CSF) and 30% compared to normal controls (p < 0.001 vs. NC; <bold>A</bold>). ALS-CSF caused significant increase in LDH activity when compared to control groups (p < 0.001 vs. NC and <sup>$$$</sup>p < 0.001 vs. N-CSF; <bold>B</bold>). Gel image representing depletion of abundant proteins prior to mass spectrometric analysis (<bold>C</bold>). Representative MS/MS spectra of the peptides of CHIT1, Osteopontin, CHI3L1 and CHI3L2 (<bold>D</bold>). Tests of significance was Student’s <italic>t</italic> test, One way Anova followed by Tukey’s post hoc analysis.</p></caption><graphic xlink:href="1559-0275-10-19-1"/></fig></sec><sec><title>Proteomic analysis of control and ALS-CSF samples</title><p>Ten CSF samples each from controls and ALS were pooled, depleted of abundant proteins and electrophoresed on SDS-PAGE (Figure <xref ref-type="fig" rid="F1">1</xref>C). Total protein from control and ALS-CSF were subjected to tryptic digestion, followed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) after labeling with isobaric tags for relative and absolute quantitation (iTRAQ). LC-MS/MS analysis identified 819 proteins using SEQUEST and Mascot (Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Table S1). Approximately 31 proteins showed more than 1.5-fold increase, suggesting an up- regulation (Additional file <xref ref-type="supplementary-material" rid="S2">2</xref>: Table S2) and about 17 proteins were down-regulated (decrease of 0.5 fold or more) in ALS-CSF samples compared to the normal controls (Additional file <xref ref-type="supplementary-material" rid="S3">3</xref>: Table S3). Four of the prominently up-regulated proteins were CHIT-1 (10 fold), osteopontin isoform-b (3 fold), chitinase-3-like protein 2 (CHI3L2; 2 fold) and chitinase-3-like protein 1 (CHI3L1; 1.7 fold). Thus CHIT-1 showed the most dramatic increase (Table <xref ref-type="table" rid="T1">1</xref>, Figure <xref ref-type="fig" rid="F1">1</xref>D).</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>List of proteins upregulated in ALS-CSF</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>S. No</bold></th><th align="left"><bold>Gene symbol</bold></th><th align="left"><bold>Protein name</bold></th><th align="left"><bold>Relative expression (ALS-CSF/N-CSF) fold change</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">1<hr/></td><td align="left" valign="bottom">CHIT1<hr/></td><td align="left" valign="bottom">chitotriosidase-1 precursor [Homo sapiens]<hr/></td><td align="left" valign="bottom">10<hr/></td></tr><tr><td align="left" valign="bottom">2<hr/></td><td align="left" valign="bottom">SPP1<hr/></td><td align="left" valign="bottom">osteopontin isoform b precursor [Homo sapiens]<hr/></td><td align="left" valign="bottom">3<hr/></td></tr><tr><td align="left" valign="bottom">3<hr/></td><td align="left" valign="bottom">CHI3L2<hr/></td><td align="left" valign="bottom">chitinase-3-like protein 2 isoform c [Homo sapiens]<hr/></td><td align="left" valign="bottom">2<hr/></td></tr><tr><td align="left">4</td><td align="left">CHI3L1</td><td align="left">chitinase-3-like protein 1 precursor [Homo sapiens]</td><td align="left">1.7</td></tr></tbody></table></table-wrap></sec><sec><title>Validation of LC-MS/MS data by ELISA</title><p>Observations obtained by LC-MS/MS were confirmed by ELISA. The basal values of CHIT-1 in the control CSF ranged as wide as 80 – 1250 pg/ml, with a mean of 982 ± 245 (n = 11). In the ALS-CSF, the level of CHIT-1 ranged between 5000 – 54,000 pg/ml, with a mean of 17570 ± 4883 pg/ml (n = 16). The mean increase in the ALS-CSF was approximately 17 fold (p < 0.01 vs N-CSF; Figure <xref ref-type="fig" rid="F2">2</xref>A). Amongst the tested CSF samples of ALS patients, 27% showed a 4-fold increase, whereas 73% showed an increase of 10-fold or more.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Validation of upregulated proteins by ELISA and CHIT-1 activity assay.</bold> Chitotriosidase was found to be 17 fold upregulated (p < 0.01 vs. N-CSF; <bold>A</bold>) and chitinase 3- like-2 (CHI3L2) showed ~2 fold increase (p < 0.01 vs. N-CSF; <bold>B</bold>) in ALS-CSF. Osteopontin showed ~ 1.9 fold upregulation (p < 0.05 vs. N-CSF; <bold>C</bold>) whereas, upregulation in Chitinase 3- like-1 (CHI3L1) levels was statistically non- significant (<bold>D</bold>). Histogram showing CHIT-1 enzymatic activity in CSF samples. The activity was found to be 10 fold higher in ALS-CSF (p < 0.001 vs. N-CSF; <bold>E</bold>).</p></caption><graphic xlink:href="1559-0275-10-19-2"/></fig><p>Similar to the LC-MS/MS data, CHI3L2 levels showed approximately 2-fold increase (p < 0.01 v/s N-CSF; no. of samples: N-CSF =13, ALS-CSF = 16) while osteopontin levels showed 1.9-fold increase in ALS-CSF compared to the control CSF (p < 0.05 v/s N-CSF; no. of samples: N-CSF =13, ALS-CSF = 16) (Figure <xref ref-type="fig" rid="F2">2</xref>B & C). Although the CHI3L1 level was also increased, the change was not statistically significant (Figure <xref ref-type="fig" rid="F2">2</xref>D).</p></sec><sec><title>Validation of LC-MS/MS data by enzyme assay</title><p>The CHIT-1 in the CSF samples catalyzed the conversion of 4-methylumbelliferyl-β - d N, N’, N” –triacetylchitotriose to 4-methylumbelliferone confirming that the enzyme was biologically active. The CHIT-1 activity in the control CSF ranged between 0.0169 – 0.1856 μmol/min/μl (Mean: 0.02459 ± 0.01499; n = 13) whereas in the ALS-CSF it was 0.0809 – 4.1658 μmol/min/μl (Mean: 0.9932 ± 0.3023; n = 16). Thus the patient CSF samples revealed approximately a 10-fold higher enzymatic activity (*p < 0.001 vs N-CSF; Figure <xref ref-type="fig" rid="F2">2</xref>E).</p></sec><sec><title>CHIT-1 expression in microglia</title><p>The Iba-1 immunoreactive pure microglial cultures when exposed to ALS-CSF, showed an elevated expression of CHIT-1 compared to control cultures (Figure <xref ref-type="fig" rid="F3">3</xref>).</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Expression of CHIT-1 in microglial cultures.</bold> Immunoflourescence photomicrographs of pure microglial cultures labeled with CHIT-1 <bold>(green; A, D)</bold> and Iba-1, a marker for microglia <bold>(red; B, E)</bold>. Note the increased expression of CHIT-1 in the cultures exposed to ALS-CSF <bold>(D)</bold> as compared to the normal control <bold>(A)</bold>. The merged images <bold>(C and F)</bold> depict the co-labeling of CHIT-1 with Iba-1 in normal control <bold>(A)</bold> and ALS group <bold>(D)</bold> respectively.</p></caption><graphic xlink:href="1559-0275-10-19-3"/></fig></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>This is the first report demonstrating an increase in the levels of four proteins namely, CHIT-1, osteopontin, CHI3L2 and CHI3L1 in the CSF of ALS patients using the novel and precise quantitative proteomics and ELISA. Similar patterns were observed with both LC-MS/MS of pooled CSF and ELISA of individual CSF. Amongst these the increase in CHIT-1 levels being most dramatic; further experiments were focused on this protein. A profound increase in CHIT-1 enzymatic activity was confirmed in ALS-CSF compared to normal CSF. Accordingly, it is likely to be an important biomarker which may help track the progression of the disease. CHIT-1 could have a role in causing toxicity propagated by the CSF, since the ALS-CSF samples used in the present study also contained neurodegenerative attributes as confirmed by MTT and LDH assays.</p><p>CHIT-1 is classically associated with lysosomal storage diseases like Gaucher’s disease, due to its increased levels in the patients’ CSF [<xref ref-type="bibr" rid="B18">18</xref>]. Chitin is the natural substrate of CHIT-1, but is thought to be absent in the human brain. However, it is intriguing to know that even in the absence of the substrate the enzyme is synthesized by microglia or infiltrating macrophages [<xref ref-type="bibr" rid="B19">19</xref>].</p><p>It is uncertain whether CHIT-1 activity directly affects the CNS or symbolizes an archaic macrophage response against chitin-containing pathogens [<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>]. Chitin is an insoluble N-acetylglucosamine polymer found in invertebrates and human parasites [<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B22">22</xref>]. In its absence, alternative substances e.g. chitin-like glucosamine polymers may be its likely substrate. In post mortem brains of AD patients, presence of glucosamine polymers in amyloid plaques hints at its role in disease pathogenesis [<xref ref-type="bibr" rid="B16">16</xref>]. It is also suggested that chitin-like polysaccharides provide scaffolding for β-amyloid deposition thus facilitating the disease process [<xref ref-type="bibr" rid="B23">23</xref>]. Higher CHIT-1 expression in post mortem brains of AD patients [<xref ref-type="bibr" rid="B24">24</xref>] underlines the co-activation of macrophages and microglia in response to deposition of β-amyloid [<xref ref-type="bibr" rid="B25">25</xref>].</p><p>It is reported that higher CSF CHIT-1 levels in MS patients correlate well with the disease progression [<xref ref-type="bibr" rid="B26">26</xref>]. In the MS patients, the increase in CSF CHIT-1 levels was approximately two-fold, however we observed approximately 10-fold increase in CHIT-1 levels in the patient CSF. Although it is not justified to compare the differences in the patients’ CSF as their study was on MS whereas we studied ALS patients, it is also likely that the differences in fold levels of CSF CHIT-1 are indeed high in ALS compared to MS.</p><p>Contrary to AD, deposits of chitin-like substances were absent in MS. Although microglia and infiltrating macrophages driven innate immune response was a classical molecular feature of MS, the products of the latter processes viz. cytokines, ROS etc. were not substantial enough to establish any clinico-pathological co-relation [<xref ref-type="bibr" rid="B17">17</xref>]. However, Correale and Fiol reported that the enhanced level of CSF chitinases driven by IL-13 could contribute to neuroinflammation by increasing immune cell migration across the blood–brain barrier in the CNS [<xref ref-type="bibr" rid="B26">26</xref>].</p><p>The enhanced expression of CHIT-1, by microglia, possibly indicates a neuroinflammatory response. CHIT-1 is an index of the severity of inflammation alongside the release of pro-inflammatory cytokines like IL-16 and IL-18 [<xref ref-type="bibr" rid="B27">27</xref>]. In stroke, CHIT-1, TNF-α and other pro-inflammatory cytokines are accepted as markers of microglial activation, occurring independent of pre-existing inflammatory or infectious conditions in patients [<xref ref-type="bibr" rid="B28">28</xref>]. According to an alternative hypothesis, CHIT-1 is speculated to be neuroprotective, in view of the reduction in glucosamine aggregates following intrathecal CHIT-1 administration in MS [<xref ref-type="bibr" rid="B17">17</xref>]. Even in AD, enhanced levels of CHIT-1 activity in plasma were considered as the response of the activated microglia-macrophage complex to clear the pathogenic chitin-like substances [<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B25">25</xref>]. Its role in ALS is yet to be deciphered.</p></sec><sec sec-type="conclusions"><title>Conclusion</title><p>Although studies report higher CSF CHIT-1 levels in several neurological diseases including MS, AD and stroke, no study till date documents its elevated levels in ALS, where maximum increase was observed compared to other neurodegenerative diseases. The mechanisms of CHIT-1 induction in each of the neurological disorders appear to be unique to the disease. Collectively, our findings of steep increase in CSF CHIT-1 levels in SALS along with stable bioactivity render it a biomarker status and may find applications in developing therapeutic strategies for sporadic ALS.</p></sec><sec sec-type="materials\|methods"><title>Materials and methods</title><sec><title>CSF sample collection</title><p>ALS-CSF samples from patients with a mean age of 47.38 ± 5.38 years and disease duration of 0.5 to 2.5 years were obtained. N-CSF samples were drawn from age-matched patients undergoing spinal anaesthesia for orthopaedic surgery but without any clinical history of neurological deficits (mean age 45.7 ± 7.04 years) (Tables <xref ref-type="table" rid="T2">2</xref> and Table <xref ref-type="table" rid="T3">3</xref>). CSF samples were snap frozen in liquid nitrogen and stored at –80°C. Informed consent for CSF sample collection was obtained as per the institutional human ethics committee guidelines.</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Details of ALS-CSF</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"> </th><th align="left"><bold>CSF for iTRAQ study</bold></th><th colspan="3" align="left"><bold>CSF for validation</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Gender<hr/></td><td align="left" valign="bottom">Male – 7 (70%)<hr/></td><td align="left" valign="bottom">Female – 3 (30%)<hr/></td><td align="left" valign="bottom">Male – 10 (62.5%)<hr/></td><td align="left" valign="bottom">Female – 6 (37.5%)<hr/></td></tr><tr><td align="left" valign="bottom">Age at presentation (Mean ± SD)<hr/></td><td align="left" valign="bottom">47.40 ± 4.95 (38 – 54) Years<hr/></td><td colspan="3" align="left" valign="bottom">47.38 ± 5.38 (38 – 54) Years<hr/></td></tr><tr><td align="left" valign="bottom">Age at onset<hr/></td><td align="left" valign="bottom">46 ± 5.05 (37 – 53) Years<hr/></td><td colspan="3" align="left" valign="bottom">46.28 ± 5.36 (37 – 53) Years<hr/></td></tr><tr><td align="left" valign="bottom">Duration of illness (Mean ± SD)<hr/></td><td align="left" valign="bottom">15.9 ± 13.4 (4.0 – 48) months<hr/></td><td colspan="3" align="left" valign="bottom">14.19 ± 10.59 (4.0 – 48) months<hr/></td></tr><tr><td align="left" valign="bottom">Onset Patter: Bulbar<hr/></td><td align="left" valign="bottom">1 (10%)<hr/></td><td colspan="3" align="left" valign="bottom">5 (31.25%)<hr/></td></tr><tr><td align="left" valign="bottom">Limb onset<hr/></td><td align="left" valign="bottom">9 (90%)<hr/></td><td colspan="3" align="left" valign="bottom">11 (68.75%)<hr/></td></tr><tr><td align="left" valign="bottom">Upper Limbs<hr/></td><td align="left" valign="bottom">9<hr/></td><td colspan="3" align="left" valign="bottom">11<hr/></td></tr><tr><td align="left" valign="bottom">Lower Limbs<hr/></td><td align="left" valign="bottom">6<hr/></td><td colspan="3" align="left" valign="bottom">7<hr/></td></tr><tr><td align="left" valign="bottom">Speech affected<hr/></td><td align="left" valign="bottom">80% (Mild – 30%, Moderate – 30%, Severe – 20%)<hr/></td><td colspan="3" align="left" valign="bottom">87.5% (Mild – 31.25%, Moderate – 31.25%, Severe – 25%)<hr/></td></tr><tr><td align="left" valign="bottom">Dysphagia<hr/></td><td align="left" valign="bottom">80% (Mild – 30%, Moderate – 30%, Severe – 20%)<hr/></td><td colspan="3" align="left" valign="bottom">93.75% (Mild – 31.25%, Moderate – 18.75%, Severe -12.5%)<hr/></td></tr><tr><td align="left">Spasticity</td><td align="left">9 cases, with evidence of pyramidal signs in the form of spasticity and exaggerated Deep Tendon Reflexes</td><td colspan="3" align="left">11 cases, with evidence of pyramidal signs in the form of spasticity and exaggerated Deep Tendon Reflexes</td></tr></tbody></table></table-wrap><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>Details of N-CSF</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"> </th><th colspan="2" align="left"><bold>CSF for iTRAQ study</bold></th><th colspan="2" align="left"><bold>CSF for validation</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Gender<hr/></td><td align="left" valign="bottom">Male – 8 (80%)<hr/></td><td align="left" valign="bottom">Female – 2 (20%)<hr/></td><td align="left" valign="bottom">Male – 11 (84.6%)<hr/></td><td align="left" valign="bottom">Female – 2 (15.4%)<hr/></td></tr><tr><td align="left" valign="bottom">Age (Mean ± SD in years)<hr/></td><td colspan="2" align="left" valign="bottom">47.3 ± 6.99 (39 – 60)<hr/></td><td colspan="2" align="left" valign="bottom">45.7 ± 7.04 (39 – 60)<hr/></td></tr><tr><td colspan="5" align="left">Patients undergoing orthopaedic surgery</td></tr></tbody></table></table-wrap></sec><sec><title>Cell culture</title><p>We followed a modified protocol to establish pure microglial cultures from P0 Wistar rat pups [<xref ref-type="bibr" rid="B29">29</xref>]. Briefly, the spinal cords were dissected, freed of meninges and mechanically triturated in Dulbecco’s Modified Eagle Medium (DMEM) and propagated in DMEM with 10% FBS (GIBCO-BRL). The mixed glial cultures were allowed to attain 100% confluence. On 10th day <italic>in-vitro</italic> (DIV), the cultures were placed on incubated orbital shaker at 200 rpm for 3–4 hrs. The supernatant containing microglia was centrifuged at 1500 rpm for 5 min, and the pelleted cells were seeded onto poly-l-lysine coated coverslips at the density of 4.4×104 cells/ml. The cultures were maintained in DMEM with 10% FBS (GIBCO-BRL). The cells were then exposed to ALS- CSF or allowed to propagate under normal conditions.</p><p>NSC-34 cell line (Cedarlane Corporation, Canada) routinely maintained in DMEM with 10% FBS [<xref ref-type="bibr" rid="B30">30</xref>] was used to analyze the neurotoxic effects of individual ALS-CSF samples.</p></sec><sec><title>Cell viability/death assays</title><p>NSC-34 cells seeded into 96 well plates (500 cells/well) were exposed to 10% v/v N-CSF or ALS-CSF on 5th DIV for 48 hrs, and then subjected to MTT assay (3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) and LDH assay [<xref ref-type="bibr" rid="B30">30</xref>].</p></sec><sec><title>Mass spectrometry</title><sec><title><bold> <italic>Depletion of abundant proteins in CSF</italic> </bold></title><p>CSF samples from controls (n = 10) and ALS patients (n = 10) were pooled individually and used for iTRAQ based quantitative proteomic comparison by mass spectrometry (Figure <xref ref-type="fig" rid="F4">4</xref>). Samples were centrifuged at 10,000 rpm for 10 min to remove cell debris. Agilent’s multiple affinity removal system (MARS-14), generally used for depletion of albumin, IgG, transferrin, heptoglobin, IgM, IgA, fibrinogen, alpha antitrypsin, apolipoprotein A1, alpha1 acid glycoprotein, alpha2 macroglobin, transthyretin, complement C3 and apolipoproteins was used to deplete the abundant proteins from CSF samples [<xref ref-type="bibr" rid="B31">31</xref>]. Briefly, MARS-14 cartridge was conditioned using 4 ml of Buffer-A (50 g, 1 min). CSF (300 μl) was loaded on to the cartridge and spun at 50 g for 1 min. Flow through was collected and reloaded on to the cartridge, incubated for 5 minutes at room temperature (RT) and spun at 50 g for 1 min. Flow through was re-collected and the bound proteins were eluted using 2.4 ml of Buffer-B. The cartridge was conditioned again using buffer-A and the process was repeated until the abundant proteins were depleted. The fractions were concentrated using Millipore filter (3 kDa). This step also ensures removal of salts contributed by the buffers used during depletion steps.</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>A schematic representation of the procedural steps for mass spectrometric analysis.</bold> Pooled CSF samples were subjected to mass spectrometry after sequential treatment procedures like, depletion of abundant proteins, tryptic digestion, iTRAQ labelling and SCX. Selected molecules were validated using ELISA.</p></caption><graphic xlink:href="1559-0275-10-19-4"/></fig><p>Protein estimation was carried out using Lowry’s method. 200 μg protein each from controls and ALS was used for proteomic comparison. Sample normalization was based on total protein amount which was further verified by SDS-PAGE.</p></sec><sec><title><bold> <italic>iTRAQ labelling and strong cation exchange chromatography</italic> </bold></title><p>Individual depleted extract (125 μg) was treated with 2% SDS and reducing agent (2 μl; 60°C for 1 hr) and alkylated with 1 μl of cysteine blocking agent for 10 min at RT. Proteins were digested overnight at 37°C using sequencing grade trypsin at 1:20 (w/w) ratio. Thereafter, peptides from N-CSF and ALS-CSF were labeled using iTRAQ reagents yielding reporter ions of m/z114 and 115 respectively, for 2 hrs at RT. The reaction was quenched by adding 100 μl of water and the peptides from both N-CSF and ALS-CSF were pooled and fractionated using strong cation exchange (SCX) chromatography. The labeled samples were distributed into 19 fractions using SCX. The fractions were desalted using C18 zip tips, dried and reconstituted in 10 μl of 0.1% trifluoroacetic acid before mass spectrometry analysis [<xref ref-type="bibr" rid="B32">32</xref>].</p></sec><sec><title><bold> <italic>LC-MS/MS analysis</italic> </bold></title><p>LC-MS/MS analysis of iTRAQ-labeled peptides was carried out on an LTQ-Orbitrap Velos mass spectrometer (Thermo Electron, Bremen, Germany) interfaced with Agilent’s 1200 series nanoflow liquid chromatography system (Agilent Technologies, Santa Clara, CA). Each sample was loaded on to the enrichment column (75 μm × 2 cm, 5 μm, 120 Å, Magic C18 AQ MichromBioresources) at a flow rate of 3 μl/min and then resolved on an analytical column (75 μm × 10 cm, 5 μm, 120 Å, Magic C18 AQ MichromBioresources) at a flow rate of 300 nl/min using a linear gradient of 10% - 40% solvent B (90% acetonitrile in 0.1% TFA) over a period of 70 min. The total run time per sample was 115 min. The resolved peptides from analytical column were delivered to LTQ Orbitrap Velos mass spectrometer through an emitter tip (8 μm, New Objective, Woburn, MA). LC-MS/MS data was acquired in a data dependent manner in FT- FT mode. MS spectra were acquired with a mass range of m/z 350 to 1800. Twenty most abundant precursor ions were selected for fragmentation from each MS scan. Data was acquired at MS resolution of 60,000 (m/z 400) and MS/MS resolution of 15,000. Precursor ion fragmentation was carried out using higher energy collision (HCD) mode with normalized collision energy of 41%. Monoisotopic precursor selection was enabled and the precursor ions that were selected for fragmentation were dynamically excluded for 30 sec [<xref ref-type="bibr" rid="B32">32</xref>].</p></sec><sec><title><bold> <italic>Data analysis</italic> </bold></title><p>The MS data was analyzed using the Proteome Discoverer software (Thermo Scientific, version 1.2). The data was searched against human protein database (NCBI RefSeq 45) along with known contaminants using SEQUEST and MASCOT search algorithms. The parameters used for data analysis included trypsin as a protease (allowed one missed cleavage), iTRAQ labeling at N-terminus and lysine residues, and cysteine modification by methyl methanethiosulfonate (MMTS) as fixed modifications and oxidation of methionine as a variable modification. The precursor ion mass error tolerance was set to 20 ppm and product ion mass error tolerance was set to 0.1 Da. The peptide data was extracted using 1% FDR as a threshold. Relative abundance of proteins between N-CSF and ALS-CSF was determined by Proteome Discoverer based on difference in the peak intensity of reporter ions in the MS/MS spectra of each peptide that was ultimately used for quantifying the corresponding protein.</p></sec><sec><title><bold> <italic>Enzyme linked immunosorbent assays (ELISA)</italic> </bold></title><p>N-CSF (n = 13) and ALS-CSF samples (n = 16) were subjected to ELISA based analysis using commercially available kits for CHIT-1 (MBL, Italy; CSF dilution factor (cdf): N-CSF: undiluted; ALS-CSF 1:20), CHI3L1 (Quidel, USA), CHI3L2 (USCN, China; cdf 1:1) and Osteopontin (R&D Systems; cdf 1:50). The protocol was followed as per the manufacturer’s instructions.</p></sec><sec><title><bold> <italic>CHIT-1 activity assay</italic> </bold></title><p>4-methylumbelliferyl-β - d N, N’, N” –triacetylchitotriose (Sigma-Aldrich USA) was used as a substrate to assay the enzyme activity. CSF (2.5 g protein) was added to 150 μl of 22 μmol solution prepared in 0.5 M citrate-phosphate buffer (pH 5.2). Following incubation for 15 min at 37°C, the reaction was stopped using 100 μl of 0.5 mol Na<sub>2</sub>CO<sub>3</sub> -NaHCO<sub>3</sub> buffer (pH 10.7). The fluorescence was recorded at 365 nm excitation/450 nm emissions (Tecan 2500 flouorimeter, USA) and measured as micromoles of substrate hydrolysed/min/l.</p></sec><sec><title><bold> <italic>Immunocytochemistry</italic> </bold></title><p>Fixed primary microglial cultures were stained with rabbit polyclonal anti-CHIT-1 antibody (1:1000, Santacruz, USA) for 24 hr after blocking with 3% bovine serum albumin (BSA) (Sigma–Aldrich, USA) and detected using FITC-conjugated anti-rabbit secondary antibody (1:200, Sigma–Aldrich, USA). The cultures were further incubated with a goat polyclonal anti Iba-1 antibody (1:400, Abcam, UK) for 24 hr and detected with CY3-conjugated anti-goat antibody (1:200, Sigma–Aldrich, USA). The staining was viewed using confocal microscopy (488 nm and 514 nm for FITC and Cy3, respectively; Leica-TCS-SL, Germany). The emission frequencies were segregated to avoid non-specific overlap of labelling [<xref ref-type="bibr" rid="B11">11</xref>].</p></sec></sec><sec><title>Statistical analysis</title><p>The data was statistically assessed for significance by either Student’s <italic>t</italic> test or one way ANOVA as applicable, followed by Tukey’s post hoc tests.</p></sec></sec><sec><title>Abbreviations</title><p>SALS: Sporadic amyotrophic lateral sclerosis; ALS-CSF: Cerebrospinal fluid from amyotrophic lateral sclerosis patients; N-CSF: CSF from patients undergoing orthopaedic surgery; CHIT-1: Chitotriosidase-1; LC-MS/MS: Liquid chromatography-tandem mass spectrometry; Iba-1: Ionised calcium binding adapter molecule-1; GFAP: Glial fibrillary acidic protein; NSC-34: Spinal cord motor neurons fused with neuroblastoma; iTRAQ: Isobaric tags for relative and absolute quantitation; MARS-14: Agilent’s multiple affinity removal system.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>AMV collected samples, performed the experiments, analysed and wrote the manuscript. AS performed the experiments, analysed and wrote the manuscript. PM carried out microglial experiments, analysed and wrote the manuscript. VK analysed and wrote the manuscript. HHC supervised mass spectrometry, analysed and wrote the manuscript. TNS facilitated obtaining control CSF samples and critically evaluated manuscript. SB designed and supervised the experiments and critically reviewed the manuscript. NA enrolled patients with ALS, performed clinical evaluations and provided ALS-CSF. PAA designed, performed experiments, analyzed and wrote the manuscript. TRR conceptualized the project, obtained funding, supervised the study and critically reviewed the manuscript. All the authors read and approved the final manuscript.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional file 1: Table S1</title><p>List of proteins identified in ALS-CSF. Description of Data: A list of the peptides and their fold changes of the 819 proteins identified in ALS-CSF using SEQUEST and Mascot.</p></caption><media xlink:href="1559-0275-10-19-S1.pdf"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S2"><caption><title>Additional file 2: Table S2</title><p>Up-regulated proteins in ALS-CSF. Description of data: Table showing 31 up-regulated proteins with more than 1.5-fold increase in ALS-CSF compared to normal CSF.</p></caption><media xlink:href="1559-0275-10-19-S2.pdf"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S3"><caption><title>Additional file 3: Table S3</title><p>Proteins down-regulated in ALS-CSF. Description of Data: List of 17 down-regulated proteins which showed a decrease of 0.5 fold or more, in ALS-CSF samples.</p></caption><media xlink:href="1559-0275-10-19-S3.pdf"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
Prevalence of tobacco smoking among school teachers in Botswana	<sec><title>Background</title><p>Tobacco is a leading cause of death worldwide, and nearly 80% of all smokers live in low to middle income countries. Previous research has suggested that smoking rates vary by occupation, with relatively low rates commonly seen among educators. Despite this fact, little is known about the smoking habits of teachers in Botswana. The objective of this study, therefore, was to investigate prevalence and correlates of tobacco use among school teachers in Botswana.</p></sec><sec><title>Results</title><p>The prevalence of smoking among school teachers in Botswana was found to be relatively low. Of the 1732 participants in the study, only 3.2% reported being current smokers, 5.3% were ex-smokers and 91.5% had never smoked. Smoking was more common among male teachers when compared to females, being 10.8% and 0.4%, respectively. Factors such as school level, marital status and body mass index were found to be positively associated with tobacco smoking, whereas age, length of employment and weekly working hours were not.</p></sec><sec><title>Conclusion</title><p>This study suggests that Botswana school teachers have a low prevalence of tobacco smoking. While this result may be attributed to tobacco control measures that have been put in place, there is still need to put in place systems to monitor compliance and programs to help those who want to quit smoking. Such protocols would represent a major step forward in further reducing the prevalence of smoking in the education profession.</p></sec>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Erick</surname><given-names>Patience N</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>patience.erick@uon.edu.au</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Smith</surname><given-names>Derek R</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>Derek.Smith@newcastle.edu.au</email></contrib>	Tobacco Induced Diseases	<sec sec-type="intro"><title>Introduction</title><p>Tobacco use represents one of the most important public health problems worldwide. Tobacco endemic is a leading cause of death, illness and impoverishment, resulting in nearly six million fatalities annually. Over 90% of these deaths are caused directly by tobacco use whilst about 10% are the results of non-smokers being exposed to second-hand smoke [<xref ref-type="bibr" rid="B1">1</xref>]. If current trends are not changed, these figures are expected to increase to more than 8 million deaths per year by 2030 [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>]. Nearly 80% of the more than one billion smokers worldwide, a percentage projected to rise [<xref ref-type="bibr" rid="B3">3</xref>], live in low and middle income countries where the burden of tobacco related illness and death is substantial. Premature deaths which may be caused by tobacco use deprive families of those who died of income, raise the cost of health care and hinder economic development [<xref ref-type="bibr" rid="B1">1</xref>]. Additionally, tobacco smoking is a prevalent risk factor for cardiovascular and respiratory disease such as coronary heart disease, lung cancer and tuberculosis [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B3">3</xref>]. In a study that was conducted in Botswana, it was found that, 66.4% of patients that were diagnosed and treated for cancer in three referral hospitals were associated with tobacco use [<xref ref-type="bibr" rid="B4">4</xref>]. Moreover, tobacco use represents an important issue in occupational health because of its significant impact in the workplace [<xref ref-type="bibr" rid="B2">2</xref>]. Previous research has suggested that smoking rates vary by occupation, with relatively low rates commonly seen among educators [<xref ref-type="bibr" rid="B5">5</xref>-<xref ref-type="bibr" rid="B7">7</xref>], although little is known about the smoking habits of teachers in African countries.</p><p>Teachers have an important responsibility in tobacco control given that they are highly respected in their communities as they influence the evolution for each aspect of life [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B9">9</xref>]. It has been recognised that teachers are important role models for students, conveyors of tobacco prevention curricula and key opinion leaders for school tobacco control policies [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. In addition, teachers have daily interaction with students and thus represent an influential group in tobacco smoking control. However, this potential can be limited if teachers use tobacco especially in the presence of students in school premises [<xref ref-type="bibr" rid="B10">10</xref>]. The results of a study carried out in Nairobi, Kenya to determine the prevalence and risk factors of smoking among secondary school students indicated that, smoking among students started very early in their life due to the smoking habits of their parents at home and teachers at school [<xref ref-type="bibr" rid="B11">11</xref>]. Similar results were found in the study conducted to assess the influence of smoking and tombak (local smokeless tobacco) dipping by parents, teachers and friends on cigarette smoking and tombak dipping by school going Sudanese adolescents [<xref ref-type="bibr" rid="B12">12</xref>].</p><p>Despite the important role of teachers on tobacco smoking control, few studies have been conducted to investigate tobacco smoking behaviours of school teachers. As far as the authors of this study could ascertain, no study on tobacco smoking has been conducted among teachers in Botswana. The aim of this study was, therefore, to investigate and report on the prevalence of tobacco smoking among teachers in Botswana.</p></sec><sec sec-type="methods"><title>Methods</title><p>As part of a larger descriptive cross sectional study of occupational health issues, 3 100 school teachers in Botswana were surveyed. The study was approved by the University of Newcastle Human Research Ethics Committee and Botswana Ministry of Education and Skills Development. From seven education regions, 107 primary and 57secondary schools were randomly selected. All school teachers in those schools were invited to take part in the study. Permission to conduct the research in the selected schools was sought from school heads. Informed consent of teachers was implied by completing and returning the questionnaire. Data was collected from August to December 2012 by means of an anonymous, self-reporting questionnaire. Tobacco smoking variables were constructed to estimate cigarette smoking prevalence, and proportions of ex-smokers and those who have never smoked. Data was also collected on the number of cigarettes smoked daily and number of years since quitting to smoke. SPSS 20.0 was used to analyse the collected data. Pearson’s chi-square tests were used to determine statistical associations with smoking.</p></sec><sec sec-type="results"><title>Results</title><p>An overall response rate of 56.3% was obtained in this study. Out of the total respondents 1260 (72.7%) were females, 832 (66.0%) of which were primary school teachers with mean age of 39.34 ± 9.02 years and working experience of 13.36 ± 8.82 years. The results of this study show that 3.2% of school teachers in Botswana reported that they were current smokers, while 5.3% were ex-smokers and 91.5% have never smoked. The results of the current study, as indicated in Table <xref ref-type="table" rid="T1">1</xref>, reveal that gender was significantly associated with smoking among school teachers. The prevalence of smoking among female teachers (0.4%) was substantial lower than of their male counterpart (10.8%), p < 0.001. Marital status was significantly associated with tobacco smoking (p = 0.001). School level has also been positively associated with tobacco smoking among teachers. Majority of smokers were 30 years or less. Age and length of employment were not significantly associated with tobacco smoking.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Prevalence of tobacco smoking among teachers in Botswana</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left" valign="bottom"> <hr/></th><th align="left" valign="bottom"><bold>Current smoker</bold><hr/></th><th align="left" valign="bottom"><bold>Ex-smoker</bold><hr/></th><th align="left" valign="bottom"><bold>Never smoked</bold><hr/></th><th align="left" valign="bottom"><bold>p-value</bold><hr/></th></tr><tr><th align="left"> </th><th align="left"><bold>n (%)</bold></th><th align="left"><bold>n (%)</bold></th><th align="left"><bold>n (%)</bold></th><th align="left"> </th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Total<hr/></td><td align="left" valign="bottom">56 (3.2)<hr/></td><td align="left" valign="bottom">91 (5.3)<hr/></td><td align="left" valign="bottom">1585 (91.5)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"><bold>Gender</bold><hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"><bold><0.001</bold><hr/></td></tr><tr><td align="left" valign="bottom">Male<hr/></td><td align="left" valign="bottom">51 (10.8)<hr/></td><td align="left" valign="bottom">64 (13.6)<hr/></td><td align="left" valign="bottom">357 (74.6)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Female<hr/></td><td align="left" valign="bottom">5 (0.4)<hr/></td><td align="left" valign="bottom">27 (2.1)<hr/></td><td align="left" valign="bottom">1228 (97.5)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"><bold>Age range (years)</bold><hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">0.172<hr/></td></tr><tr><td align="left" valign="bottom">≤30<hr/></td><td align="left" valign="bottom">19 (5.3)<hr/></td><td align="left" valign="bottom">19 (5.3)<hr/></td><td align="left" valign="bottom">319 (89.4)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">31-40<hr/></td><td align="left" valign="bottom">22 (3.4)<hr/></td><td align="left" valign="bottom">35 (5.4)<hr/></td><td align="left" valign="bottom">591 (91.2)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">41-50<hr/></td><td align="left" valign="bottom">12 (2.3)<hr/></td><td align="left" valign="bottom">28 (5.3)<hr/></td><td align="left" valign="bottom">485 (92.4)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">>50<hr/></td><td align="left" valign="bottom">2 (3.2)<hr/></td><td align="left" valign="bottom">7 (4.3)<hr/></td><td align="left" valign="bottom">154 (94.5)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"><bold>School level</bold><hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"><bold>0.002</bold><hr/></td></tr><tr><td align="left" valign="bottom">Primary school<hr/></td><td align="left" valign="bottom">26 (2.6)<hr/></td><td align="left" valign="bottom">40 (4.0)<hr/></td><td align="left" valign="bottom">937 (93.4)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Junior secondary<hr/></td><td align="left" valign="bottom">27 (4.8)<hr/></td><td align="left" valign="bottom">36 (6.4)<hr/></td><td align="left" valign="bottom">493 (88.7)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Senior secondary<hr/></td><td align="left" valign="bottom">3 (1.8)<hr/></td><td align="left" valign="bottom">15 (8.8)<hr/></td><td align="left" valign="bottom">152 (89.4)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"><bold>Marital status</bold><hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"><bold>0.001</bold><hr/></td></tr><tr><td align="left" valign="bottom">Single<hr/></td><td align="left" valign="bottom">42 (4.6)<hr/></td><td align="left" valign="bottom">49 (5.3)<hr/></td><td align="left" valign="bottom">827 (90.1)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Married<hr/></td><td align="left" valign="bottom">8 (1.1)<hr/></td><td align="left" valign="bottom">39 (5.5)<hr/></td><td align="left" valign="bottom">666 (93.4)<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left">Separated/divorced/widowed</td><td align="left">6 (5.9)</td><td align="left">3 (3.0)</td><td align="left">92 (91.1)</td><td align="left"> </td></tr></tbody></table></table-wrap></sec><sec sec-type="discussion"><title>Discussion</title><p>About 3.2% of teachers in this study reported that they were smokers. This prevalence is relatively lower compared to results of other studies that have been carried out around the world. Supporting this are the results of studies from Kingdom of Bahrain and Kenya in which prevalence of smoking among Bahraini and Kenyan teachers were 7% for each [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B14">14</xref>]. As shown on Table <xref ref-type="table" rid="T2">2</xref>, quite similar findings were found in studies conducted among Malay and Yemen teachers where 7.8% and 8% prevalence were reported, respectively [<xref ref-type="bibr" rid="B9">9</xref>]. Similarly high prevalence of tobacco smoking has been reported among school teachers around the world. A study of school teachers in India, for example, found that 14.5% of primary school teachers where smokers [<xref ref-type="bibr" rid="B15">15</xref>] while in Bangladesh prevalence of tobacco smoking among secondary school teachers was 17% [<xref ref-type="bibr" rid="B16">16</xref>] and 17.8% in Sousse, Tunisia [<xref ref-type="bibr" rid="B17">17</xref>]. Furthermore, another study of Malay secondary school teachers in Kelantan found that 20% are smokers [<xref ref-type="bibr" rid="B18">18</xref>]. A much higher prevalence was reported in a study from Tunisian Sahel which found that 29.3% of school teachers smoked [<xref ref-type="bibr" rid="B19">19</xref>] and 29.7% of primary and secondary Spanish teachers were smokers [<xref ref-type="bibr" rid="B20">20</xref>]. The highest smoking prevalence (58.1%) has been reported by Turkish primary teachers. In the same study, 36.1% teachers reported that they were ex-smokers whilst 5.8% had never smoked [<xref ref-type="bibr" rid="B21">21</xref>]. A similar smoking prevalence (52.1%) was reported among Syrian male primary and secondary school teachers [<xref ref-type="bibr" rid="B22">22</xref>], whilst in Malaysia, 40.6% secondary school teachers were smokers [<xref ref-type="bibr" rid="B23">23</xref>].</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Prevalence of smoking among school teachers reported from international studies</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Country</bold></th><th align="left"><bold>Participants</bold></th><th align="left"><bold>Number of participants</bold></th><th align="left"><bold>Response rate (%)</bold><sup> <bold>a</bold> </sup></th><th align="left"><bold>Smokers (%)</bold></th><th align="left"><bold>Ex-smokers (%)</bold></th><th align="left"><bold>Never smoked (%)</bold></th><th align="left"><bold>Year</bold><sup> <bold>b</bold> </sup></th><th align="left"><bold>Authors</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">India<hr/></td><td align="left" valign="bottom">Primary school teachers<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">400<hr/></td><td align="left" valign="bottom">14.5<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">2013<hr/></td><td align="left" valign="bottom">Savadi <italic>et al.</italic>[<xref ref-type="bibr" rid="B15">15</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Malaysia<hr/></td><td align="left" valign="bottom">Secondary school teachers<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">495<hr/></td><td align="left" valign="bottom">7.8<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">2012<hr/></td><td align="left" valign="bottom">Al-Naggar <italic>et al.</italic>[<xref ref-type="bibr" rid="B9">9</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Tunisia<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">800<hr/></td><td align="left" valign="bottom">92.4<hr/></td><td align="left" valign="bottom">17.8<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">2011<hr/></td><td align="left" valign="bottom">Harrabi <italic>et al.</italic>[<xref ref-type="bibr" rid="B17">17</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Bangladesh<hr/></td><td align="left" valign="bottom">Secondary school teachers<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">559<hr/></td><td align="left" valign="bottom">17<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">2011<hr/></td><td align="left" valign="bottom">Rahman <italic>et al.</italic>[<xref ref-type="bibr" rid="B16">16</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Turkey<hr/></td><td align="left" valign="bottom">Primary school teachers<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">468<hr/></td><td align="left" valign="bottom">58.1<hr/></td><td align="left" valign="bottom">36.1<hr/></td><td align="left" valign="bottom">5.8<hr/></td><td align="left" valign="bottom">2008<hr/></td><td align="left" valign="bottom">Unsal <italic>et al.</italic>[<xref ref-type="bibr" rid="B21">21</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Tunisia<hr/></td><td align="left" valign="bottom">Primary and secondary school teachers<hr/></td><td align="left" valign="bottom">402<hr/></td><td align="left" valign="bottom">89.1<hr/></td><td align="left" valign="bottom">29.3<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">2006<hr/></td><td align="left" valign="bottom">Abdelaziz <italic>et al.</italic>[<xref ref-type="bibr" rid="B19">19</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Yemen<hr/></td><td align="left" valign="bottom">Secondary school teachers<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">317<hr/></td><td align="left" valign="bottom">8<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">2006<hr/></td><td align="left" valign="bottom">Bin Ghouth <italic>et al.</italic>[<xref ref-type="bibr" rid="B8">8</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Kenya<hr/></td><td align="left" valign="bottom">Primary school teachers<hr/></td><td align="left" valign="bottom">910<hr/></td><td align="left" valign="bottom">87.9<hr/></td><td align="left" valign="bottom">7<hr/></td><td align="left" valign="bottom">8.4<hr/></td><td align="left" valign="bottom">84.6<hr/></td><td align="left" valign="bottom">2001<hr/></td><td align="left" valign="bottom">Kwamanga <italic>et al.</italic>[<xref ref-type="bibr" rid="B11">11</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Malaysia<hr/></td><td align="left" valign="bottom">Secondary school teachers<hr/></td><td align="left" valign="bottom">180<hr/></td><td align="left" valign="bottom">180<hr/></td><td align="left" valign="bottom">40.6<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">2001<hr/></td><td align="left" valign="bottom">Naing & Ahmad [<xref ref-type="bibr" rid="B23">23</xref>]<hr/></td></tr><tr><td rowspan="2" align="left" valign="top">Japan<hr/></td><td rowspan="2" align="left" valign="top">Kindergarten, elementary and secondary school teachers<hr/></td><td rowspan="2" align="left" valign="top">16000<hr/></td><td rowspan="2" align="left" valign="top">87.5<hr/></td><td align="left" valign="bottom">44.7 Male<hr/></td><td rowspan="2" align="left" valign="bottom"> <hr/></td><td rowspan="2" align="left" valign="bottom"> <hr/></td><td rowspan="2" align="left" valign="top">2000<hr/></td><td rowspan="2" align="left" valign="top">Ohida <italic>et al.</italic>[<xref ref-type="bibr" rid="B26">26</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">3.1 Female<hr/></td></tr><tr><td align="left" valign="bottom">Spain<hr/></td><td align="left" valign="bottom">Primary and secondary school teachers<hr/></td><td align="left" valign="bottom">8000<hr/></td><td align="left" valign="bottom">38.1<hr/></td><td align="left" valign="bottom">29.7<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">2000<hr/></td><td align="left" valign="bottom">Barrueco <italic>et al.</italic>[<xref ref-type="bibr" rid="B20">20</xref>]<hr/></td></tr><tr><td rowspan="2" align="left" valign="top">Syria<hr/></td><td rowspan="2" align="left" valign="top">Primary and secondary school teachers<hr/></td><td rowspan="2" align="left" valign="bottom"> <hr/></td><td rowspan="2" align="left" valign="top">90<hr/></td><td align="left" valign="bottom">52.1 Male<hr/></td><td rowspan="2" align="left" valign="bottom"> <hr/></td><td rowspan="2" align="left" valign="bottom"> <hr/></td><td rowspan="2" align="left" valign="top">2000<hr/></td><td rowspan="2" align="left" valign="top">Maziak <italic>et al.</italic>[<xref ref-type="bibr" rid="B22">22</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">12.3 Female<hr/></td></tr><tr><td align="left" valign="bottom">Malaysia<hr/></td><td align="left" valign="bottom">Secondary school teachers<hr/></td><td align="left" valign="bottom">5112<hr/></td><td align="left" valign="bottom">63<hr/></td><td align="left" valign="bottom">20<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">1994<hr/></td><td align="left" valign="bottom">Bin Yaacob & Bin Harum [<xref ref-type="bibr" rid="B18">18</xref>]<hr/></td></tr><tr><td align="left">Bahrain</td><td align="left">Primary and secondary school teachers</td><td align="left">1284</td><td align="left">89</td><td align="left">7</td><td align="left">3.1</td><td align="left">89.9</td><td align="left"> </td><td align="left">Alnasir [<xref ref-type="bibr" rid="B14">14</xref>]</td></tr></tbody></table><table-wrap-foot><p><sup>a</sup>Response rate of the study (*Total number of respondents listed as the response rate was not provided), <sup>b</sup>Publication year.</p></table-wrap-foot></table-wrap><p>The low smoking prevalence among Botswana teachers can be, perhaps, attributed to a general non acceptance of smoking in the country, generally. The prevalence of any tobacco smoking and cigarette smoking in Botswana as of 2011 was 17% and 13% respectively [<xref ref-type="bibr" rid="B24">24</xref>]. Low prevalence of smoking in Botswana could also be attributed to tobacco control measures that have been put in place in the country. The Government of Botswana long recognised and accepted the need to sensitize its population to the harmful effects of tobacco. Botswana is one of the first African countries to become signatories to the Framework Convention on Tobacco Control (FCTC). Botswana signed FCTC in June 2003 and ratified in 2005. Prior to this development, Botswana had enacted her first tobacco control legislation, the Control of Smoking Act (CSA) in 1992. The main focus of the act is on controlling Environmental Tobacco Smoke in enclosed public and workplace, educational institutions and hospitals as well as to ban tobacco advertising. To date, the country has by far successfully implemented several key aspects of the FCTC guidelines such as smoke free places, a ban on advertising and promotion of tobacco products, and sale to minors. However, the are no systems in place to check compliance [<xref ref-type="bibr" rid="B25">25</xref>].</p><p>The results of this study demonstrated that male teachers had a significantly higher prevalence of tobacco smoking than their female colleagues (10.8% vs 0.4%, p < 0.001). Similar results have been found in other studies conducted in Japan where, only 3.1% and 44.7% of female and male teachers respectively, were smokers [<xref ref-type="bibr" rid="B26">26</xref>], and in Syria where 12.3% of female and 52.1% male teachers were smokers [<xref ref-type="bibr" rid="B22">22</xref>]. In addition, 94% of smoking teachers in Bahrain were male teachers [<xref ref-type="bibr" rid="B14">14</xref>]. Comparably, other studies have also reported that smoking was higher among male than female teachers [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B27">27</xref>]. Interestingly, the results of studies conducted among primary school teachers in Belgaum City, India [<xref ref-type="bibr" rid="B15">15</xref>] and secondary school teachers in Yemen [<xref ref-type="bibr" rid="B8">8</xref>], indicated that female teachers in these studies did not smoke. Low prevalence of smoking among female teachers could be because traditionally it is a taboo for women to smoke. It has been suggested that there are few female smokers than males especially in developing countries which could probably be related to social norm that has been long formed in many societies [<xref ref-type="bibr" rid="B9">9</xref>].</p><p>In this study, cigarette smoking was found to be associated with marital status (p = 0.001). Similar findings were reported by Malay secondary school teachers [<xref ref-type="bibr" rid="B9">9</xref>]. School level (p = 0.002) and body mass index (p = 0.027) were also significantly associated with smoking among school teachers in Botswana. However, age, education level, number of children less than six years, length of employment, working hours and number of students taught were not significantly associated with smoking.</p><p>Smokers in this study indicated that they have been smoking for periods ranging from a year to 31 years with an average smoking duration of 8.62 years, smoking between one to 20 cigarettes a day. The average number of cigarettes smoked was 5.6 per day. The results also show that 5.3% of teachers in the study were ex-smokers having smoked for one to 27 years with average smoking years of 7.83 years.</p><p>Various strengths and limitation were found for this study. Firstly, the study covered seven out of 10 educational regions in Botswana, so the results can be generalised. Secondly, the study helped to find the prevalence of smoking among teachers as they are considered to be students’ role models. A limitation of the study is that the data reflect respondents’ subjective perceptions.</p></sec><sec sec-type="conclusions"><title>Conclusion</title><p>Prevalence of tobacco smoking among Botswana teachers was relatively low. Factors such as gender, school level and body mass index have been associated with smoking. Measures should be put in place to monitor compliance with measures that have been put in place to control tobacco smoking.</p></sec><sec><title>Competing interests</title><p>The authors declare they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>PNE and DRS conceived and designed the study. PNE carried out data collection and analysis. PNE and DRS read and approved the final manuscript.</p></sec>
Random unstimulated pediatric luteinizing hormone levels are not reliable in the assessment of pubertal suppression during histrelin implant therapy	<sec><title>Background</title><p>Gonadotropin-releasing hormone agonist (GnRHa)-stimulated luteinizing hormone (LH) is the standard hormonal assessment for both diagnosis and therapeutic monitoring of children with central precocious puberty (CPP). Use of unstimulated (random) LH levels may be helpful in diagnosis and has gained popularity in monitoring GnRHa therapy despite lack of validation against stimulated values. The objective of this investigation was to assess the suitability of random LH for monitoring pubertal suppression during GnRHa treatment.</p></sec><sec><title>Methods</title><p>Data from a multi-year, multicenter, open-label trial of annual histrelin implants for CPP was used for our analysis. Children meeting clinical and hormonal criteria for CPP, either naïve to GnRHa therapy or previously treated with another GnRHa for at least 6 months who were being treated at academic pediatric centers were included in the study. Subjects received a single 50-mg subcutaneous histrelin implant annually until final explant at an age determined at the discretion of each investigator. Monitoring visits for physical examination and GnRHa-stimulation testing were performed at regular intervals. The main outcome measure was pubertal suppression during treatment defined by peak LH < 4 mIU/mL after GnRHa stimulation.</p></sec><sec><title>Results</title><p>During histrelin treatment, 36 children underwent a total of 308 monitoring GnRHa stimulation tests. Unstimulated and peak LH levels were positively correlated (r = 0.798), and both declined from the first to second year of treatment. Mean ± SD peak LH level during therapy was 0.62 ± 0.43 mIU/mL (range, 0.06–2.3), well below the normal prepubertal mean. Mean random LH was 0.35 ± 0.25 mIU/mL (range, 0.04–1.5), 10-fold higher than the normal prepubertal mean. The random LH levels were above the prepubertal upper threshold (<0.3 mIU/mL) in 48.4% of all tests and in 88.9% of subjects at some point during therapy.</p></sec><sec><title>Conclusions</title><p>In contrast with GnRHa-stimulated LH, unstimulated LH values frequently fail to demonstrate suppression to prepubertal values during GnRHa therapy for CPP, despite otherwise apparent pubertal suppression, and are thus unsuitable for therapeutic monitoring.</p></sec><sec><title>Trial registration</title><p>ClinicalTrial.gov <ext-link ext-link-type="uri" xlink:href="http://www.clinicaltrials.gov/NCT00779103">NCT00779103</ext-link>.</p></sec>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Neely</surname><given-names>E Kirk</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>neely@stanford.edu</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Silverman</surname><given-names>Lawrence A</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>Lawrence.Silverman@atlantichealth.org</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Geffner</surname><given-names>Mitchell E</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>MGeffner@chla.usc.edu</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Danoff</surname><given-names>Theodore M</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>tdanoff@clarustherapeutics.com</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Gould</surname><given-names>Errol</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>egould5@comcast.net</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Thornton</surname><given-names>Paul S</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>Paul.Thornton@cookchildrens.org</email></contrib>	International Journal of Pediatric Endocrinology	<sec><title>Background</title><p>Gonadotropin-releasing hormone agonist (GnRHa) therapy is considered the standard of care for treatment of children with central precocious puberty (CPP) [<xref ref-type="bibr" rid="B1">1</xref>]. Monthly and 3-monthly depot leuprolide [<xref ref-type="bibr" rid="B2">2</xref>-<xref ref-type="bibr" rid="B5">5</xref>] and the annual histrelin implant [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref>] are the most commonly used therapies in the U.S. Serum LH response to GnRH or GnRHa stimulation, typically using aqueous leuprolide acetate as the stimulating agent, is the conventional test used to diagnose CPP [<xref ref-type="bibr" rid="B8">8</xref>-<xref ref-type="bibr" rid="B15">15</xref>] in conjunction with sex steroid levels and characteristic clinical features. Unstimulated, i.e. random, LH levels can also be used for diagnosis of CPP, although with limited sensitivity in early puberty [<xref ref-type="bibr" rid="B16">16</xref>].</p><p>During treatment of CPP, GnRHa stimulation tests are utilized to monitor and confirm continued LH suppression, with peak levels dropping into or below the normal prepubertal range [<xref ref-type="bibr" rid="B2">2</xref>-<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B18">18</xref>]. During depot GnRHa therapy, stimulation testing can be performed easily by drawing a post-injection sample, whereas monitoring during histrelin implant therapy requires an additional injection. Many practitioners alternatively use a random LH level during therapeutic monitoring instead of stimulation testing, despite the lack of data demonstrating its utility. Here we investigate the suitability of random LH levels for assessment of pubertal suppression during histrelin implant therapy using data collected from the pivotal, long-term, U.S. multicenter trial of histrelin for treatment of CPP [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref>].</p></sec><sec sec-type="methods"><title>Methods</title><p>Data were derived from an open-label study of 36 children with CPP [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref>] in which all subjects received a 50-mg subcutaneous histrelin implant (Supprelin LA, Endo Pharmaceuticals, Malvern, PA) in the upper arm annually until final explant at an age determined at the discretion of each investigator. The study was open to girls aged 2–8.99 years and to boys aged 2–9.99 years who had evidence of CPP and who had not previously received GnRHa therapy (“treatment naïve”) and to girls aged 2–10.99 years and boys aged 2–11.99 years who had received treatment with a GnRHa regimen for at least 6 months (“previously treated”). Prior to any treatment, all participants had breast Tanner stage ≥ 2 (girls) or testicular volume ≥ 4 mL (boys), stimulated LH > 7 mIU/mL after GnRH or > 10 mIU/mL after leuprolide acetate, and bone age ≥ +2 standard deviations [SD]. Written informed consent was obtained at each site, as was assent when appropriate.</p><p>Monitoring visits with physical examinations and hormone testing were performed at months 1, 3, 6, 9, 12, 13, 18, 24, 36, and every 6 months thereafter. Blood samples were collected at 0 minutes (baseline) before subcutaneous injection of leuprolide acetate (20 μg/kg) and 30 and 60 minutes post-injection; beginning in the third year of the study, samples were obtained at 0 and 40 minutes. All hormone measurements were performed at Esoterix Clinical Trial Services (East Windsor, NJ). LH and follicle-stimulating hormone (FSH) were measured by immunochemiluminescent (ICMA) assay with a lower limit of quantification of 0.02 mIU/mL [<xref ref-type="bibr" rid="B15">15</xref>]. Mean values are stated with standard deviations. Estradiol was initially measured on baseline samples using radioimmunoassay (RIA) with a lower limit of detection of 5 pg/mL, but, beginning in the third year of the study, assays were performed by liquid chromatography and tandem mass spectrometry (LCMS/MS) with a lower limit of 1 pg/mL. For consistency of data analysis, any estradiol value ≤ 5 pg/mL was imputed as 5 pg/mL. Testosterone measurement was performed on baseline samples using RIA with a lower limit of 3 ng/dL.</p></sec><sec sec-type="results"><title>Results</title><p>Mean age at the onset of GnRHa therapy was 7.1 ± 1.4 years in the treatment-naïve group (n = 20) and 8.9 ±1.5 years in the previously treated group (n = 16). Mean ± SD unstimulated (0 minute) and peak stimulated (30 or 60 minute) LH levels on the day of initial implant were 1.54 ± 1.67 mIU/mL and 28.2 ± 20 mIU/mL, respectively, in the naïve group, and 0.36 ± 0.33 mIU/mL and 2.09 ± 2.15 mIU/mL, respectively, in the pretreated group. Duration of implant therapy ranged from 1 to 5 years. There were no treatment failures or withdrawals for adverse events.</p><p>Baseline pretreatment testing provided valuable adjunct data regarding the possible lack of utility of random LH values in both the diagnosis of CPP and its monitoring. Prior to the initial implant, 7/20 (35%) treatment-naive subjects had prepubertal 0 minute LH levels (< 0.3 mIU/mL) despite peak GnRHa-stimulated LH levels diagnostic of CPP (> 10 mIU/mL), demonstrating the relatively limited sensitivity of a random LH to diagnose CPP. Also, 6/16 (38%) of previously treated patients on depot leuprolide at the time of implant had unstimulated LH levels > 0.3 mIU/mL (range, up to 1.1 mIU/mL) at the baseline visit.</p><p>During histrelin treatment, a total of 308 GnRHa stimulation tests were performed in the 36 children. All subjects in both groups maintained suppression of LH levels for the duration of the study based on the primary outcome measure of peak GnRHa-stimulated LH < 4 mIU/mL. The mean of all peak LH levels obtained during the leuprolide stimulation tests while on therapy was 0.62 ± 0.43 mIU/mL (range, 0.06–2.3 mIU/mL). Levels at 30 and 60 minutes were not significantly different from one another. In females, all estradiol levels drawn at stimulation testing were below the <italic>a priori</italic> threshold of 20 pg/mL, and in boys all testosterone levels were below the <italic>a priori</italic> testosterone threshold of 30 ng/dL.</p><p>The mean of all 0 minute LH levels obtained in leuprolide stimulation tests during therapy was 0.35 ± 0.25 mIU/mL (range, 0.04–1.5). A strong positive correlation was found between the random (0 minute) and peak leuprolide-stimulated LH levels (r = 0.798, Figure <xref ref-type="fig" rid="F1">1</xref>) (random LH values did not correlate with estradiol or peak FSH). Despite the fact that all peak LH levels were suppressed to < 2.5 IU/L, random LH levels remained at or above the pubertal threshold of 0.3 mIU/mL in 149/308 tests (48.4%) and were ≥ 0.7 mIU/mL in 28/308 (9.1%) and ≥ 1.0 mIU/mL in 7/308 (2.3%). The seven values ≥ 1.0 mIU/mL occurred in 7 different children, and only 4/36 children (11.1%) had random LH < 0.3 at all treatment visits (1/20 naïve, 3/16 previously treated).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Scattergram of unstimulated versus peak GnRHa-stimulated LH levels during the histrelin implant therapy.</bold> Data from all follow-up visits (n = 308) during the first year of histrelin implant therapy (open circles) and the second through fifth year of therapy (solid circles) are shown. The horizontal dashed line represents the a priori peak GnRHa-stimulated LH threshold level of 4 mIU/mL defining adequate suppression. The vertical dashed line indicates the published 0.3 mIU/mL pubertal threshold for unstimulated LH levels [<xref ref-type="bibr" rid="B15">15</xref>].</p></caption><graphic xlink:href="1687-9856-2013-20-1"/></fig><p>A temporal decline in both random and peak LH (Figure <xref ref-type="fig" rid="F2">2</xref>) was observed over the course of study. Mean random LH levels in the first and subsequent years were 0.42 ± 0.27 mIU/mL (n = 174) and 0.25 ± 0.19 mIU/mL (n = 141), respectively. The highest random LH value was 1.5 mIU/mL during year one, compared with 1.0 mIU/mL during years 2 to 5. Of the 28 basal LH values > 0.7 IU/L during therapy, 24 occurred in the first year. Mean random LH levels were slightly higher in the naïve group compared with the previously treated group during the first year of implant therapy. Six out of 7 random LH values > 1 mIU/mL occurred in the naïve group in the first year of therapy. After the first year, the mean random LH levels in the 2 groups were not different.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Mean (± SE) unstimulated LH levels (0 minutes) at each visit over 5 years of histrelin therapy.</bold> The horizontal dashed line indicates the 0.3 mIU/mL pubertal threshold for unstimulated LH levels [<xref ref-type="bibr" rid="B15">15</xref>]. Note that the time scale is not proportionate. SE is not shown beginning with the 36-month visit due to small sample size, and values in the fifth year are not shown due to small sample size (3 or less).</p></caption><graphic xlink:href="1687-9856-2013-20-2"/></fig></sec><sec sec-type="discussion"><title>Discussion</title><p>All study subjects on histrelin implant therapy for CPP were clinically and biochemically suppressed using the standard outcome measures of peak LH returning to prepubertal levels, reduced sex steroid levels, cessation of advancement of Tanner stage, diminished growth velocity, and reduction in the rate of bone age advancement [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref>]. In this multi-year study of 36 children treated with histrelin implants (20 naïve to treatment and 16 pretreated with another GnRHa), all peak LH values during implant therapy were < 2.5 mIU/mL, demonstrating that puberty in all subjects was unequivocally suppressed. Nonetheless, the random unstimulated LH value exceeded the 0.3 mIU/mL pubertal threshold for the Esoterix assay at 48.4% of the treatment visits, the value was sporadically > 1 mIU/mL in 7 different children, and the lack of suppression assessed by random LH was nearly universal, occurring in 89% of children, despite every other parameter indicating pubertal suppression.</p><p>In other words, stimulated LH returns to prepubertal norms during histrelin therapy, but random unstimulated LH levels do not, making the usefulness of random LH in therapeutic monitoring somewhat dubious. The physiologic reason for persistence of basal LH in the pubertal range is not understood. GnRH superagonist therapy might result in tonic LH secretion, but the observed decline over years of therapy would need to be explained. Alternatively, consistently low but measurable LH might be related to the circulating alpha subunits during therapy, although modern assays do not cross-react with those subunits.</p><p>Persistence of pubertal random LH values was reported in short-term depot leuprolide therapy nearly 2 decades ago when ultrasensitive LH assays became available [<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B16">16</xref>]. It was again recently noted during short-term histrelin therapy [<xref ref-type="bibr" rid="B19">19</xref>]. That study was performed with a much smaller sample size, non-simultaneous random and stimulated LH testing, and a relatively insensitive LH assay, but the message is effectively the same. Nonetheless, random LH measurement is commonly performed and used as a criterion to confirm suppression. The persistent elevation in random LH and the temporal decline during the first years of therapy have also been observed during long-term depot leuprolide therapy [<xref ref-type="bibr" rid="B2">2</xref>]. It is likely not coincidental that the mean random LH level in the previously treated group in our study was essentially unchanged by subsequent histrelin therapy. Persistent elevation of the random LH level is characteristic of both leuprolide and histrelin therapy, whereas a three-fold reduction in peak LH levels was seen following the change in our pretreated subjects from depot leuprolide to histrelin implant.</p><p>Unstimulated LH values trended lower during successive years of histrelin therapy, as did the GnRHa-stimulated LH levels [<xref ref-type="bibr" rid="B7">7</xref>]. Our data confirm that random LH levels during GnRHa therapy correlate positively with stimulated LH levels. Nevertheless, a considerable percentage of random LH levels remained at or well above the pubertal threshold in the later years of histrelin treatment. The mean random LH of 0.35 mIU/mL during therapy remained approximately 10-fold higher than the prepubertal mean for this LH ICMA assay, 0.03 ± 0.03 mIU/mL [<xref ref-type="bibr" rid="B15">15</xref>], even in the later years of therapy. This extremely low prepubertal norm for random LH using an accurately performed ICMA has been corroborated by other assays [<xref ref-type="bibr" rid="B18">18</xref>].</p><p>In comparison with the continued elevation of random LH levels, the mean peak LH of 0.62 mIU/mL during therapy in the current study is markedly less than the normal mean prepubertal peak LH (2.0 ± 1.5 mIU/mL) for this ICMA assay [<xref ref-type="bibr" rid="B15">15</xref>]. Mean GnRHa-stimulated levels in the later years of therapy fall to near-equivalence with random LH levels. These findings imply that chronic GnRH superagonist therapy results in low-level tonic LH secretion, but nearly complete suppression of pulsatility. As a consequence, only the GnRHa-stimulated LH level during therapy provides clear biochemical confirmation of suppression of the pubertal axis.</p><p>During histrelin therapy, assessment of peak LH requires an aqueous leuprolide injection, unlike GnRHa injection therapies in which the therapeutic injection itself can be used as the stimulating agent [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B18">18</xref>]. Thus, it is not surprising that some practitioners have been using unstimulated LH levels for convenience or cost savings in monitoring histrelin therapy, along with clinical features and random sex steroid levels. Some utilize 24-hour leuprolide-stimulated estradiol, which circumvents estradiol assay limitations but is likely more inconvenient to obtain than stimulated LH because of the necessity of a return visit. A consensus statement 5 years ago on pediatric uses of GnRHa [<xref ref-type="bibr" rid="B1">1</xref>] did not take a position on the utility of random LH for monitoring because the practice was common and published data using sensitive LH assays were at that time limited. Our findings clearly demonstrate that the practice of relying upon random LH for monitoring should be used cautiously, if at all.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>In this study of 36 children, no treatment failures occurred during histrelin implant therapy as assessed by leuprolide stimulation testing and by clinical parameters such as cessation of pubertal progression and diminished growth velocity. Considering that 89% of these subjects exhibited a random LH in the pubertal range at some point during therapy, random LH is unsuitable for routine therapeutic monitoring. A two-step test sequence of random LH followed by as-needed stimulation testing seems futile when failure of the first test is so common. An argument can be made that clinical parameters alone (particularly slowing of growth and decrease in breast size) suffice for treatment monitoring, at least during histrelin therapy. A GnRHa stimulation test with estradiol should be performed if laboratory confirmation is desired, specifically if there are doubts about clinical suppression.</p></sec><sec><title>Abbreviations</title><p>CPP: Central precocious puberty; FSH: Follicle-stimulating hormone; GnRHa: Gonadotropin-releasing hormone agonist; ICMA: Immunochemiluminescent assay; LCMS/MS: Liquid chromatography and tandem mass spectrometry; LH: Luteinizing hormone; RIA: Radioimmunoassay; SD: Standard deviation; SE: Standard error.</p></sec><sec><title>Competing interests</title><p>EKN, LAS, MEG, and PST have received research support and have been a consultant, advisory board member, or speaker’s bureau member for Endo Pharmaceuticals Inc. EKN, LAS, and MEG have received research support and have been a consultant, advisory board member, or speaker’s bureau member for Abbott. TMD is an employee, and EG was formerly an employee of Endo Pharmaceuticals Inc. and is currently an employee of Synchrony Healthcare, West Chester, PA). Funding was provided by Endo Pharmaceuticals Inc to support editorial assistance in the preparation of the manuscript.</p></sec><sec><title>Authors’ contributions</title><p>The authors take full responsibility for all content. The authors verify that they have met all of the journal’s requirements for authorship and that they have not received compensation for this work. All authors were involved in study design and the collection, analysis, and interpretation of data. EKN was primarily responsible for writing the manuscript. All authors have read, approved the final manuscript, and made the decision to submit the manuscript for publication.</p></sec>
Opening a can of centipedes: new insights into mechanisms of body segmentation	<p>The search for a common developmental genetic mechanism of body segmentation appears to become more difficult, and more interesting, as new segmented organisms are added to the roster. Recent work in this journal by Brena and Akam on segmentation of the geophilomorph centipede <italic>Strigamia maritima</italic>, an arthropod distantly related to the standard insect models, contains developmental and evolutionary surprises that highlight the importance of a wider sampling of phyla.</p><p>See research article: <ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1741-7007/11/112">http://www.biomedcentral.com/1741-7007/11/112</ext-link></p>	<contrib contrib-type="author" id="A1"><name><surname>Valentin</surname><given-names>Guillaume</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>gvalent@nimr.mrc.ac.uk</email></contrib><contrib contrib-type="author" corresp="yes" id="A2"><name><surname>Oates</surname><given-names>Andrew C</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>aoates@nimr.mrc.ac.uk</email></contrib>	BMC Biology	<sec><title/><p>The evolution of body axis segmentation is the subject of an historical debate in which the search for homologies has recently focused on the developmental mechanisms underlying segment formation in three taxa: annelids, arthropods and chordates. Originally, in his classical Articulata hypothesis based on morphological traits, Cuvier (1817) proposed that annelids and arthropods shared a common segmented ancestor, whereas the chordates had independently evolved segmentation. With the elucidation of the new animal phylogeny based on ribosomal RNA gene sequences in which Bilateria are divided into three ancient clades - Lophotrochozoa, Ecdysozoa, and Deuterostomia (containing annelids, arthropods, and chordates, respectively) - came the argument from parsimony that because segmented bodies are a minority in each clade, they are most likely independently evolved. Two more arguments further defend this hypothesis. First, there is variety of developmental processes underlying segmentation among the three clades; second, the germ layers that are initially segmented are different: with some exceptions, most chordates and arthropods primarily segment mesoderm and ectoderm, respectively - annelids segment both layers at the same time.</p><p>Modern evo-developmental biology has now entered this discussion with findings of homology between segmentally expressed genes giving rise to the hypothesis that the last common ancestor of all three clades, <italic>Urbilateria</italic>, was segmented [<xref ref-type="bibr" rid="B1">1</xref>]. A corollary is that segmentation of the body axis must have been lost at several points during evolution [<xref ref-type="bibr" rid="B2">2</xref>]. In this issue, Brena and Akam have extended the analysis of segmentation expression dynamics in the centipede <italic>Strigamia maritime</italic>[<xref ref-type="bibr" rid="B3">3</xref>]. Their new work raises several fundamental questions about the mechanisms and evolution of segmentation in arthropods, and its similarities to that in chordates.</p><p>Before examining the candidate molecular systems that have come to light, it is important to highlight the similarities and differences in how various embryos grow and elongate their body axis (germband), and how morphological segmentation is integrated into this growth mode (Figure <xref ref-type="fig" rid="F1">1</xref>). One extreme is provided by the long germband insects, with the most famous example being the beloved fruitfly <italic>Drosophila</italic>. In these embryos, segmentation occurs simultaneously along the body axis in the absence of elongation. Others, such as the short germband flour beetle <italic>Tribolium</italic>, exhibit simultaneous segmentation of the head parts, but sequentially segment their bodies in concert with posterior growth at a terminal growth zone [<xref ref-type="bibr" rid="B4">4</xref>]. The bee <italic>Apis</italic> shows a remarkable intermediate mode: segments form sequentially in a body that, much like <italic>Drosophila</italic>, does not elongate during the process [<xref ref-type="bibr" rid="B5">5</xref>]. Overall, the sequential mode, where elongation and segmentation are tightly coordinated, appears to be the most prevalent across the invertebrates, and is shared with the vertebrates.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Bilaterian phylogeny highlighting the key features associated with body segmentation.</bold> For each segmented phylum (red) we listed a representative model organism, the germ layers that are primarily segmented (blue), the germ band development adopted among arthropods (purple), the segmentation dynamics (either sequential or simultaneous addition of segments, green), the initial patterning periodicity (orange), the tissue patterning that underlies segment formation (pink), and whether posterior growth occurs during segmentation (yellow). Onycophorans and Echiurans have a less pronounced, or partial segmentation of the body. The phylogeny is a broad consensus of molecular and morphological traits. Note that we have displayed a trichotomy of pancrustaceans, myriapods and chelicerates, as these relationships are contested. The branch lengths are arbitrary.</p></caption><graphic xlink:href="1741-7007-11-116-1"/></fig><p>What molecular mechanisms underlie these various segmentation systems? In <italic>Drosophila</italic>, the best understood case, maternally supplied signal gradients along the anterior-posterior axis trigger a genetic cascade of transcription factors that subdivide the embryo. This process results in expression of the so-called Pair-rule genes, which initially demarcate a two-segment periodicity [<xref ref-type="bibr" rid="B4">4</xref>]: two morphological segments form in the interval along the axis defined by one repeat of pair-rule gene expression. Two-segment periodicity is common in insects, but apart from <italic>Strigamia</italic>, a single segment periodicity is the rule in other arthropods and in vertebrates.</p><p>In contrast, evidence from <italic>Tribolium</italic>, cockroach <italic>Periplaneta americana</italic>, and spider <italic>Cupiennius salei</italic>, which all have short germband growth and sequential segmentation, has suggested that a clock-based mechanism is at work across the arthropods [<xref ref-type="bibr" rid="B6">6</xref>-<xref ref-type="bibr" rid="B8">8</xref>]. Homologs of the <italic>Drosophila</italic> pair-rule gene <italic>Hairy</italic> were among the genes observed with wave-like, cyclic expression patterns in most of these arthropods.</p><p>An oscillating molecular mechanism underlying segmentation was first discovered in vertebrates, where a periodic gene expression signal involving <italic>Hairy</italic> gene homologs is converted into regularly sized mesodermal segments, called somites [<xref ref-type="bibr" rid="B9">9</xref>]. In this case, each segment along the body axis is formed by the same mechanism, repeating over and over. Therefore, the growth zone in some arthropods and the pre-somitic mesoderm (PSM) in vertebrates can be thought of as a population of genetic oscillators that act as a rhythmic patterning system, or, in other words, a segmentation clock [<xref ref-type="bibr" rid="B10">10</xref>]. Strikingly, however, the homology of the oscillating genetic circuits appears weak. The only genes observed with cyclic expression stripes (implying a candidate oscillator component) in any members of both Chordata and Arthropoda are <italic>Hairy</italic> and <italic>Delta</italic> homologs. And yet cyclic expression of these genes is not seen in all arthropods; it is notably absent from the growth zone of <italic>Tribolium</italic>.</p><p>Investigating oscillatory or other dynamic genetic processes in species with well-developed sets of molecular and transgenic tools is a formidable challenge. In a species without these tools, or where samples must be collected in the wild, as is the case for <italic>Strigamia</italic>, it is more difficult still. Previous observations of wave-like gene expression patterns, including a <italic>Hairy</italic> homolog, suggested that segmentation in <italic>Strigamia</italic> might be under the control of a segmentation clock. However, without knowledge of the relative movement of cells and dynamics of gene expression, a lineage-based pair-rule mechanism could not be ruled out. In the current paper, Brena and Akam looked in carefully age-matched embryos at the expression of a pair-rule gene, <italic>even-skipped</italic>, and the Notch ligand <italic>Delta</italic>, comparing their wave-like patterns to morphological changes during trunk segmentation (Figure <xref ref-type="fig" rid="F2">2</xref>). They were able to exclude a prominent contribution of cell movement to the patterns of <italic>eve</italic> or <italic>Delta</italic> gene expression. Furthermore, to demonstrate that these dynamic expression patterns reflect intracellular changes in gene expression they used an intron probe to detect the onset of cyclic gene transcription. Even in the absence of live embryo imaging of cyclic gene expression or explant culture these results converge towards demonstrating the existence of a segmentation clock operating in centipede. Although this conclusion may have been anticipated, three new questions arise from the precise description of segmentation provided in the paper.</p><fig id="F2" position="float"><label>Figure 2.</label><caption><p><bold>Graphical representation of </bold><bold><italic>Delta</italic></bold><bold>, </bold><bold><italic>Eve1 </italic></bold><bold>and </bold><bold><italic>Engrailed </italic></bold><bold>expression in </bold><bold><italic>Strigamia</italic></bold><bold>.</bold> Formation of the posterior head and trunk segments is under the control of a clock-like mechanism that manifests as a burst of gene expression in the peri-proctodeal area. <italic>Delta</italic> and <italic>Eve1</italic> expression oscillate out of phase and propagate anteriorly through the posterior disk as a cyclic wave of gene expression. Once the primary stripes of either <italic>Delta</italic> or <italic>Eve1</italic> reach the forming germ-band they stop and <italic>Eve1</italic> and <italic>Delta</italic> intercalary stripes appear. Shortly after, <italic>Engrailed</italic> is expressed in every stripe and a new morphological segment becomes visible. However, when the last nine segments are added to the trunk <italic>Eve1</italic> and <italic>Delta</italic> oscillations cease. At this stage <italic>Eve1</italic> is homogenously expressed in the posterior disc, and a single stripe emerges from this domain in the germ band. <italic>Delta</italic> expression is limited to a stripe that co-localizes with the <italic>Eve1</italic> stripe observed in the germ band. As described for the more anterior segments, <italic>Engrailed</italic> is then expressed and segments are formed.</p></caption><graphic xlink:href="1741-7007-11-116-2"/></fig></sec><sec><title>Head patterned like the body</title><p>In several arthropods, including, for example, <italic>Tribolium</italic>, the head appears to be segmented by a distinct mechanism, as described above for <italic>Drosophila</italic>, which occurs prior to and independently of sequential body segmentation. In <italic>Strigamia</italic>, however, Brena and Akam now show that the segments of the posterior head are demarcated by early expression waves of <italic>Eve</italic> and <italic>Delta</italic> that sweep across most of the blastoderm. These waves appear to be contiguous with those that segment the body, suggesting that the posterior (gnathal) head and body segments are generated by the same mechanism. Given <italic>Strigamia</italic>’s phylogenetic position, this raises the possibility that a clock-like mechanism ancestrally patterned much of the head, and that extant head segmentation modes have been subsequently elaborated from this base. Examination of head segmentation in other basal arthropods, or in members from deeper outgroups, should shed light on this possibility.</p></sec><sec><title>Two-segment to one-segment periodicity within one body</title><p>The body of <italic>Strigamia</italic> is generated in phases with two timescales. The first phase is characterized by rapid formation of the first 38 to 40 leg-bearing segments. During this time, <italic>Delta</italic> and <italic>Eve</italic> primary expression stripes establish a double segment periodicity that predates the formation of morphologically defined segments<italic>.</italic> After this, the remaining nine or so segments are added much more slowly. A fascinating observation is the molecular signature of this switching of gears: the genetic network underlying segmentation appears to shift from an oscillatory to a non-oscillatory mode that correlates with the transition from a double to single segment periodicity. At this time, <italic>Even-skipped2</italic> and <italic>Delta</italic> expression are turned off and <italic>Eve1</italic> is expressed broadly across the growth zone. The existence of another oscillating molecular network that acts as a clock during the final phase of segmentation cannot be ruled out, but perhaps this represents the evolutionary acquisition of a novel segmentation strategy in the posterior. Alternatively, given that the short-bodied centipede <italic>Lithobius</italic> appears to make its segments singly, the oscillating double-segment periodic mechanism in <italic>Strigamia</italic> may have been acquired in the trunk on top of an ancestral single segment periodicity mechanism. Regardless of the exact evolutionary transitions involved, <italic>Strigamia</italic> has shown us two new ways to segment a single axis.</p></sec><sec><title>The role of Delta in segmentation</title><p>Comparison of the known cyclic genes in various arthropods reveals a curious split. In <italic>Periplaneta</italic>, <italic>Cupiennius</italic> and <italic>Strigamia</italic> expression of both <italic>Delta</italic> and <italic>Hairy</italic> homologs appear to oscillate. However, in <italic>Tribolium</italic> neither <italic>Delta</italic> nor <italic>Hairy</italic> homologs are cyclic; instead, <italic>eve</italic> and <italic>odd</italic> show cyclic patterns [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B11">11</xref>]. In vertebrates, <italic>in vivo</italic> experiments indicate that intercellular coupling via the Notch-Delta signaling system synchronizes oscillations between neighboring cells [<xref ref-type="bibr" rid="B12">12</xref>]. This is required to maintain coherent tissue-level stripes of cyclic gene expression, and consequently sharp somite boundaries. Functional evidence in spider and cockroach has shown that inhibition of Notch signaling impairs segmentation (and alters growth) and this has been interpreted as evidence that a Notch-based mechanism is responsible for the oscillations [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B8">8</xref>]. However, an alternative hypothesis is that Delta-Notch signaling is an ancestral mode of coupling cell oscillations during segmentation. In this case, the parts of the internal oscillator might be able to diverge while the coupling mechanism is maintained.</p><p>It is interesting to note that long germband insects lack the patterns of gene expression expected from a segmentation clock; the notion that a clock co-evolved with posterior growth has been proposed [<xref ref-type="bibr" rid="B13">13</xref>]. In vertebrates, both theoretical work and <italic>in vivo</italic> experiments indicate that intercellular coupling via Delta-Notch signaling confers robustness to the system in the presence of developmental noise [<xref ref-type="bibr" rid="B14">14</xref>]. Potential sources include cell proliferation, local cell rearrangement such as migration or convergent-extension, and stochasticity in gene expression. All these processes are inextricably linked to embryonic growth and axis extension. Although there is currently no way to compare gene expression noise between these species, <italic>Strigamia</italic> is a striking example of very strong tissue deformation driving posterior body axis elongation; the posterior progenitor pool occupies the majority of the germband at the onset of segmentation and its cells likely undergo significant mixing during elongation.</p><p>The flour beetle <italic>Tribolium</italic> has a much smaller pool of posterior progenitors in which cell division is likely to be a major contributor to elongation; <italic>Delta</italic> does not oscillate and it is not required for proper segmentation even though <italic>Hairy</italic> is expressed in stripes along the body axis of the embryo [<xref ref-type="bibr" rid="B11">11</xref>]. In this case, one can ask whether the oscillating cells of <italic>Tribolium</italic> need active synchronization. This question is still open and will no doubt generate a lot of excitement in the segmentation microcosm. However functional analysis of Delta-Notch signaling in other arthropods and in centipede in particular will be needed to understand if the segmentation clock and coupling via Notch-Delta signaling co-evolved and why this function may have been lost in <italic>Tribolium</italic>.</p><p>We hypothesize that patterning the growth zone or the PSM via coupled oscillators may be an elegant and robust mechanism to ensure segmental pattern in a tissue where the scale of cellular rearrangements accompanying germband extension would prevent any lineage-based mechanism from working. Thus, a Delta-based mechanism for coupling may be essential in species where a large pool of posterior progenitors is used. Whether this is an ancestral role or not could be investigated by systematically comparing the phylogenetic distribution of cyclic Delta expression with that of the ‘large progenitor pool’ mode of elongation.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>Even without clear homologies, we may nevertheless find common organizing principles of animal segmentation. The use of some form of segmentation clock is clearly one of these. Needing an active way to synchronize cells if they mix significantly during the movements that drive body elongation might be another, and Notch signaling may perform this role.</p></sec>
Do people with HIV infection have a normal life expectancy in the era of combination antiretroviral therapy?	<p>There is evidence that the life expectancy (LE) of individuals infected with the human immunodeficiency virus (HIV) has increased since the introduction of combination antiretroviral therapy (cART). However, mortality rates in recent years in HIV-positive individuals appear to have remained higher than would be expected based on rates seen in the general population. A low CD4 count, whether due to late HIV diagnosis, late initiation of cART, or incomplete adherence to cART, remains the dominant predictor of LE, and thus the individual’s disease stage at initiation of cART (or thereafter) certainly contributes to these higher mortality rates. However, individuals with HIV also tend to exhibit lifestyles and behaviors that place them at increased risk of mortality, particularly from non-AIDS causes. Thus, although mortality rates among the HIV population may indeed remain slightly higher than those seen in the general population, they may be no higher than those seen in a more appropriately matched control group. Thus, further improvements in LE may now only be possible if some of the other underlying issues (for example, modification of lifestyle or behavioral factors) are tackled.</p>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Sabin</surname><given-names>Caroline A</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>c.sabin@ucl.ac.uk</email></contrib>	BMC Medicine	<sec sec-type="intro"><title>Introduction</title><p>Approximately 34.3 million people worldwide are thought to be infected with the human immunodeficiency virus (HIV) [<xref ref-type="bibr" rid="B1">1</xref>]. Left untreated, HIV is inevitably fatal, with a median survival time from seroconversion of 8 to 10 years [<xref ref-type="bibr" rid="B2">2</xref>]. However, the widespread introduction of combination antiretroviral therapy (cART) in many countries in the mid-1990s resulted in a rapid and dramatic reduction in mortality in those living with HIV [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B4">4</xref>]. Although the early cART regimens often included drugs with side effects that limited their efficacy, the drugs used as part of modern cART combinations are generally easier to take, have fewer side effects, and are more forgiving of minor lapses in adherence. As a result, immunological and virological responses to cART have continued to improve over time, with resulting continued reductions in mortality [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B6">6</xref>]. HIV has now come to be viewed by many as a chronic disease and, for the first time, the HIV research community has started to discuss the possibility that life expectancy (LE) in those infected with HIV may now be approaching that seen in the general population.</p><p>The aim of this review is to describe changes in LE in the HIV-positive population since the introduction of cART, and to consider whether this has now reached the same level as in those without HIV infection.</p><sec><title>What is life expectancy?</title><p>LE is an important indicator of health that is used widely by governments, healthcare agencies, and insurance companies to monitor trends in survival over time, and to determine resource allocation [<xref ref-type="bibr" rid="B7">7</xref>]. Formally, LE indicates the average number of years that a person would be expected to survive beyond a given age. That given age would usually be birth [<xref ref-type="bibr" rid="B8">8</xref>]; however, in the context of HIV, the given age may be difficult to interpret as, in most cases, individuals are not born with HIV but acquire it at some point during their life. Thus, LE is commonly quoted from a specific given age [<xref ref-type="bibr" rid="B9">9</xref>-<xref ref-type="bibr" rid="B11">11</xref>] or after some specific event, such as HIV diagnosis [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B13">13</xref>]. Of note, LE at a particular age is not the same as LE at birth minus that age, as LE at a particular age is calculated after conditioning on the fact that the individual has already survived to that age.</p><p>To describe the effect of a particular infection, such as HIV, on LE, investigators may prefer to report the potential years of life lost due to that infection. These may be ‘productive’ life years lost before the age of 65 years [<xref ref-type="bibr" rid="B9">9</xref>], or may be overall years of life lost [<xref ref-type="bibr" rid="B14">14</xref>]. Alternatively, investigators may report the potential gains in LE that could be achieved if that infection (in this case, HIV) were to be eliminated from the population [<xref ref-type="bibr" rid="B15">15</xref>-<xref ref-type="bibr" rid="B17">17</xref>], the excess mortality rates due to HIV [<xref ref-type="bibr" rid="B18">18</xref>], or the standardized mortality ratio (SMR) or mortality rate ratio [<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>], both of which provide a relative measure of the mortality rate in HIV-positive individuals compared with the expected mortality rate in an age-matched uninfected population. The variety of statistics that may be quoted, and the different ages at which LE may be expressed, complicates attempts to summarize LE in the cART era. Table <xref ref-type="table" rid="T1">1</xref> lists reported estimates of LE in the cART era from resource-rich settings, which range from 19.9 years at the age of 25 years in Denmark [<xref ref-type="bibr" rid="B21">21</xref>], to around 75 years from birth in the UK [<xref ref-type="bibr" rid="B8">8</xref>].</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Estimates of LE reported in the cART era</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Reference</bold></th><th align="left"><bold>Cohort/study name</bold></th><th align="left"><bold>Country of study</bold></th><th align="left"><bold>LE in HIV-positive population</bold></th><th align="left"><bold>LE in general population</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Nakagawa <italic>et al</italic>. [<xref ref-type="bibr" rid="B8">8</xref>]<hr/></td><td align="left" valign="bottom">Computer simulation (HIV Synthesis)<hr/></td><td align="left" valign="bottom">UK<hr/></td><td align="left" valign="bottom">LE at birth: 75.0 years if diagnosed with HIV with high CD4 count; 71.5 years if diagnosed with HIV with low CD4 count<hr/></td><td align="left" valign="bottom">LE at birth: estimated from model to be 82.0 years if not infected with HIV<hr/></td></tr><tr><td align="left" valign="bottom">The Antiretroviral Therapy Cohort Collaboration [<xref ref-type="bibr" rid="B9">9</xref>]<hr/></td><td align="left" valign="bottom">ART-CC<hr/></td><td align="left" valign="bottom">Multi-country study (Europe and North America)<hr/></td><td align="left" valign="bottom">LE at age 20: 43.1 years. LE at age 35: 31.7 years<hr/></td><td align="left" valign="bottom">Not stated<hr/></td></tr><tr><td align="left" valign="bottom">Johnson <italic>et al</italic>. [<xref ref-type="bibr" rid="B10">10</xref>]<hr/></td><td align="left" valign="bottom">IeDEA-SA<hr/></td><td align="left" valign="bottom">South Africa<hr/></td><td align="left" valign="bottom">LE at age 20: 27.6 years in men; 36.8 years in women. LE at age 60: 10.1 years in men; 14.4 years in women<hr/></td><td align="left" valign="bottom">Not stated<hr/></td></tr><tr><td align="left" valign="bottom">Mills <italic>et al</italic>. [<xref ref-type="bibr" rid="B11">11</xref>]<hr/></td><td align="left" valign="bottom">The AIDS Support Organization (TASO) cohort<hr/></td><td align="left" valign="bottom">Uganda<hr/></td><td align="left" valign="bottom">LE at age 20: 26.7 years. LE at age 35 years: 27.9 years<hr/></td><td align="left" valign="bottom">LE at age 20: 41 years<hr/></td></tr><tr><td align="left" valign="bottom">Losina <italic>et al.</italic>[<xref ref-type="bibr" rid="B12">12</xref>]<hr/></td><td align="left" valign="bottom">Computer simulation (CEPAC)<hr/></td><td align="left" valign="bottom">USA<hr/></td><td align="left" valign="bottom">LE at age 33: 22.66 years if optimally diagnosed and treated; 19.36 years if treated with cART and adherence follows normal patterns<hr/></td><td align="left" valign="bottom">LE at age 33: 42.91 years for general population; 34.58 years if risk profile similar to those with HIV<hr/></td></tr><tr><td align="left" valign="bottom">Bor <italic>et al</italic>. [<xref ref-type="bibr" rid="B17">17</xref>]<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">KwaZulu-Natal, South Africa<hr/></td><td align="left" valign="bottom">No specific estimates<hr/></td><td align="left" valign="bottom">LE at birth: 52.3 years in 2000; 49.2 years in 2003; 60.5 years in 2011<hr/></td></tr><tr><td align="left" valign="bottom">Lohse <italic>et al</italic>. [<xref ref-type="bibr" rid="B21">21</xref>]<hr/></td><td align="left" valign="bottom">Danish HIV Cohort Study<hr/></td><td align="left" valign="bottom">Denmark<hr/></td><td align="left" valign="bottom">LE at age 25: 8 years in 1995 to 1996; 23 years in 1997 to 1999; 33 years in 2000 to 2005<hr/></td><td align="left" valign="bottom">LE at age 25: 51 years<hr/></td></tr><tr><td align="left" valign="bottom">May <italic>et al</italic>. [<xref ref-type="bibr" rid="B23">23</xref>]<hr/></td><td align="left" valign="bottom">UK Collaborative HIV Cohort Study<hr/></td><td align="left" valign="bottom">UK<hr/></td><td align="left" valign="bottom">LE at age 20: 39.5 years in men; 50.2 years in women. LE at age 35: 30.1 years in men; 37.7 years in women<hr/></td><td align="left" valign="bottom">LE at age 20: 57.8 years in men; 61.6 years in women. LE at age 35: 43.5 years in men; 46.9 years in women<hr/></td></tr><tr><td align="left">van Sighem <italic>et al</italic>. [<xref ref-type="bibr" rid="B41">41</xref>]</td><td align="left">ATHENA Cohort</td><td align="left">The Netherlands</td><td align="left">LE at age 25: 52.7 years in men; 57.8 years in women</td><td align="left">LE at age 25: 53.1 years in men; 58.1 years in women</td></tr></tbody></table><table-wrap-foot><p><italic>Abbreviations</italic>: cART, combination antiretroviral therapy; LE, life expectancy.</p></table-wrap-foot></table-wrap></sec><sec><title>Changes in LE in the cART era</title><p>It is clear that LE has increased since the introduction of cART. Using data from the large CASCADE Collaboration, Bhaskaran [<xref ref-type="bibr" rid="B18">18</xref>] found a continued narrowing in the gap in mortality rates between those seen in individuals infected with HIV with known dates of HIV seroconversion and those that would have been expected based on a demographically similar HIV-negative population. Excess mortality rates in the HIV-positive population dropped by 94% from 31.4 per 1000 person-years (PYRS) prior to 1996 to 6.1 per 1000 PYRS in 2004 to 2006. Mortality rates among 43,355 cART-naive participants in the Antiretroviral Therapy Cohort Collaboration (ART-CC) dropped similarly from 16.3 per 1000 PYRS in 1996 to 1999, to 10.0 per 1000 PYRS in 2003 to 2005 [<xref ref-type="bibr" rid="B9">9</xref>]. LE at 20 and 35 years increased from 36.1 and 25.0 years to 49.4 and 37.3 years, respectively, over the same period, with the potential years of life lost decreasing from 366 per 1000 PYRS to 189 per 1000 PYRS. Among participants with acquired immune deficiency syndrome (AIDS) in the Longitudinal Study of Ocular Complications in AIDS [<xref ref-type="bibr" rid="B22">22</xref>], excess mortality decreased by 8.0% per year from the period 1999 to 2001 to the period 2006 to 2007. LE at age 25 years in the Danish HIV cohort increased from only 8 years in the pre-cART era (1995 to 1996) to 33 years in 2000 to 2005, with LE for a similarly aged uninfected Danish person during that period being 51 years [<xref ref-type="bibr" rid="B21">21</xref>]. Among individuals starting cART in the UK Collaborative HIV Cohort (CHIC) Study, LE at 20 years increased from 30.0 years if cART was started during 1996 to 1999 to 45.8 years if cART was started during 2006 to 2008 [<xref ref-type="bibr" rid="B23">23</xref>]. Of note, improvements in LE in the cART era are not restricted to resource-rich settings: the overall population LE at birth in KwaZulu-Natal, South Africa, is reported to have increased from 49.2 years in 2003 (prior to the scale-up of antiretroviral therapy), to 60.5 years in 2011 [<xref ref-type="bibr" rid="B17">17</xref>].</p></sec><sec><title>Predictors of LE in the cART era: the role of disease stage</title><p>Despite the dramatic improvements in LE witnessed since the introduction of cART, LE may still not have reached the levels seen in the uninfected population. Bhaskaran [<xref ref-type="bibr" rid="B18">18</xref>] reported that even by 2003 to 2005, excess mortality rates in the CASCADE Collaboration remained elevated at 6.1 per 1000 PYRS, and in the ART-CC, potential years of life lost remained high (189 per 1000 PYRS) over the period 2003 to 2005 [<xref ref-type="bibr" rid="B9">9</xref>]. LE in patients starting cART in 2008 in the UK CHIC Study remained lower than that seen in the UK general population (59 years at age 20) [<xref ref-type="bibr" rid="B23">23</xref>]. Among women in the US Women’s Interagency HIV Study (WIHS), the SMR dropped from 24.7 in 1996 to a plateau of 10.3 during 2001 to 2003, despite the addition of a group of younger and healthier women into the cohort in 2001 to 2002 [<xref ref-type="bibr" rid="B19">19</xref>].</p><p>The disease stage of individuals at the time of initiation of cART, and shortly thereafter, may at least partly contribute to the higher than expected mortality rates seen in recent years (Table <xref ref-type="table" rid="T2">2</xref>). In the Longitudinal Study of Ocular Complications in AIDS [<xref ref-type="bibr" rid="B22">22</xref>], excess death rates ranged from 128 per 1000 PYRS in individuals who had cytomegalovirus retinitis, a viral load of greater than 400 copies/ml, and a CD4 count of less than 200 cells/mm<sup>3</sup>, to only 8 per 1000 PYRS for individuals lacking these factors. Interestingly, although excess mortality rates in this study dropped in the cART era by 8.3% per year in those with a CD4 count of less that 200 cells/mm<sup>3</sup>, no significant reduction was seen in those with higher CD4 cell counts. In ART-CC participants [<xref ref-type="bibr" rid="B24">24</xref>], the lowest SMR was seen in men who have sex with men (MSM), who did not have AIDS at cART initiation and who had attained a viral load of 500 copies/ml or lower and a CD4 cell count of 350 cells/mm<sup>3</sup> or higher by 6 months after starting cART. By contrast, the highest SMR was seen in injection drug users who failed to attain a suppressed viral load by 6 months and in whom the CD4 cell count remained at less than 50 cells/mm<sup>3</sup>.</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Summary of factors that may influence LE in people with HIV infection</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Sociodemographic factors</bold></th><th align="left"><bold>Lifestyle/behavioral factors</bold></th><th align="left"><bold>HIV-related factors</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Gender<hr/></td><td align="left" valign="bottom">Smoking<hr/></td><td align="left" valign="bottom">Late HIV diagnosis<hr/></td></tr><tr><td align="left" valign="bottom">Age<hr/></td><td align="left" valign="bottom">Alcohol use<hr/></td><td align="left" valign="bottom">CD4 count at cART initiation<hr/></td></tr><tr><td align="left" valign="bottom">Co-morbidities related to aging<hr/></td><td align="left" valign="bottom">Recreational and injection drug use<hr/></td><td align="left" valign="bottom">CD4 count and HIV RNA attained on cART<hr/></td></tr><tr><td align="left" valign="bottom">Ethnic group/country of origin<hr/></td><td align="left" valign="bottom">Viral hepatitis co-infection<hr/></td><td align="left" valign="bottom">Clinical AIDS prior to cART initiation<hr/></td></tr><tr><td align="left" valign="bottom">Place of residence/neighborhood<hr/></td><td align="left" valign="bottom">Sexually transmitted infections<hr/></td><td align="left" valign="bottom">Attendance at outpatient clinics<hr/></td></tr><tr><td align="left">Socioeconomic status</td><td align="left"> </td><td align="left">Adherence to cART</td></tr></tbody></table><table-wrap-foot><p><italic>Abbreviations</italic>: cART, combination antiretroviral therapy; LE, life expectancy.</p></table-wrap-foot></table-wrap><p>The important association between the pre-cART CD4 count and LE has been described in several other studies. In the UK CHIC Study [<xref ref-type="bibr" rid="B23">23</xref>], individuals started on cART in line with UK guidelines (at a CD4 cell count of 200 to 350 cells/mm<sup>3</sup>) experienced a LE at age 20 of 53.4 years, only marginally shorter than that seen in the general male (57.8 years) and female (61.6 years) populations. By contrast, LEs at age 20 were only 41.0 and 37.9 years among those started on cART at a CD4 count of 100 to 199 and less than 100 cells/mm<sup>3</sup>, respectively. Among cART-treated South African individuals, LE at age 20 ranged from 43.1 years if the CD4 count was 200 cells/mm<sup>3</sup> or higher to 29.5 years if the CD4 count was 50 cells/mm<sup>3</sup> or lower [<xref ref-type="bibr" rid="B10">10</xref>]. In Australian cART-treated individuals [<xref ref-type="bibr" rid="B25">25</xref>], the SMR increased from 1.5 among individuals with a CD4 count of 500 cells/mm<sup>3</sup> or higher to 8.6 among those with a CD4 cell count of 350 cells/mm<sup>3</sup> or lower. Finally, among HIV-positive individuals in the Study of Fat Redistribution and Metabolic change in HIV Infection (FRAM), mortality rates were 2.3 times higher than in HIV-negative controls in individuals with a CD4 count of greater than 350 cells/mm<sup>3</sup>, but 6.3 times higher in those with a CD4 count of less than 350 cells/mm<sup>3</sup>[<xref ref-type="bibr" rid="B26">26</xref>]. Thus, it is clear that a low CD4 count, whether due to late diagnosis of HIV, late initiation of cART, or incomplete adherence to cART, remains the dominant predictor of LE in the cART era.</p></sec><sec><title>Predictors of LE in the cART era: the role of non-HIV factors</title><p>Although stage of HIV disease at cART initiation is strongly associated with LE, other factors may also play a role. (Table <xref ref-type="table" rid="T2">2</xref>) Individuals with HIV are known to exhibit lifestyles and behaviors that put them at higher risk of mortality than the general population, regardless of HIV status, including higher rates of smoking, alcohol and recreational drug use, and viral and sexually transmitted co-infections [<xref ref-type="bibr" rid="B27">27</xref>-<xref ref-type="bibr" rid="B30">30</xref>]. Current smoking was an additional risk factor for death in HIV-positive individuals in the FRAM Study [<xref ref-type="bibr" rid="B26">26</xref>], and in a recent study from the Danish HIV Cohort, Helleberg <italic>et al</italic>. [<xref ref-type="bibr" rid="B31">31</xref>] reported that those with HIV may now lose more life years to smoking than to HIV itself. Among participants in the ART-CC [<xref ref-type="bibr" rid="B9">9</xref>], injection drug users had a LE that was around 13 years shorter at age 20, and 10 years shorter at age 35, than non-injection drug users. The percentage of participants in this study with a SMR less than 2 (that is, individuals whose mortality patterns most closely resembled those in the general population) was 46% in MSM, 42% in those infected with HIV through heterosexual sex, and 0% among injection drug users; the corresponding percentages of participants with a SMR greater than 10 (individuals with the worst mortality patterns) were 4%, 14%, and 47%,s respectively [<xref ref-type="bibr" rid="B24">24</xref>], confirming the negative impact of injection drug use and/or hepatitis co-infection on overall mortality rates [<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B21">21</xref>].</p><p>Although these non-HIV factors may have only a limited influence on deaths from AIDS-related causes, they may play a more major role in deaths from non-AIDS causes, which appear to have increased in frequency in the cART era. In the WIHS Study [<xref ref-type="bibr" rid="B19">19</xref>], deaths from non-AIDS causes increased in the cART era, and by 2001 to 2004, they accounted for the majority of deaths that occurred; it was this increase in non-AIDS deaths that was thought to contribute to the plateau in the SMR seen from 2001 among women in the study. Whereas the mortality rate ratio for deaths from non-AIDS causes in non-injection drug users in the Danish HIV Cohort had dropped from 4.5 in 1995 to 1.3 in 2008, it had increased from 7.0 to 10.3 over the same period in injection drug users [<xref ref-type="bibr" rid="B32">32</xref>]. In a direct comparison with the Multicenter AIDS Cohort Study (MACS), Wada [<xref ref-type="bibr" rid="B33">33</xref>] reported that median LE for non-AIDS causes was almost 10 years shorter in women in the WIHS (55.9 years) than in men in the MACS (66.0 years), contributing to an overall difference in age at death between men and women of 11.6 years. Further evidence of the potential role of non-HIV factors in mortality rates comes from Alabama [<xref ref-type="bibr" rid="B34">34</xref>], where patients who missed visits in the first year after initiating outpatient treatment for HIV had over twice the rate of long-term mortality compared with those attending all scheduled appointments, and from Canada, where a three-fold increased risk of death was seen in cART-treated HIV-positive individuals who lived in neighborhoods with a high concentration of injection drug users, relative to those who lived in neighborhoods with a high concentration of MSM [<xref ref-type="bibr" rid="B35">35</xref>].</p><p>To investigate the potential effect of these external factors on the mortality rates seen, Lohse [<xref ref-type="bibr" rid="B36">36</xref>] used data from the Danish general population to show that only around 55% of deaths that occurred in the Danish HIV cohort could be attributed to HIV, with 32% of deaths being attributed to hepatitis C virus co-infection and/or other co-morbidities, and the remaining 14% being unrelated to either HIV or co-morbidities. Losina and colleagues [<xref ref-type="bibr" rid="B12">12</xref>] used the CEPAC model, a state-transition model of HIV infection, to quantify the potential influence on LE of various lifestyle and behavioral factors. They found that in the general US population, LE at age 33 (the mean age at seroconversion in the USA) was around 43 years [<xref ref-type="bibr" rid="B12">12</xref>], but this dropped to 34.58 years when the authors selected a cohort from the HIV-negative population that matched their HIV-positive population in terms of several lifestyle and sexual risk factors. The authors were then able to estimate that HIV infection, when appropriately treated and diagnosed at an early stage, would lead to a further loss of LE of around 11.92 years, with late diagnosis, late initiation of cART, and early discontinuation of cART further reducing LE by an additional 3.3 years [<xref ref-type="bibr" rid="B12">12</xref>].</p></sec><sec><title>Can we improve LE further?</title><p>Late HIV diagnosis remains extremely common in many countries [<xref ref-type="bibr" rid="B37">37</xref>], and has been reported to be a major risk factor for mortality [<xref ref-type="bibr" rid="B38">38</xref>]. In Brazil, it was estimated that 95.5% of deaths occurring in the first year after diagnosis were attributable to late diagnosis [<xref ref-type="bibr" rid="B39">39</xref>]; study investigators estimated that averting late diagnosis would have reduced the AIDS mortality rate 2003 to 2006 by 39.5%, a similar reduction to that produced by cART. In the UK, earlier diagnosis would have reduced short-term (first year after diagnosis) mortality by 84% in MSM [<xref ref-type="bibr" rid="B38">38</xref>] and by 56% in those infected heterosexually [<xref ref-type="bibr" rid="B40">40</xref>]. Using the HIV Synthesis model, a stochastic computer simulation model of HIV progression, Nakagawa [<xref ref-type="bibr" rid="B8">8</xref>] showed that LE from birth was 71.5 years, with 10.5 years lost to HIV infection, in a scenario in which diagnosis occurred at a late stage of HIV infection (median CD4 count 140 cells/mm<sup>3</sup>), but under a scenario of earlier diagnosis (median CD4 count 432 cells/mm<sup>3</sup>), LE from birth was 75.0 years, with only 7.0 years lost, on average, due to HIV. Thus, earlier diagnosis of HIV might go some way to improve LE further.</p><p>Among those diagnosed and receiving cART, efforts to ensure that all individuals attain optimal CD4 levels may also lead to improvements in LE. Lewden calculated SMR for individuals in the COHERE collaboration who had attained a CD4 count of 500 cells/mm<sup>3</sup> or higher on cART [<xref ref-type="bibr" rid="B20">20</xref>]. For men, attaining a CD4 count of 500 cells/mm<sup>3</sup> or higher for just over 1 year was sufficient to ensure that their mortality rates were similar to those in the general population. For women, however, SMR remained above 1, even among those who had maintained a CD4 count of 500 cells/mm<sup>3</sup> or higher for over 5 years. The potential for further improvement in LE was also studied in the Dutch ATHENA cohort [<xref ref-type="bibr" rid="B41">41</xref>]; LE at age 25 among HIV-positive participants who had been diagnosed during 1998 to 2007 and who remained AIDS-free and untreated for 24 weeks after diagnosis was 52.7 years in men (versus 53.1 years in the general population) and 57.8 years in women (versus 58.1 years). The authors noted that individuals included in the study were highly selected (injection drug users were excluded) with a median CD4 count at 24 weeks after diagnosis of 480 cells/mm<sup>3</sup>, and therefore the outcomes reported reflect the potential outcomes that might be feasible in a group of patients diagnosed and treated at an early stage of infection. Of note, there is some evidence to suggest a small potential benefit of cART (through a reduction in CD4 loss) if it is initiated during primary HIV infection [<xref ref-type="bibr" rid="B42">42</xref>]. Although such benefits may translate into further improvements in LE, any effect at a population level is likely to be small, given the difficulties in diagnosing individuals with HIV infection at such an early stage.</p><p>Earlier HIV diagnosis and optimal cART initiation aside, do we still have some way to go to improve LE, or have we already reached the maximum LE that might be anticipated in this population? Although LE in those with HIV infection is generally compared with that seen in the general population in the same country, LEs vary tremendously both between and within countries. In the UK, for example, male LE at birth in 2007 to 2009 ranged from 84.4 years for those living in parts of London to 73.1 years for those living in parts of Glasgow [<xref ref-type="bibr" rid="B43">43</xref>]. Even within a city such as London, there may be large differences in LE in different areas, as shown by the Lives on the Line project (<ext-link ext-link-type="uri" xlink:href="http://life.mappinglondon.co.uk/">http://life.mappinglondon.co.uk/</ext-link>). These differences may be explained by differences in the characteristics of those living in different regions, particularly socioeconomic status, lifestyle factors, and dietary factors. When LE is compared between the HIV-positive and the general populations, therefore, the two populations may have a different underlying risk of mortality, and LEs may be expected to differ from that in the general population. The identification of appropriately matched HIV-negative control populations, with similar lifestyle and behavioral characteristics, for the provision of comparative estimates of LE, would go some way to addressing this concern.</p><p>This inability to eliminate residual confounding is a limitation of any comparison based on observational data. However, LE also suffers from several other limitations. Firstly, LE is generally based on current mortality rates and does not take into consideration any improvements to patient management that may occur in the future (leading to an underestimate of future LE) nor to any longer-term possible adverse outcomes of cART or HIV infection (leading to an overestimate of future LE). Secondly, the estimation of LE often requires long-term extrapolation of mortality rates from individuals followed over a relatively short period. After all, HIV has only been around for 30 years or so, a relatively short time compared with the length of an individual’s lifetime. Finally, LE is only as good as the ascertainment of deaths within a cohort; where deaths are not fully ascertained, LE may appear artificially high. Using information collected from cohort studies in West Africa, Cote d’Ivoire, and Burkina Faso, Lewden <italic>et al</italic>. [<xref ref-type="bibr" rid="B44">44</xref>] reported that the highest estimates of mortality were seen in cohorts with the lowest rates of loss to follow-up. Verguet <italic>et al</italic>. [<xref ref-type="bibr" rid="B45">45</xref>] subsequently reported that whereas the best estimate of life years gained by a person in Africa in the first 5 years after starting cART was 2.1 [<xref ref-type="bibr" rid="B45">45</xref>], this estimate could drop by approximately 14% if mortality rates among those lost to follow-up were assumed to be 100%, or could increase by 19% if zero mortality was assumed in this group. In cohorts participating in the ART-CC, incomplete death ascertainment was reported to contribute to the higher mortality rates seen in the North American compared with European cohorts, although other patient factors also played a role [<xref ref-type="bibr" rid="B46">46</xref>].</p></sec></sec><sec sec-type="conclusions"><title>Conclusions</title><p>With the limitations described above in mind, it is possible that LE may now have reached levels that we would expect to see in this population. At this stage, it is possible that further major improvements in LE may only be achievable by tackling some of the other underlying issues, such as earlier HIV diagnosis (through enhanced opportunities for testing and greater awareness of the early signs of HIV infection) and improved retention in HIV care, earlier cART initiation, or the modification of lifestyle or behavioral factors.</p></sec><sec><title>Abbreviations</title><p>AIDS: Acquired immune deficiency syndrome; ART-CC: Antiretroviral therapy cohort collaboration; cART: Combination antiretroviral therapy; HIV: Human immunodeficiency virus; LE: Life expectancy; MACS: Multicenter AIDS cohort study; MSM: Men who have sex with men; PYRS: Person-years; SMR: Standardised mortality ratio; UK CHIC study: UK collaborative HIV cohort study; US: United States; WIHS: Women’s interagency HIV study.</p></sec><sec><title>Competing interests</title><p>The author declares that she has no competing interests.</p></sec><sec><title>Author’s information</title><p>CS is a professor of Medical Statistics and Epidemiology at University College London (UCL). She has worked for many years on the analysis of large observational HIV databases, with a particular interest in raising awareness of the biases inherent in these databases. She is the principal investigator on the UK CHIC Study, principal statistician on the D:A:D Study, and has worked with many other research groups in the UK and elsewhere.</p></sec>
Containing the accidental laboratory escape of potential pandemic influenza viruses	Could not extract abstract	<contrib contrib-type="author"><name><surname>Merler</surname><given-names>Stefano</given-names></name><address><email>merler@fbk.eu</email></address><xref ref-type="aff" rid="Aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Ajelli</surname><given-names>Marco</given-names></name><address><email>ajelli@fbk.eu</email></address><xref ref-type="aff" rid="Aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Fumanelli</surname><given-names>Laura</given-names></name><address><email>lfumanelli@fbk.eu</email></address><xref ref-type="aff" rid="Aff1">1</xref><xref ref-type="aff" rid="Aff2">2</xref></contrib><contrib contrib-type="author" corresp="yes"><name><surname>Vespignani</surname><given-names>Alessandro</given-names></name><address><email>a.vespignani@neu.edu</email></address><xref ref-type="aff" rid="Aff3">3</xref><xref ref-type="aff" rid="Aff4">4</xref><xref ref-type="aff" rid="Aff5">5</xref></contrib><aff id="Aff1"><label>1</label><institution-wrap><institution-id institution-id-type="GRID">grid.38225.3e</institution-id><institution>Bruno Kessler Foundation, </institution></institution-wrap>Trento, Italy </aff><aff id="Aff2"><label>2</label><institution-wrap><institution-id institution-id-type="GRID">grid.11696.39</institution-id><institution-id institution-id-type="ISNI">0000000419370351</institution-id><institution>Department of Mathematics, </institution><institution>University of Trento, </institution></institution-wrap>Trento, Italy </aff><aff id="Aff3"><label>3</label><institution-wrap><institution-id institution-id-type="GRID">grid.261112.7</institution-id><institution-id institution-id-type="ISNI">0000000121733359</institution-id><institution>Laboratory for the Modeling of Biological and Socio-technical Systems, </institution><institution>Northeastern University, </institution></institution-wrap>Boston, 02115 MA USA </aff><aff id="Aff4"><label>4</label><institution-wrap><institution-id institution-id-type="GRID">grid.418750.f</institution-id><institution-id institution-id-type="ISNI">0000000417593658</institution-id><institution>Computational Epidemiology Laboratory, </institution><institution>Institute for Scientific Interchange (ISI), </institution></institution-wrap>Torino, Italy </aff><aff id="Aff5"><label>5</label><institution-wrap><institution-id institution-id-type="GRID">grid.38142.3c</institution-id><institution-id institution-id-type="ISNI">000000041936754X</institution-id><institution>Institute for Quantitative Social Sciences at Harvard University, </institution></institution-wrap>Cambridge, MA 02138 USA </aff>	BMC Medicine	<sec id="Sec1"><title>Background</title><p>The risk associated with the accidental laboratory escape of potential pandemic pathogens is under the magnifying lens of research and policy making communities [<xref ref-type="bibr" rid="CR1">1</xref>, <xref ref-type="bibr" rid="CR2">2</xref>]. The recent debate on the genetic manipulation of highly virulent influenza viruses [<xref ref-type="bibr" rid="CR3">3</xref>, <xref ref-type="bibr" rid="CR4">4</xref>] has made clear the necessity for quantitative risk/benefit assessment before starting research projects involving biosafety level (BSL) 3 and 4 agents. According to data collected in 2010 and 2011, the number of BSL 4 laboratories worldwide is 38 [<xref ref-type="bibr" rid="CR5">5</xref>], mostly concentrated in the US (10) and Europe (14). The official number of BSL 3 facilities worldwide is unknown, since most laboratories where research on infectious diseases is carried out and many hospital laboratories operate at safety level 3. Their number, however, is of the order of several thousands: there were 1,362 in the US alone in 2008 [<xref ref-type="bibr" rid="CR6">6</xref>]. According to data collected in 2010, the number of US workers with approved access to biological select agent and toxin (BSAT) was 10,639 [<xref ref-type="bibr" rid="CR7">7</xref>]. From 2004 to 2010, 639 release reports were reported to the Centers for Disease Control (CDC), 11 of them reporting laboratory-acquired infections that, however, did not result in fatalities or secondary transmission [<xref ref-type="bibr" rid="CR7">7</xref>]. A list of recently reported laboratory-acquired infections is available (see [<xref ref-type="bibr" rid="CR8">8</xref>]). A rigorous risk assessment is a scientific challenge <italic>per se</italic> [<xref ref-type="bibr" rid="CR9">9</xref>–<xref ref-type="bibr" rid="CR11">11</xref>]. Although the estimates of the probability of accidental escape are relatively low (0.3% risk of release per lab per year [<xref ref-type="bibr" rid="CR11">11</xref>]), the increased number of laboratories working on BSL 3 and 4 agents gives rise to estimates projecting an appreciable combined escape risk of potential pandemic pathogens (PPP) in a 10-year window [<xref ref-type="bibr" rid="CR11">11</xref>]. In addition, for PPP, the relatively small risk of release has to be weighted against the size of the population that could be affected by such an event, the risk of severe or fatal cases and the likelihood of containment before the event could escalate to global proportions. Furthermore, the quantitative analysis of the post-release scenario is complicated by the different social and environmental settings that apply to the more than 1,500 BSL 3 and 4 laboratories around the world [<xref ref-type="bibr" rid="CR9">9</xref>].</p><p>Here, we perform a quantitative analysis of (accidental) post-release scenarios from a BSL facility, focusing on the likelihood of containment of the accidental release event. Although BSL 4 agents, such as Ebola virus and Marburg virus, are considered the most dangerous to handle because of the often fatal outcome of the disease, they are unlikely to generate global risk because of their inefficient mechanism of person-to-person transmission and other features of the natural history of the induced diseases [<xref ref-type="bibr" rid="CR12">12</xref>, <xref ref-type="bibr" rid="CR13">13</xref>]. It is therefore understood that the major threat of a pandemic escalation is provided by modified influenza viruses [<xref ref-type="bibr" rid="CR10">10</xref>], and for this reason we focused our work on the accidental release of novel influenza strain in a densely populated area of Europe. We used a highly detailed agent-based model that specifically considers laboratory workers and their household in order to test the detailed implementation of non-pharmaceutical containment measures in the very early stage of the release/outbreak scenario. The model allowed analysis of the progression of the epidemic at the level of single individual. We could therefore assess the likelihood of containment as a function of a wide range of interventions, and provide a discussion of different geographical settings (for example, rural vs urban seeding) by analyzing the effects of population density and structure. Differently from methods employed to estimate the probability of containing naturally emerging pathogens at the source, here we assumed that epidemiological surveillance is presumably enhanced in areas where BSL laboratories are located, thus increasing the likelihood of quickly detecting symptomatic cases. Moreover, we assumed that this makes it possible to put in place intervention measures (for example, social distancing measures and contact tracing) at the very beginning of the epidemic. A number of factors determine the controllability of an outbreak, including the uncertainty in the efficacy of the containment policies recorded in the literature. For this reason we performed a very extensive sensitivity analysis on the efficacy of implemented policies and the disease natural history. In terms of specific interventions implemented, our analysis is inspired by the experience of an accidental release of severe acute respiratory syndrome (SARS) in August 2003 from a laboratory in Singapore [<xref ref-type="bibr" rid="CR14">14</xref>]: a total of 8 household contacts, 2 community contacts, 32 hospital contacts, and 42 work contacts were identified, of whom 25 were placed under home quarantine. Both laboratories where the patient had worked were closed as a precautionary measure. Specifically as regards contact tracing, its efficacy for tuberculosis (TB) is ascertained (large-scale studies tracing contacts of TB patients in the US and Canada found high incidence rates of active TB (200 to 2,200 cases per 100,000 individuals) against 5 to 10 per 100,000 in the general population [<xref ref-type="bibr" rid="CR15">15</xref>–<xref ref-type="bibr" rid="CR17">17</xref>]). In contrast, contact tracing was performed in the case (described above) of accidental release of SARS and in another case of SARS [<xref ref-type="bibr" rid="CR18">18</xref>] (1,000 persons traced), but no secondary infections were detected. The two most critical quantities affecting the temporal pattern of spread of influenza viruses, and containment probabilities as well, are the generation time (the distribution of the time interval between infection of a primary case and infection of a secondary case caused by the primary case), and the basic reproduction number R<sub>0</sub>. We analyzed different scenarios by assuming transmissibility comparable to that observed in past influenza pandemics, for example, the 2009 H1N1 virus (namely R<sub>0</sub> or effective transmissibility in the range 1.2 to 1.6 [<xref ref-type="bibr" rid="CR19">19</xref>–<xref ref-type="bibr" rid="CR24">24</xref>]) or 1918 Spanish influenza (R<sub>0</sub> = 1.8 or higher [<xref ref-type="bibr" rid="CR25">25</xref>]), and generation time distributions consistent with current estimates for influenza (in the range 2.5 to 4 days [<xref ref-type="bibr" rid="CR23">23</xref>, <xref ref-type="bibr" rid="CR26">26</xref>–<xref ref-type="bibr" rid="CR29">29</xref>]). Beyond these factors, intervention efficacy depends on probability of developing clinical symptoms and length of the incubation period, as they affect, respectively, the probability of detecting cases and the probability of stopping the transmission chain through rapid identification of secondary cases. All these factors make influenza different from other potential pandemic pathogens. For instance, SARS is characterized by a very long incubation period (1 to 2 days for influenza, up to 10 days for SARS [<xref ref-type="bibr" rid="CR30">30</xref>]) and by a low proportion of infections generated by asymptomatic infections (up to 50% for influenza, negligible for SARS [<xref ref-type="bibr" rid="CR30">30</xref>]). The R<sub>0</sub> of SARS was estimated to be slightly larger than that of influenza, namely in the range 2 to 3 [<xref ref-type="bibr" rid="CR30">30</xref>]. Smallpox, similar to SARS, is another potentially pandemic pathogen characterized by a low proportion of infections generated by asymptomatic infections [<xref ref-type="bibr" rid="CR30">30</xref>], though characterized by a larger R<sub>0</sub> (in the range of 5 to 10 [<xref ref-type="bibr" rid="CR30">30</xref>]). In contrast, Marburg hemorrhagic fever is characterized by a low R<sub>0</sub> (about 1.5 [<xref ref-type="bibr" rid="CR12">12</xref>]) and short incubation period (about 2 days, with an overall generation time of 8 to 10 days [<xref ref-type="bibr" rid="CR12">12</xref>]).</p></sec><sec id="Sec2"><title>Methods</title><p>In order to provide a quantitative assessment of the containment likelihood and the detailed modeling of interventions we used a stochastic microsimulation model structurally similar to the one used elsewhere (see [<xref ref-type="bibr" rid="CR19">19</xref>, <xref ref-type="bibr" rid="CR31">31</xref>]) to generate simulations of pandemic events. The model is a spatially explicit stochastic individual-based model of influenza transmission with force of infection decreasing with the distance and explicit transmission in households, schools and workplaces (see Additional file <xref rid="MOESM1" ref-type="media">1</xref> for details). This model has been validated with data from the H1N1 2009 pandemic [<xref ref-type="bibr" rid="CR19">19</xref>] and compared and tested against other large-scale computational approaches [<xref ref-type="bibr" rid="CR32">32</xref>]. The model integrates highly detailed data on country-specific sociodemographic structures (for example, household size and composition, age structure, rates of school attendance, and so on) available from the Statistical Office of the European Commission [<xref ref-type="bibr" rid="CR33">33</xref>]. These data were used to generate highly detailed synthetic populations. More specifically, census data on frequencies of household size and type, and age of household components by size were used to group individuals into households. Data on rates of employment/inactivity and school attendance by age, structure of educational systems, school and workplace size allowed the assignment of individuals to schools and workplaces or their tagging as inactive, according to their age. Following the available estimates [<xref ref-type="bibr" rid="CR34">34</xref>–<xref ref-type="bibr" rid="CR37">37</xref>], the transmission model is parameterized so that 18% of transmission occurs through contacts made at school, 30% within households, 19% in workplaces and 33% in the general community. We made use of state of the art estimates of generation time for influenza viruses in the different settings [<xref ref-type="bibr" rid="CR38">38</xref>], namely age dependent Weibull distributions (see Additional file <xref rid="MOESM1" ref-type="media">1</xref> for details on the natural history of the virus) with a latent period of 1 day, consistent with estimates of generation time in the range 2.5 to 4 days [<xref ref-type="bibr" rid="CR23">23</xref>, <xref ref-type="bibr" rid="CR26">26</xref>–<xref ref-type="bibr" rid="CR29">29</xref>]. As it is nearly impossible to predict the reproduction number R<sub>0</sub> of a modified influenza strain (typical values for past influenza pandemics are in the range 1.3 to 2 [<xref ref-type="bibr" rid="CR23">23</xref>–<xref ref-type="bibr" rid="CR26">26</xref>, <xref ref-type="bibr" rid="CR39">39</xref>–<xref ref-type="bibr" rid="CR44">44</xref>]) we analyzed scenarios with R<sub>0</sub> varying from 1.1 to 2.5, accounting for the possible larger transmissibility of the modified virus with respect to past influenza viruses. The resulting doubling time of simulations without intervention is shown in Figure <xref rid="Fig1" ref-type="fig">1</xref>. We considered containment successful if the disease was eliminated in less than 5 months and resulted in less than 1,000 cumulative cases. The rationale for this choice is that, beyond the obvious requirement of disease elimination, epidemics should be characterized by a relatively low, socially acceptable, cumulative number of cases in a relatively short period of time; otherwise we speak of outbreak. See Additional file <xref rid="MOESM1" ref-type="media">1</xref> for methodological details.<fig id="Fig1"><label>Figure 1</label><caption><p> <bold>Doubling time.</bold> Average doubling time (dots) and 95% CI (vertical lines) as a function of R<sub>0</sub>. For each value of R<sub>0</sub> results were obtained by analyzing 100 uncontrolled (no intervention) simulated epidemics.</p></caption><graphic xlink:href="12916_2013_Article_878_Fig1_HTML" id="d29e518"/></fig> </p><p>Once the initial conditions for the outbreak were set the model generated stochastic ensemble estimates of the unfolding of the epidemic. The infection transmission chain can be analyzed at the level of each single individual and all the microscopic details of the progression of the epidemic in the population can be accessed for each stochastic realization of the escape event. The escape events were identically initialized in a BSL facility in the Netherlands (see Figure <xref rid="Fig2" ref-type="fig">2</xref>), by assuming 1 initial infected worker (among 50 to 150 workers; results obtained by assuming a different number of initial infections are analyzed in Additional file <xref rid="MOESM1" ref-type="media">1</xref>). This is a fundamental difference of the proposed method with respect to methods employed to estimate the probability of containing naturally emerging pathogens at the source or to analyze the potential effects of bioterrorist attacks: we assume to exactly know the starting point of the outbreak. A second key difference from other studies is the following: we assume that, if ascertained, initial infections generated by the first infected laboratory worker in the network of contacts comprising laboratory colleagues and laboratory workers’ household members may generate an initial warning, and a set of medical/epidemiological analyses are conducted very early to identify the origin of reported symptoms.<fig id="Fig2"><label>Figure 2</label><caption><p> <bold>Study area.</bold> The map shows population density of the Netherlands (colors from yellow to dark brown indicate increasing densities, from 1 to 3,500 inhabitants per km<sup>2</sup>), the location of the laboratory in a randomly chosen simulation (in Rotterdam, red point), the location of the workers houses (blue points), the location of workplaces and schools attended by household members of laboratory workers (green). Black concentric circles indicate distances of 10 km, 20 km, 30 km from the laboratory. The inset shows the probability of commuting to (at) a certain distance by laboratory workers.</p></caption><graphic xlink:href="12916_2013_Article_878_Fig2_HTML" id="d29e542"/></fig> </p><p>We assumed the warning to be issued at the time T<sub>w</sub> corresponding to the first identification of one of the initial cases. Two key parameters determine the efficacy of subsequent interventions: the first one is probability (P<sub>c</sub>) of identifying initial infections, which is related to the virus specific probability of developing clinical symptoms and the probability of individuals to be actually concerned and report their health status. The second one is the time (T<sub>i</sub>) required to link the initial infections to an accidental release of the modified influenza strain in the laboratory (and not, for instance, to other circulating seasonal influenza viruses) and to activate the containment interventions.</p><p>Once the PPP escape event has been detected we considered the following set of containment interventions: (i) isolation of the laboratory, (ii) laboratory workers’ household quarantine, (iii) contact tracing of cases and subsequent household quarantine of identified secondary cases, (iv) school and workplace closure both preventive, on a spatial basis, at the very beginning of the epidemic, and reactive during the entire epidemic.</p><p>For contact tracing, we assumed that once one case is detected, infected close contacts (that is household, school and workplace contacts) of the case are detected with probability P<sub>c</sub> and can transmit the infection for a certain time (T<sub>t</sub>) before isolation and household quarantine. Cases generated through random contacts in the general population are detected with lower probability (P<sub>g</sub>). We also assume that undetected cases may self-report their health status with a certain probability (P<sub>r</sub>). Parameters characterizing interventions along with reference values and explored ranges are described in Table <xref rid="Tab1" ref-type="table">1</xref> (see also Additional file <xref rid="MOESM1" ref-type="media">1</xref> for model details). Detailed descriptions of the contact tracing procedure and initial detection of the accidental release are shown in Figure <xref rid="Fig3" ref-type="fig">3</xref>A,B respectively. In the following we explore different implementations of the containment interventions and assess their effectiveness by generating stochastic scenario output (SSO) sets, providing for each point in space and time, as given by the resolution of the model, an ensemble of possible epidemic evolutions. We use as a benchmark SSO set the no intervention case, in which the epidemic is assumed to progress without external intervention and a reference SSO set where all the above containment measures are implemented according to the reference value reported in Table <xref rid="Tab1" ref-type="table">1</xref>.<table-wrap id="Tab1"><label>Table 1</label><caption><p> <bold>Model parameters regulating efficacy of interventions</bold> </p></caption><table frame="hsides" rules="groups"><thead><tr><th>Variable</th><th>Description</th><th>Reference (range)</th></tr></thead><tbody><tr><td>P<sub>c</sub> </td><td>Infected close contacts detection probability</td><td>0.6 (0.4 to 1)</td></tr><tr><td>P<sub>g</sub> </td><td>Infected random contacts detection probability</td><td>P<sub>c</sub> × 0.5 (0.1 to 1)</td></tr><tr><td>P<sub>r</sub> </td><td>Infected random contacts self-reporting probability</td><td>P<sub>g</sub> × 0.8 (0.5 to 1)</td></tr><tr><td>T<sub>i</sub> </td><td>Delay from initial warning to intervention</td><td>3 (0 to 30) days</td></tr><tr><td>T<sub>t</sub> </td><td>Delay from case detection to household quarantine</td><td>1 (0 to 4) days</td></tr><tr><td>T<sub>p</sub> </td><td>Duration of schools and workplaces closure</td><td>21 (0, 7, 14, 21, 28) days</td></tr><tr><td>D<sub>p</sub> </td><td>Radius for schools and workplaces closure</td><td>30 (0, 5, 10, 20, 30, 50, 50>) km</td></tr><tr><td>F<sub>s</sub> </td><td>Fraction of closed schools</td><td>0.9 (0 to 0.9)</td></tr><tr><td>F<sub>w</sub> </td><td>Fraction of closed workplaces</td><td>0 (0 to 0.5)</td></tr></tbody></table></table-wrap> <fig id="Fig3"><label>Figure 3</label><caption><p> <bold>Contact tracing. (A)</bold> Probabilities of detecting first and second generation cases (the latter conditioned to the detection of first generation cases) triggered by a traced index case. <bold>(B)</bold> Example of network of cases triggered by the initial infected laboratory worker (undetected in this example; the initial warning is triggered by a secondary case in the laboratory), and probability of case detection at time of intervention (T<sub>w</sub> + T<sub>i</sub>).</p></caption><graphic xlink:href="12916_2013_Article_878_Fig3_HTML" id="d29e754"/></fig> </p></sec><sec id="Sec3"><title>Results and discussion</title><p>Below we discuss the likelihood that the escape of PPP virus will spread into the local population and the ensuing outbreak will be contained by non-pharmaceuticals interventions that are likely the only ones to be available in the early stage of the outbreak.</p><sec id="Sec4"><title>Proportion of escape events that will trigger an outbreak</title><p>In order to set a baseline for our investigation it is worth stressing that there is a certain probability that the epidemic goes extinct without any intervention. In general, it is very difficult to estimate this probability, as it depends, beyond other factors, on seeding location (for example, urban vs rural) and contact network of the initial case. In our simulations, all these factors did not vary much as we simulated the initial epidemic seeding to occur always in a BSL facility in a populated area, thus drastically reducing the uncertainty of estimates. The probability of observing an epidemic outbreak in the absence of any interventions (no intervention scenario) is shown in Figure <xref rid="Fig4" ref-type="fig">4</xref>A, and it increases from about 25% for R<sub>0</sub> = 1.1 to values larger than 80% if R<sub>0</sub> >2.<fig id="Fig4"><label>Figure 4</label><caption><p> <bold>Reference scenario. (A)</bold> Probability of outbreak for different values of R<sub>0</sub> by assuming no intervention scenario (uncontrolled epidemics) and reference scenario. <bold>(B)</bold> Probability of undetected epidemics for different values of R<sub>0</sub> by assuming reference scenario (in red) and reference scenario with different values of P<sub>c</sub>. <bold>(C)</bold> Upper panel: overall number of cases in contained outbreaks by assuming reference scenario and R<sub>0</sub> = 1.5 (not considering autoextinct epidemics). Middle panel: as upper panel but for the number of traced cases. Lower panel: as upper panel but for the number of isolated individuals (including the laboratory’s contact network). A total of 1,000 simulations were undertaken for each parameter set to produce the results shown.</p></caption><graphic xlink:href="12916_2013_Article_878_Fig4_HTML" id="d29e805"/></fig> </p></sec><sec id="Sec5"><title>Proportion of undetected escape events</title><p>Notably, model simulations suggest that there is a non-negligible probability that the escape event is not detected at all. This may happen when no initial cases are detected among laboratory workers and laboratory workers’ household members, but secondary cases are generated through random contacts in the general population. In this case it is reasonable to assume that it is very difficult to ascertain the accidental release of a PPP from the BSL facility and to put in place timely control measures. As shown in Figure <xref rid="Fig4" ref-type="fig">4</xref>B, the probability of undetected epidemics increases with R<sub>0</sub> and it is strongly influenced by the probability of detecting cases. If R<sub>0</sub> >1.5, it may be as high as 5% when P<sub>c</sub> = 60% and 15% when P<sub>c</sub> = 40%. In general, the probability of case detection affects the outcome of intervention options. As we note, to a large extent the detection probability depends on the rate of asymptomatic cases and non-detectable transmissions. In the case of accidental release, the situation is even worse because the probability of detecting cases affects the probability of the timely implementation of the control and containment interventions. As shown in Additional file <xref rid="MOESM1" ref-type="media">1</xref>, this probability decreases and eventually vanishes when the number of initial cases is larger than 1.</p></sec><sec id="Sec6"><title>Controllability of the escape event</title><p>By assuming reference values for the parameters regulating the containment plan, the probability of observing an epidemic outbreak is drastically reduced for all values of R<sub>0</sub>. In particular, containment is likely to succeed for values of R<sub>0</sub> below 1.5 (probability of outbreak less than 10%, see Figure <xref rid="Fig4" ref-type="fig">4</xref>A). The SSO set indicates that for those values of R<sub>0</sub> the probability of outbreak is largely due to the probability of not detecting the outbreak itself; when the accidental release of the PPP agent is detected in a timely manner, outbreaks are contained with probability close to 100%. The resources required to contain epidemic outbreaks with reference intervention may vary considerably. As shown in Figure <xref rid="Fig4" ref-type="fig">4</xref>C, most epidemics are contained at the very beginning, when only few cases are present in the population (median: three infections), thus requiring little effort in terms of contact tracing (median: two traced cases) and overall number of quarantined households. However, it is possible, though not very likely, that containment requires the tracing of several cases (up to 58 traced cases for R<sub>0</sub> = 1.5, corresponding to the isolation of about 500 individuals). Even more demanding, especially from the social point of view, is the closure of 90% of schools for 21 days in a radius of 30 km around location of initial cases, as assumed by the reference SSO set. The number of cases observed can be easily related to the fatality associated to the outbreak if the case fatality rate (CFR) of the specific PPP agent is known. Unfortunately, the CFR is often not obviously correlated with the transmissibility of the pathogen. In addition, it is extremely difficult to obtain reliable estimates of the CFR during the early stage of an outbreak. A sensitivity analysis of the fatality of the virus can however be performed by applying plausible CFR to the number of cases observed with our approach.</p><p>The timeline of simulated epidemics with R<sub>0</sub> = 1.5 is shown in Figure <xref rid="Fig5" ref-type="fig">5</xref>. Autoextinction occurs in very few days (maximum 57 days) after only few cumulative cases (maximum 10 to 20 cases). A similar pattern is observed for contained epidemics, which may be characterized by a slightly longer duration (maximum 100 days) and slightly larger number of cases (maximum 100 to 200). In both cases, incidence is always less than 20 daily cases. Uncontained epidemics result in long-lasting epidemics (more than 1 year) and produce a large number of cases in a short period of time (larger than 10,000 in 5 months; peak incidence between 10,000 and 15,000 daily cases). Undetected epidemics are shorter (less than 1 year) but are characterized by a much larger number of cases (overall attack rate: 49.5% on average) and peak incidence (between 200,000 and 300,000 daily cases). In addition, these results show the mitigation efficacy of the proposed interventions (specifically household quarantine and reactive school closure on the basis of contact tracing procedures). Moreover, as only 2 different patterns may occur (either the disease quickly dies out after a very limited number of cases or it results in an epidemic outbreak, with many cases in the very first days), these results justify our definition of contained epidemic (disease elimination in less than 5 months and less than 1,000 cumulative cases), though many others are of course equivalent.<fig id="Fig5"><label>Figure 5</label><caption><p> <bold>Epidemic timing. (A)</bold> Average number of daily cases as observed in autoextinct simulated epidemics (red points) with R<sub>0</sub> = 1.5. Vertical lines represent minimum and maximum daily incidence. <bold>(B)</bold> As in <bold>(A)</bold> but for the average cumulative number of cases. <bold>(C,D)</bold> As <bold>(A)</bold> and <bold>(B)</bold> but for contained epidemics by assuming reference interventions. <bold>(E,F)</bold> As <bold>(A)</bold> and <bold>(B)</bold> but for uncontained epidemics by assuming reference interventions. <bold>(G,H)</bold> As <bold>(A)</bold> and <bold>(B)</bold> but for undetected epidemics. A total of 1,000 simulations were undertaken to produce the results shown.</p></caption><graphic xlink:href="12916_2013_Article_878_Fig5_HTML" id="d29e912"/></fig> </p></sec><sec id="Sec7"><title>Sensitivity analysis of containment policies</title><p>Results are very sensitive to most of the parameters describing intervention options. By restricting our analysis to parameters regulating contact tracing (thus excluding self-reporting of cases and preventive closure of schools and workplaces) we found that the probability of detecting infections among close contacts of cases and time from initial warning to interventions are the two most critical variables (see Additional file <xref rid="MOESM1" ref-type="media">1</xref> for details). Figure <xref rid="Fig6" ref-type="fig">6</xref>A shows sensitivity of results obtained by assuming reference parameters but varying the values of these two parameters. For low values of R<sub>0</sub> containment is very likely to succeed when P<sub>c</sub> is larger than 60% (for R<sub>0</sub> = 1.2) or 80% (for R<sub>0</sub> = 1.5) even when the delay from initial warning to interventions (T<sub>i</sub>) is much larger than the one assumed by reference simulations (up to 30 or 10 days for R<sub>0</sub> = 1.2 and 1.5 respectively). For larger values of R<sub>0</sub>, containment is feasible only when P<sub>c</sub> is larger than 60% and T<sub>i</sub> is no larger than 3 to 5 days. Figure <xref rid="Fig6" ref-type="fig">6</xref>B,C show that other parameters regulating contact tracing can play an important role. In particular, Figure <xref rid="Fig6" ref-type="fig">6</xref>B shows that a timely intervention during contact tracing is very critical and Figure <xref rid="Fig6" ref-type="fig">6</xref>C shows that it may be important to identify a high number of contacts infected in the general population. This may be difficult in practice but it might be a critical factor for the successful containment. If contact tracing allows the identification of cases in the general community with approximately the same probability of identifying secondary cases in household, school and workplaces, epidemic outbreaks with R<sub>0</sub> up to 1.6 to 1.7 could be reasonably expected to be contained.<fig id="Fig6"><label>Figure 6</label><caption><p> <bold>Sensitivity analysis: contact tracing. (A)</bold> Probability (×100) of outbreak for different values of R<sub>0</sub> by assuming reference scenario and by varying T<sub>i</sub> and P<sub>c</sub>. <bold>(B)</bold> Probability of outbreak for different values of R<sub>0</sub> by assuming no intervention scenario, reference scenario, and reference scenarios with different delays in the isolation of traced cases. <bold>(C)</bold> Probability of outbreak for different values of R<sub>0</sub> by assuming no intervention scenario, reference scenario, and reference scenarios with different probabilities of identifying cases in the general community. A total of 1,000 simulations were undertaken for each parameter set to produce the results shown.</p></caption><graphic xlink:href="12916_2013_Article_878_Fig6_HTML" id="d29e999"/></fig> </p></sec><sec id="Sec8"><title>Effectiveness of preventive school and workplace closure</title><p>Figure <xref rid="Fig7" ref-type="fig">7</xref>B shows that closure of schools (with probability 90%) may be relevant while the additional closure of workplaces (with probability 50%) may be relevant only to decrease the outbreak probability when R<sub>0</sub> is larger than 1.4. Figure <xref rid="Fig7" ref-type="fig">7</xref>A shows that distance for spatial closure of places and duration of closure are irrelevant when R<sub>0</sub> is 1.2 (as the overall impact of the strategy is not very relevant), while for values of R<sub>0</sub> = 1.5 or larger, model simulations show that, as expected, the longer the duration and the greater the distance are the lower the probability of outbreak is: duration of 21 days and distance of 30 km represent a good compromise between feasibility and impact. A distance of 30 km for spatially targeted interventions is remarkably larger than that considered in [<xref ref-type="bibr" rid="CR25">25</xref>] for containing an epidemic in Thailand. This can be explained by looking at the different human mobility patterns in Thailand, where most of commuting is within 5 km, and the Netherlands, where commutes of 10 to 30 km to go to work or school are common [<xref ref-type="bibr" rid="CR45">45</xref>] (see inset of Figure <xref rid="Fig2" ref-type="fig">2</xref> and Additional file <xref rid="MOESM1" ref-type="media">1</xref> for details).<fig id="Fig7"><label>Figure 7</label><caption><p> <bold>Sensitivity analysis: school and workplace closure. (A)</bold> Probability (×100) of outbreak for different values of R<sub>0</sub> by assuming reference scenario with additional workplaces closure (F<sub>w</sub> = 0.5) and by varying D<sub>p</sub> and T<sub>p</sub>. <bold>(B)</bold> Probability of outbreak for different values of R<sub>0</sub> by assuming no intervention scenario, reference scenario, and reference scenarios with different policies regulating school and workplaces closure. A total of 1,000 simulations were undertaken for each parameter set to produce the results shown.</p></caption><graphic xlink:href="12916_2013_Article_878_Fig7_HTML" id="d29e1064"/></fig> </p></sec><sec id="Sec9"><title>Geographical context analysis</title><p>The probability of containing an epidemic outbreak may also depend on the BSL laboratory location and the sociodemographic structure of the population. This is shown in Figure <xref rid="Fig8" ref-type="fig">8</xref>A where we compare results obtained for Rotterdam (The Netherlands) with those obtained by simulating the epidemic spread emerging from BSL facilities in other urban areas of Europe. We found that Rotterdam likely represents the best case scenario among those analyzed in this paper: for instance, the probability of observing an epidemic outbreak in Paris, by assuming reference interventions, may be 200% to 300% larger than that estimated for the Netherlands if R<sub>0</sub> <1.5. Differences reduce drastically for larger values of R<sub>0</sub>. Without considering control measures, the probability of observing an epidemic outbreak after virus escape is quite similar to that in the Dutch scenario: slight differences can be observed for low values of R<sub>0</sub>. Such large differences may be due to dissimilarities in sociodemographic characteristics of French and Dutch populations because, despite a general similarity, some marked country-specific features such as age structure and average household size exist. However, although quantitatively different, the general patterns obtained by varying P<sub>c</sub> and T<sub>i</sub> are the same observed in the Dutch case. Detailed results for Paris are discussed in Additional file <xref rid="MOESM1" ref-type="media">1</xref>. We also found that the probability of observing an epidemic outbreak when the BSL laboratory is located in a rural region is systematically lower than that estimated for urban areas (see Figure <xref rid="Fig8" ref-type="fig">8</xref>B). For instance, given a BSL facility located in the UK, we found that the probability of epidemic outbreak when the pathogen is accidentally released from a hypothetical BSL laboratory in Wales (UK) may be three to five times lower than that estimated for a BSL laboratory in London if R<sub>0</sub> <1.5. These differences are ascribable to differences in population density and sociodemographic structure, as discussed in [<xref ref-type="bibr" rid="CR31">31</xref>]. These results are discussed in detail in Additional file <xref rid="MOESM1" ref-type="media">1</xref>.<fig id="Fig8"><label>Figure 8</label><caption><p> <bold>Geographical variability. (A)</bold> Ratio between probability of outbreak in different urban areas and probability of outbreak in Rotterdam (The Netherlands) for different values of R<sub>0</sub> by assuming reference scenario. <bold>(B)</bold> Ratio between probability of outbreak in urban and rural areas in different countries for different values of R<sub>0</sub> by assuming reference scenario. Urban areas as in <bold>(A)</bold>; rural areas are low population density areas in Wales (UK, 80 km north of Cardiff), Uppland (SE, 100 km north of Uppsala), Sardinia island (IT, 50 km east of Sassari), Andalusia - Castile la Mancha (ES, 50 km northeast of Cordoba), Centre-Burgundy (France, 80 km southeast of Orleans). Note that the reported value of R<sub>0</sub> refers to that of simulations carried out for Rotterdam (The Netherlands); comparative results for other countries are obtained by assuming the same transmission rates in the different social contexts (that is the same probability of infection transmission given a contact in a certain setting) as in Rotterdam. A total of 1,000 simulations were undertaken for each parameter set to produce the results shown.</p></caption><graphic xlink:href="12916_2013_Article_878_Fig8_HTML" id="d29e1133"/></fig> </p></sec><sec id="Sec10"><title>Impact of additional intervention</title><p>We found that results are not very sensitive to the probability of self-reporting (P<sub>r</sub>) and to the initial set of interventions on the initial network of contacts comprising laboratory workers and laboratory workers’ household members. The reference scenario assumes the closure of the laboratory and the quarantine of the households of laboratory workers. We explored the possibility of extending these interventions and to preventively close all workplaces and schools attended by relatives of laboratory workers. We found that closing the laboratory is the only intervention leading to a certain reduction of the outbreak probability. Additional interventions are of little impact. We report on these findings in Additional file <xref rid="MOESM1" ref-type="media">1</xref>.</p></sec></sec><sec id="Sec11"><title>Conclusions</title><p>Our results suggest that containment is likely to succeed by employing social distancing measures only if R<sub>0</sub> is no larger than 1.5. Containment could be feasible even for larger values of R<sub>0</sub> in cases of very timely intervention both in recognizing the accidental release and during contact tracing and high probability of detecting secondary cases in the same household, school or workplace as a newly identified case. Overall, these results suggest that success in containing an accidentally released potentially pandemic influenza virus by employing social distancing measures only is uncertain: containment probability for a virus with transmissibility comparable to many of the estimates for the 2009 H1N1 virus (namely R<sub>0</sub> or effective transmissibility in the range 1.2 to 1.6 [<xref ref-type="bibr" rid="CR19">19</xref>–<xref ref-type="bibr" rid="CR24">24</xref>]) is reassuring, even though containment is not guaranteed. Should the transmissibility of the pathogen be comparable to that of the 1918 Spanish influenza (R<sub>0</sub> = 1.8 or higher [<xref ref-type="bibr" rid="CR25">25</xref>]), containment success would be seriously compromised. A further relevant finding is the strong impact of the BSL laboratory location. Rural areas have a fivefold increase in containment probability with respect to densely populated urban areas. Similarly, we observe differences according to the sociodemographic structure of the geographical region. These results provide data with potential use in defining policies for deciding the most appropriate location of BSL laboratories.</p><p>Our simulations do not account for the possible use of pharmaceutical interventions. While the availability of an effective vaccine is highly questionable in case of accidental release of genetically manipulated influenza viruses from BSL facilities, the use of antivirals at the very beginning of the epidemic is an option that could be considered. If used for treatment of cases and prophylaxis of close contacts (for example, household and school contacts) only, however, the benefit should not be very different from that obtained by assuming household quarantine and reactive school closure, as this paper does. Moreover, it requires a timely administration (within 2 days from symptoms onset [<xref ref-type="bibr" rid="CR25">25</xref>, <xref ref-type="bibr" rid="CR46">46</xref>–<xref ref-type="bibr" rid="CR50">50</xref>]) to be effective. Geographical targeting of a large fraction of the population is a completely different option that could be considered: on the one hand it could lead to drastically increasing the probability of containment but on the other hand also poses serious logistical challenges [<xref ref-type="bibr" rid="CR25">25</xref>, <xref ref-type="bibr" rid="CR47">47</xref>].</p><p>The preventive immunization of laboratory workers (see for instance the Special Immunization Program in the US [<xref ref-type="bibr" rid="CR51">51</xref>]) is another option not considered in this work. Although for diseases for which a vaccine is available this is a measure to take into account (for instance, the incidence of hepatitis B virus (HBV) infection among laboratory workers in the UK has significantly dropped because of the availability of immunization [<xref ref-type="bibr" rid="CR52">52</xref>]), this measure is highly questionable for genetically modified influenza viruses, not to speak of influenza viruses for which no vaccine is currently available, for example, A(H7N9).</p><p>In summary, our results suggest that public health authorities should be prepared to put in place a set of social distancing interventions, for example, contact tracing and closure of schools and workplaces on a geographical basis. Moreover, as it is nearly impossible to get accurate estimates of R<sub>0</sub> (as well as case fatality rate) for a new virus at the very beginning of the outbreak, in order to maximally reduce the risk of a global pandemic the possibility of timely targeting a large fraction of the population with antivirals (as a prophylactic measure on a geographical basis) or establishing quarantine areas should not be set aside, even though this calls for the development of detailed intervention plans and requires public health agencies to put in place containment efforts hardly achievable in most places in the world. Where the pandemic pathogens are concerned, short generation time and asymptomaticity are among the most critical factors that make accidental release of influenza viruses difficult to contain.</p><p>Qualitatively, the results do not vary much by considering different seeding locations. However, containment probabilities are affected by several factors, including population density and sociodemographic structure. These findings may have an important impact on policies: our results strongly suggest the location of new BSL facilities worldwide should be carefully chosen, for instance with priority given to rural areas and, when this is not feasible, by taking into account density and structure of the population in urban areas. This may make the difference, especially for pathogens with low to moderate transmissibility. Of course, these decisions should also be based on other factors not considered in this study, for example, population vulnerability to infectious agents, risk factors, structure of the health system, possibility of putting in place a rapid response program. Simulated scenarios emerging from detailed models such as the one presented here may inform quantitatively the process of identifying locations that minimize risk. Finally, it is worth remarking that the presented approach can be generally extended to other pathogens that can be classified as dual use research of concern if we have the appropriate information on the pathogens, mechanism of transmission and natural history of the disease.</p></sec><sec sec-type="supplementary-material"><title>Electronic supplementary material</title><sec id="Sec12"><p> <supplementary-material content-type="local-data" id="MOESM1"><media xlink:href="12916_2013_878_MOESM1_ESM.pdf"><caption><p>Additional file 1: Supporting material. Model details, additional results. (PDF 359 KB)</p></caption></media></supplementary-material> </p></sec></sec>
Management of HIV-associated tuberculosis in resource-limited settings: a state-of-the-art review	<p>The HIV-associated tuberculosis (TB) epidemic remains a huge challenge to public health in resource-limited settings. Reducing the nearly 0.5 million deaths that result each year has been identified as a key priority. Major progress has been made over the past 10 years in defining appropriate strategies and policy guidelines for early diagnosis and effective case management. Ascertainment of cases has been improved through a twofold strategy of provider-initiated HIV testing and counseling in TB patients and intensified TB case finding among those living with HIV. Outcomes of rifampicin-based TB treatment are greatly enhanced by concurrent co-trimoxazole prophylaxis and antiretroviral therapy (ART). ART reduces mortality across a spectrum of CD4 counts and randomized controlled trials have defined the optimum time to start ART. Good outcomes can be achieved when combining TB treatment with first-line ART, but use with second-line ART remains challenging due to pharmacokinetic drug interactions and cotoxicity. We review the frequency and spectrum of adverse drug reactions and immune reconstitution inflammatory syndrome (IRIS) resulting from combined treatment, and highlight the challenges of managing HIV-associated drug-resistant TB.</p>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Lawn</surname><given-names>Stephen D</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><xref ref-type="aff" rid="I3">3</xref><email>stephen.lawn@lshtm.ac.uk</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Meintjes</surname><given-names>Graeme</given-names></name><xref ref-type="aff" rid="I3">3</xref><xref ref-type="aff" rid="I4">4</xref><email>Graeme.Meintjes@uct.ac.za</email></contrib><contrib contrib-type="author" id="A3"><name><surname>McIlleron</surname><given-names>Helen</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>helen.mcilleron@uct.ac.za</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Harries</surname><given-names>Anthony D</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I6">6</xref><email>adharries@theunion.org</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Wood</surname><given-names>Robin</given-names></name><xref ref-type="aff" rid="I2">2</xref><xref ref-type="aff" rid="I3">3</xref><email>Robin.Wood@hiv-research.org.za</email></contrib>	BMC Medicine	<sec sec-type="intro"><title>Introduction</title><p>The global epidemics of HIV/AIDS and tuberculosis (TB) both remain huge challenges to international public health, causing illness and death in millions of people worldwide each year (Table <xref ref-type="table" rid="T1">1</xref>) [<xref ref-type="bibr" rid="B1">1</xref>]. TB is the most important AIDS-related opportunistic disease globally and is the leading cause of HIV/AIDS-related mortality, accounting for an estimated 25% of such deaths [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B3">3</xref>]. Sub-Saharan Africa suffers disproportionately, with 79% of global cases of HIV-associated TB [<xref ref-type="bibr" rid="B1">1</xref>]. In the countries of southern and eastern Africa where HIV prevalence is highest, the impact of HIV has severely undermined TB control over the past 20 years [<xref ref-type="bibr" rid="B4">4</xref>]. The global co-epidemic has been further compounded in recent years by the emergence of the growing challenge of multi-drug resistant TB (MDR-TB) [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B6">6</xref>].</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Burden of HIV infection, tuberculosis (TB) and HIV-associated TB globally and in sub-Saharan Africa</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="left"><bold>Disease</bold></th><th align="center"><bold>Global burden</bold></th><th align="center"><bold>Burden in sub-Saharan Africa: (% of global burden)</bold></th></tr></thead><tbody valign="top"><tr><td colspan="3" align="left" valign="bottom"><bold>HIV/AIDS</bold>:<hr/></td></tr><tr><td align="left" valign="bottom">No. of people living with HIV infection<hr/></td><td align="center" valign="bottom">34,200,000<hr/></td><td align="center" valign="bottom">23,500,000 (69%)<hr/></td></tr><tr><td align="left" valign="bottom">HIV/AIDS-related deaths<hr/></td><td align="center" valign="bottom">1,700,000<hr/></td><td align="center" valign="bottom">1,200,000 (71%)<hr/></td></tr><tr><td colspan="3" align="left" valign="bottom"><bold>Tuberculosis</bold>:<hr/></td></tr><tr><td align="left" valign="bottom">No. of incident cases of TB<hr/></td><td align="center" valign="bottom">8,700,000<hr/></td><td align="center" valign="bottom">2,300,000 (26%)<hr/></td></tr><tr><td align="left" valign="bottom">TB deaths (excluding HIV)<hr/></td><td align="center" valign="bottom">990,000<hr/></td><td align="center" valign="bottom">220,000 (22%)<hr/></td></tr><tr><td align="left" valign="bottom">Incident cases of multidrug-resistant TB<hr/></td><td align="center" valign="bottom">310,000<hr/></td><td align="center" valign="bottom">45,000 (15%)<hr/></td></tr><tr><td colspan="3" align="left" valign="bottom"><bold>HIV-</bold>a<bold>ssociated</bold> t<bold>uberculosis</bold>:<hr/></td></tr><tr><td align="left" valign="bottom">No. of incident cases<hr/></td><td align="center" valign="bottom">1,100,000<hr/></td><td align="center" valign="bottom">870,000 (79%)<hr/></td></tr><tr><td align="left">No. of HIV-associated TB deaths</td><td align="center">430,000</td><td align="center">300,000 (70%)</td></tr></tbody></table><table-wrap-foot><p>Data from [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B3">3</xref>]. Incident disease and deaths represent annual disease burden.</p></table-wrap-foot></table-wrap><p>The World Health Organization (WHO) DOTS (directly observed treatment, short-course) TB control strategy used in isolation provides far from optimum case management for individual patients with HIV-associated TB and it has failed to control TB at a population level in settings with high HIV prevalence [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B7">7</xref>]. Comprehensive packages of additional interventions are needed to address the consequences of HIV in TB patients and to reduce the burden of TB in those living with HIV infection [<xref ref-type="bibr" rid="B8">8</xref>]. An interim policy on collaborative TB/HIV activities was first published by WHO in 2004 [<xref ref-type="bibr" rid="B9">9</xref>] and approximately 1.3 million lives are estimated to have been saved by these interventions by 2011 [<xref ref-type="bibr" rid="B1">1</xref>]. An updated policy (Table <xref ref-type="table" rid="T2">2</xref>) [<xref ref-type="bibr" rid="B10">10</xref>] published in 2012 provides the overall policy framework for addressing HIV-associated TB and specific recommendations on management of HIV, TB and multidrug-resistant (MDR)-TB are provided by individual guideline documents [<xref ref-type="bibr" rid="B11">11</xref>-<xref ref-type="bibr" rid="B13">13</xref>] (Table <xref ref-type="table" rid="T3">3</xref>).</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p><bold>World Health Organization (WHO)-recommended collaborative tuberculosis (TB)/HIV activities (adapted from</bold>[<xref ref-type="bibr" rid="B10">10</xref>]<bold>)</bold></p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Key area</bold></th><th align="left"><bold>Points of action</bold></th></tr></thead><tbody valign="top"><tr><td rowspan="4" align="left" valign="bottom"><bold>Establish and strengthen the mechanisms for delivering integrated TB and HIV services</bold><hr/></td><td align="left" valign="bottom">Set up and strengthen a coordinating body for collaborative TB/HIV activities<hr/></td></tr><tr><td align="left" valign="bottom">Determine the HIV prevalence among TB patients and the TB prevalence among HIV patients<hr/></td></tr><tr><td align="left" valign="bottom">Carry out joint TB/HIV planning to integrate the delivery of TB and HIV services<hr/></td></tr><tr><td align="left" valign="bottom">Monitor and evaluate collaborative TB/HIV activities<hr/></td></tr><tr><td rowspan="3" align="left" valign="bottom"><bold>Reduce the burden of TB in people living with HIV (early ART plus the three Is)</bold><hr/></td><td align="left" valign="bottom">Intensify TB case finding and ensure high quality TB treatment<hr/></td></tr><tr><td align="left" valign="bottom">Initiate TB prevention using isoniazid preventive therapy and early antiretroviral therapy (ART)<hr/></td></tr><tr><td align="left" valign="bottom">Ensure control of TB infection in healthcare facilities and congregate settings<hr/></td></tr><tr><td rowspan="4" align="left" valign="bottom"><bold>Reduce the burden of HIV in patients with diagnosed TB and those under investigation for TB</bold><hr/></td><td align="left" valign="bottom">Provide HIV testing and counseling to both groups of patients<hr/></td></tr><tr><td align="left" valign="bottom">Provide HIV preventive interventions to both groups of patients<hr/></td></tr><tr><td align="left" valign="bottom">Provide co-trimoxazole preventive therapy for TB patients living with HIV<hr/></td></tr><tr><td align="left" valign="bottom">Provide HIV prevention interventions, treatment and care for TB patients living with HIV<hr/></td></tr><tr><td align="left"> </td><td align="left">Provide antiretroviral therapy for TB patients living with HIV</td></tr></tbody></table></table-wrap><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>World Health Organization (WHO) policy guidelines on collaborative tuberculosis (TB)/HIV activities and the management of HIV infection, TB and multidrug-resistant TB (MDR-TB)</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="center"/></colgroup><thead valign="top"><tr><th align="left"><bold>Guidelines/year</bold></th><th align="left"><bold>Details</bold></th><th align="center"><bold>Reference</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom"><bold>Guidelines for collaborative TB/HIV activities (2012)</bold><hr/></td><td align="left" valign="bottom">World Health Organization. WHO policy on collaborative TB/HIV activities. Guidelines for national programmes and stakeholders. 2012. World Health Organization, Geneva. WHO/HTM/TB/2012.1. <ext-link ext-link-type="uri" xlink:href="http://whqlibdoc.who.int/publications/2012/9789241503006_eng.pdf">http://whqlibdoc.who.int/publications/2012/9789241503006_eng.pdf</ext-link><hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B10">10</xref>]<hr/></td></tr><tr><td align="left" valign="bottom"><bold>Antiretroviral treatment guidelines (2013)</bold><hr/></td><td align="left" valign="bottom">World Health Organization. Consolidated guidelines on the use of antiretroviral drugs for treating and preventing HIV infection. Recommendations for a public health approach, June 2013. WHO, Geneva. Accessible at: <ext-link ext-link-type="uri" xlink:href="http://apps.who.int/iris/bitstream/10665/85321/1/9789241505727_eng.pdf">http://apps.who.int/iris/bitstream/10665/85321/1/9789241505727_eng.pdf</ext-link>.<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B11">11</xref>]<hr/></td></tr><tr><td align="left" valign="bottom"><bold>Tuberculosis treatment guidelines (2010)</bold><hr/></td><td align="left" valign="bottom">World Health Organization. Treatment of tuberculosis: guidelines - fourth edition. World Health Organization, Geneva, 2010. WHO/HTM/TB/2009.420 Accessible at: <ext-link ext-link-type="uri" xlink:href="http://whqlibdoc.who.int/publications/2010/9789241547833_eng.pdf">http://whqlibdoc.who.int/publications/2010/9789241547833_eng.pdf</ext-link><hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B12">12</xref>]<hr/></td></tr><tr><td align="left"><bold>Drug-resistant TB treatment guidelines (2011)</bold></td><td align="left">World Health Organization. Guidelines for the management of drug-resistant tuberculosis: 2011 update. WHO, Geneva. WHO/HTM/TB/2011.6. Accessible at: <ext-link ext-link-type="uri" xlink:href="http://whqlibdoc.who.int/publications/2011/9789241501583_eng.pdf">http://whqlibdoc.who.int/publications/2011/9789241501583_eng.pdf</ext-link>.</td><td align="center">[<xref ref-type="bibr" rid="B13">13</xref>]</td></tr></tbody></table></table-wrap><p>This article provides an up-to-date review of the current medical management of adult patients with HIV-associated TB. We review case ascertainment as the critical first step and then how clinical outcomes can be optimized by provision of effective TB treatment, use of concurrent ART, prevention of HIV-related comorbidities and management of drug cotoxicity and immune reconstitution inflammatory syndrome (IRIS). We also describe the management of HIV-associated MDR-TB. However, the management of children, models of integrated TB and HIV care delivery and prevention of TB in people living with HIV using ART and isoniazid preventive therapy lie outside the scope of this review.</p><sec><title>Diagnosis of HIV-associated TB</title><p>The prerequisite for optimum management of HIV-associated TB is early and accurate diagnosis and, for many years, this has been a key obstacle. Case ascertainment can be greatly improved by high rates of quality-assured HIV testing among those being investigated for TB as well as high rates of screening for TB in those living with HIV.</p><sec><title>Screening for TB in those living with HIV infection</title><p>In high burden settings, much prevalent TB disease remains ‘below the radar’ in those living with HIV. Postmortem studies conducted in hospitals across sub-Saharan Africa over the past 20 years have repeatedly shown that between 30% and 50% of HIV-infected adult inpatients who die have postmortem evidence of TB, much of which was neither clinically suspected nor diagnosed before death [<xref ref-type="bibr" rid="B14">14</xref>-<xref ref-type="bibr" rid="B17">17</xref>]. These studies have highlighted the abject failure of the diagnostic process and the low sensitivity of diagnostic tools available [<xref ref-type="bibr" rid="B18">18</xref>]. In the absence of more sensitive means of diagnosis, management algorithms for suspected sputum smear-negative disease were developed [<xref ref-type="bibr" rid="B19">19</xref>-<xref ref-type="bibr" rid="B21">21</xref>] and studies of empirical TB treatment for certain high risk patient groups with advanced immunodeficiency are being conducted [<xref ref-type="bibr" rid="B22">22</xref>].</p><p>However, in recent years there have been significant advances in screening and diagnosis. Traditional symptom screening for pulmonary TB based on chronic cough has low sensitivity for HIV-associated TB [<xref ref-type="bibr" rid="B23">23</xref>,<xref ref-type="bibr" rid="B24">24</xref>]. A new WHO symptom screening tool for HIV-associated TB (one or more of the following symptoms: cough, fever, weight loss or night sweats, each of any magnitude or duration) has much higher sensitivity and is recommended for routine screening of those in HIV care at each visit [<xref ref-type="bibr" rid="B25">25</xref>]. However, in view of its low specificity, further research is needed to define which of the large number of patients with a positive screen should be prioritized for subsequent microbiological testing of clinical samples.</p><p>New diagnostic tools have also increased our capacity for microbiological diagnosis. This includes the Xpert MTB/RIF assay, which was endorsed by WHO in 2010. A single test is able to detect all sputum smear-positive disease, approximately 70% of smear-negative pulmonary disease and provides rapid simultaneous screening for RIF resistance [<xref ref-type="bibr" rid="B26">26</xref>]. In addition, this assay can be used to test a wide range of extrapulmonary sample types [<xref ref-type="bibr" rid="B26">26</xref>,<xref ref-type="bibr" rid="B27">27</xref>]. The Xpert MTB/RIF assay has been incorporated into the national guidelines of many high burden countries. In South Africa, which alone accounts for approximately 30% of the global burden of HIV-associated TB, sputum smear microscopy has now been replaced by Xpert MTB/RIF as the initial diagnostic test for TB [<xref ref-type="bibr" rid="B26">26</xref>].</p><p>Determine TB-LAM is a low-cost, point-of-care lateral-flow (‘strip test’) assay that diagnoses TB through detection in urine of lipoarabinomannan (LAM): a lipopolysaccharide component of the <italic>M. tuberculosis</italic> cell wall [<xref ref-type="bibr" rid="B28">28</xref>]. It has high specificity whereas sensitivity is very strongly CD4 count dependent, at best detecting approximately two-thirds of cases in those with CD4 counts <50 cells/μl [<xref ref-type="bibr" rid="B28">28</xref>-<xref ref-type="bibr" rid="B31">31</xref>]. This assay therefore allows rapid (<30 minutes) bedside diagnosis among those who have the highest mortality risk [<xref ref-type="bibr" rid="B32">32</xref>]. The growing evidence base on this assay will be reviewed by WHO in 2014. Its role is likely to be as an add-on test within the diagnostic algorithm to permit point-of-care diagnosis and immediate TB treatment among patients with advanced immunodeficiency (CD4 counts <200 cells/μl) following admission to hospital or enrolling in ART clinics [<xref ref-type="bibr" rid="B28">28</xref>,<xref ref-type="bibr" rid="B31">31</xref>].</p></sec><sec><title>Screening for HIV in those with TB or possible TB</title><p>A major step forward in improving HIV testing rates in patients with TB was the switch from voluntary counseling and testing (VCT) to provider-initiated testing and counseling (PITC) in 2007 [<xref ref-type="bibr" rid="B33">33</xref>]. With PITC, all patients undergo routine testing unless they specifically opt out. Testing has increased globally from 3.1% in 2004 to 40% of notified TB cases in 2011, but falls well short of the goal of universal testing [<xref ref-type="bibr" rid="B1">1</xref>]. Testing rates have reached 69% in Africa, >50% in the Americas and 32% in South-East Asia. In African countries, the proportion of TB patients testing positive is 46% overall (range, 8% to 77%) and exceeds 50% in ten counties in the south and east of the continent [<xref ref-type="bibr" rid="B1">1</xref>]. A further significant policy change has been to expand PITC to include all patients being investigated for TB regardless of whether or not TB is diagnosed [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B12">12</xref>]. This change resulted from the observed high HIV prevalence and mortality among those presenting for investigation of possible TB even when this diagnosis was subsequently excluded [<xref ref-type="bibr" rid="B34">34</xref>]. It is critical, however, that improved testing rates are accompanied by improvement in the delivery of appropriate management.</p></sec></sec><sec><title>Optimized TB treatment</title><p>The first priority for patients with HIV-associated TB is to immediately start effective TB treatment using a regimen containing RIF throughout [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B35">35</xref>]. A systematic review found that the incidence of relapse and/or failure among patients treated with intermittent (thrice weekly) TB therapy throughout was two to three times higher than that in patients who received a daily intensive phase [<xref ref-type="bibr" rid="B36">36</xref>]. Thus, the recommended optimum standard regimen is 2 months of rifampicin, isoniazid, pyrazinamide and ethambutol followed by 4 months of rifampicin and isoniazid (2HRZE/4HR), with therapy administered daily throughout [<xref ref-type="bibr" rid="B12">12</xref>]. Where this is not possible, an acceptable alternative is to use a thrice-weekly continuation phase. Treatment outcomes are worse for those with isoniazid monoresistance [<xref ref-type="bibr" rid="B36">36</xref>,<xref ref-type="bibr" rid="B37">37</xref>] and, thus, in settings with high prevalence of isoniazid monoresistance, 2HRZE/4HRE is the recommended first-line regimen [<xref ref-type="bibr" rid="B12">12</xref>]. Drug susceptibility testing is recommended to guide treatment in patients who have previously been treated for TB, although ideally all patients with TB should have drug susceptibility testing. Where the Xpert MTB/RIF assay is being rolled out as the primary TB diagnostic test, RIF resistance screening is now integral to the initial diagnostic process [<xref ref-type="bibr" rid="B26">26</xref>].</p><p>After several decades with no new advances in TB treatment, there are now some promising developments. For example, several large-scale phase III randomized controlled trials (including the ReMOX, Oflotub and RIFAQUIN studies) are evaluating whether incorporation of a newer fluoroquinolone into treatment regimens can be used to shorten treatment for drug susceptible TB [<xref ref-type="bibr" rid="B38">38</xref>]. The first of these to report, the RIFAQUIN study, found treatment shortening was associated with a higher rate of adverse outcomes including failure, relapse and death [<xref ref-type="bibr" rid="B39">39</xref>]. However, none of these studies have been designed to specifically address this question in HIV-infected clinical populations. There is also a growing developmental pipeline of new TB drugs, although these are most likely to be used in the treatment of MDR-TB, at least initially [<xref ref-type="bibr" rid="B38">38</xref>].</p></sec><sec><title>Co-trimoxazole preventive therapy</title><p>Co-trimoxazole (trimethoprim sulfamethoxazole) is a low-cost, widely available and relatively safe antibiotic that reduces morbidity and mortality in people living with HIV due to prophylactic activity against a range of pathogens, including those causing bacterial sepsis, pneumocystis pneumonia, cerebral toxoplasmosis and malaria. Both observational and randomized controlled trials conducted in sub-Saharan Africa have shown that this simple intervention is associated with a substantial reduction in mortality among patients with HIV-associated TB (range, 19% to 46%) [<xref ref-type="bibr" rid="B40">40</xref>-<xref ref-type="bibr" rid="B44">44</xref>] (Table <xref ref-type="table" rid="T4">4</xref>). This beneficial effect was observed in a range of settings with high or low rates of bacterial resistance to the drug and is additive in reducing early mortality when combined with ART [<xref ref-type="bibr" rid="B45">45</xref>].</p><table-wrap position="float" id="T4"><label>Table 4</label><caption><p>Impact of co-trimoxazole prophylaxis on mortality among predominately adult patients with HIV-associated tuberculosis (TB)</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Study</bold></th><th align="left"><bold>Year of publication</bold></th><th align="left"><bold>Study design</bold></th><th align="left"><bold>Country</bold></th><th align="left"><bold>Level of bacterial resistance to co-trimoxazole</bold></th><th align="left"><bold>No. of study participants</bold></th><th align="left"><bold>Mortality reduction</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Wiktor <italic>et al</italic>. [<xref ref-type="bibr" rid="B40">40</xref>]<hr/></td><td align="left" valign="bottom">1999<hr/></td><td align="left" valign="bottom">Randomized controlled trial<hr/></td><td align="left" valign="bottom">Cote D’Ivoire<hr/></td><td align="left" valign="bottom">Low<hr/></td><td align="left" valign="bottom">771<hr/></td><td align="left" valign="bottom">46%<hr/></td></tr><tr><td align="left" valign="bottom">Zachariah <italic>et al</italic>. [<xref ref-type="bibr" rid="B41">41</xref>]<hr/></td><td align="left" valign="bottom">2003<hr/></td><td align="left" valign="bottom">Cohort study (‘before’ and ‘after’ study with historical controls)<hr/></td><td align="left" valign="bottom">Malawi (north)<hr/></td><td align="left" valign="bottom">High<hr/></td><td align="left" valign="bottom">1,986<hr/></td><td align="left" valign="bottom">19%<hr/></td></tr><tr><td align="left" valign="bottom">Mwaungulu <italic>et al</italic>. [<xref ref-type="bibr" rid="B42">42</xref>]<hr/></td><td align="left" valign="bottom">2004<hr/></td><td align="left" valign="bottom">Cohort study (‘before’ and ‘after’ study with historical controls)<hr/></td><td align="left" valign="bottom">Malawi (south)<hr/></td><td align="left" valign="bottom">High<hr/></td><td align="left" valign="bottom">717<hr/></td><td align="left" valign="bottom">22%<hr/></td></tr><tr><td align="left" valign="bottom">Grimwade <italic>et al</italic>. [<xref ref-type="bibr" rid="B43">43</xref>]<hr/></td><td align="left" valign="bottom">2005<hr/></td><td align="left" valign="bottom">Cohort study (‘before’ and ‘after’ study with historical controls)<hr/></td><td align="left" valign="bottom">South Africa<hr/></td><td align="left" valign="bottom">High<hr/></td><td align="left" valign="bottom">3,325<hr/></td><td align="left" valign="bottom">29%<hr/></td></tr><tr><td align="left">Nunn <italic>et al</italic>. [<xref ref-type="bibr" rid="B44">44</xref>]</td><td align="left">2008</td><td align="left">Randomized controlled trial</td><td align="left">Zambia</td><td align="left">High</td><td align="left">1,003</td><td align="left">21%</td></tr></tbody></table></table-wrap><p>Routine administration of co-trimoxazole to patients with HIV-associated TB is recommended (480 mg twice per day or 960 mg once per day) [<xref ref-type="bibr" rid="B10">10</xref>-<xref ref-type="bibr" rid="B12">12</xref>]. Implementation of this simple, life-saving intervention has steadily increased from a negligible proportion in 2004 to 79% of all notified TB cases with a positive HIV test in 2011 (79% of those in the African region and 89% of those in the South-East Asian region) [<xref ref-type="bibr" rid="B1">1</xref>]. Coverage needs to increase to the 100% target set in the Global Plan to Stop TB, 2011-2015 [<xref ref-type="bibr" rid="B46">46</xref>]. Evidence is unclear as to whether co-trimoxazole should be continued indefinitely or might be discontinued once the CD4 cell count has reached a threshold of either 200 or 350 cells/μl [<xref ref-type="bibr" rid="B11">11</xref>]. The potential benefits of ongoing therapy may vary according to local factors such as the safety of the water supply, the presence of malaria and the local spectrum of opportunistic pathogens.</p></sec><sec><title>Antiretroviral treatment</title><p>In observational cohort studies, concurrent ART reduces mortality risk by 64% to 95% in patients receiving treatment for HIV-associated TB [<xref ref-type="bibr" rid="B47">47</xref>]. In the South African Starting Antiretroviral Therapy at Three Points in Tuberculosis Therapy (SAPIT) randomized trial, receipt of concurrent ART was associated with survival benefit among those with CD4 cell counts of <200 cells/μl and 200 to 500 cells/μl [<xref ref-type="bibr" rid="B48">48</xref>]. Recommended first-line ART regimens for use with TB treatment are based on non-nucleoside reverse transcriptase inhibitors (NNRTI), with efavirenz (EFV) as the preferred choice and nevirapine (NVP) as an alternative. While first-line regimen choices are well established, second-line ART remains problematic. The recommended regimens and their pharmacokinetic interactions with TB treatment are shown in Table <xref ref-type="table" rid="T5">5</xref> and the hiv-druginteractions.org website provides a useful up-to-date source of information on interactions (see [<xref ref-type="bibr" rid="B49">49</xref>]). Combining the multidrug regimens used to treat TB and HIV is complicated not only by high pill burden and increased risks of drug-drug interactions, but also by cotoxicity and immune reconstitution inflammatory syndrome (IRIS).</p><table-wrap position="float" id="T5"><label>Table 5</label><caption><p>Approaches to cotreatment for HIV-infected patients with rifampicin-susceptible tuberculosis</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Combined regimens</bold></th><th align="left"><bold>Treatment recommendations</bold></th><th align="left"><bold>Drug-drug interactions</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Efavirenz + rifampicin-based TB treatment<hr/></td><td align="left" valign="bottom">No dose adjustments TDF + 3TC/FTC + EFV (WHO-recommended optimum regimen) AZT + 3TC + EFV (alternative WHO regimen)<hr/></td><td align="left" valign="bottom">Rifampicin induces CYP2B6 but inhibition of CYP2A6 by isoniazid might account for increased efavirenz concentrations during TB treatment in those patients with slow CYP2B6 metabolizer genotype<hr/></td></tr><tr><td align="left" valign="bottom">Nevirapine + rifampicin-based TB treatment<hr/></td><td align="left" valign="bottom">Omit 14 day lead-in phase of once daily dose of NVP TDF + 3TC/FTC + NVP (alternative WHO regimen) AZT + 3TC + NVP (alternative WHO regimen)<hr/></td><td align="left" valign="bottom">Rifampicin induces CYP2B6 and CYP3A4. Although TB treatment reduces nevirapine concentrations, toxicity concerns curtail increasing the dose and outcomes are acceptable (but inferior to EFV) on standard doses.<hr/></td></tr><tr><td align="left" valign="bottom">Lopinavir/ritonavir + rifampicin-based TB treatment<hr/></td><td align="left" valign="bottom">Double dose lopinavir/ritonavir (800/200 mg 12 hourly) Or superboost lopinavir (lopinavir/ritonavir 400/400 mg 12 hourly) Monitor alanine transaminase (ALT) closely.<hr/></td><td align="left" valign="bottom">Rifampicin induces CYP3A4, p-glycoprotein and OATP1B1. Ritonavir counteracts this effect and adjusted doses of ritonavir or lopinavir/ritonavir are used to compensate, but lopinavir concentrations may be more variable. Increased risk of hepatotoxicity, and gastrointestinal side effects.<hr/></td></tr><tr><td align="left" valign="bottom">PI/ritonavir + rifabutin-based TB treatment<hr/></td><td align="left" valign="bottom">Reduce rifabutin dose to 150 mg daily or thrice weekly. Monitor closely for rifabutin toxicity.<hr/></td><td align="left" valign="bottom">Ritonavir-boosted PIs markedly increase rifabutin concentrations and reduce its clearance necessitating reduction in the dose of rifabutin by 50% to 75%. Toxicity (neutropenia, uveitis, hepatoxicity, rash, gastrointestinal symptoms) and suboptimal rifamycin exposures with reduced dose are concerns.<hr/></td></tr><tr><td align="left">Triple nucleoside/tide regimen + rifampicin-based TB treatment</td><td align="left">No dose adjustments. A triple nucleoside/tide regimen should include tenofovir or abacavir. Monitor viral load.</td><td align="left">Triple nucleoside/tide regimens may perform adequately in patients with viral suppression who have not failed a first line regimen, and provide alternative ART regimens in patients with contraindications to efavirenz or nevirapine, wehre other options are unavailable. TB treatment has minimal effect on tenofovir concentrations. Although rifampicin induces the enzymes responsible for glucuronidation of abacavir and zidovudine, this effect is not thought to be clinically important.</td></tr></tbody></table><table-wrap-foot><p><italic>3TC</italic> 2′,3′-dideoxy-3′-thiacytidine, <italic>ART</italic> antiretroviral therapy, <italic>CYP</italic> cytochrome P450, <italic>EFV</italic> efavirenz, <italic>FTC</italic> emtricitabine, <italic>OATP</italic> organic anion-transporting polypeptide, <italic>NNRTI</italic> non-nucleoside reverse transcriptase inhibitor, <italic>NVP</italic> nevirapine, <italic>PI</italic> protease inhibitor, <italic>TB</italic> tuberculosis, <italic>TDF</italic> tenofovir, <italic>WHO</italic> World Health Organization.</p></table-wrap-foot></table-wrap><sec><title>Pharmacokinetic interactions with <italic>fir</italic>st-line ART</title><p>Although RIF induces the expression of cytochrome P450 2B6 (CYP2B6), which comprises the main metabolic pathway for EFV, studies have failed to demonstrate significantly reduced concentrations of EFV with concomitant RIF-based TB treatment [<xref ref-type="bibr" rid="B50">50</xref>-<xref ref-type="bibr" rid="B53">53</xref>]. This is consistent with the observed virological responses which are excellent in patients receiving RIF-based TB treatment treated with standard 600 mg daily doses of EFV [<xref ref-type="bibr" rid="B54">54</xref>-<xref ref-type="bibr" rid="B57">57</xref>] and were better than those in TB patients randomized to NVP-based ART in the recent CARINEMO trial [<xref ref-type="bibr" rid="B56">56</xref>]. Similarly, lowering the dose of EFV to 400 mg daily in the ENCORE1 trial did not compromise outcomes in non-TB patients [<xref ref-type="bibr" rid="B58">58</xref>]. Thus, although the US Federal Drug Administration (FDA) [<xref ref-type="bibr" rid="B59">59</xref>] recommends that the dose of EFV during RIF treatment is increased in adults weighing more than 50 kg, this is not supported by studies in TB patients [<xref ref-type="bibr" rid="B53">53</xref>] and is not recommended by the WHO for resource-limited settings.</p><p>Conversely, however, among patients with a slow <italic>CYP2B6</italic> metabolizer genotype, EFV concentrations are increased during TB treatment, possibly due to inhibition by INH of accessory pathways metabolizing EFV [<xref ref-type="bibr" rid="B60">60</xref>,<xref ref-type="bibr" rid="B61">61</xref>]. This genotype is relatively common in Africa, South-East Asia, and the Caribbean [<xref ref-type="bibr" rid="B50">50</xref>-<xref ref-type="bibr" rid="B52">52</xref>,<xref ref-type="bibr" rid="B62">62</xref>,<xref ref-type="bibr" rid="B63">63</xref>]. Whether EFV-induced central nervous system (CNS) adverse effects are more frequent during TB treatment or isoniazid preventive therapy in patients with this genotype needs to be evaluated.</p><p>NVP is a reasonably safe, acceptable alternative for TB patients unable to tolerate EFV. Through induction of the expression of CYP2B6, RIF treatment reduces NVP concentrations by an average of approximately 40% and NVP-based ART remains inferior to EFV-based regimens in TB patients [<xref ref-type="bibr" rid="B56">56</xref>]. During the 14-day lead-in phase of NVP dosing, plasma drug concentrations are very low in patients receiving RIF, potentially predisposing to the development of viral resistance mutations and contributing to an increased risk of virological failure [<xref ref-type="bibr" rid="B54">54</xref>]. The CARENIMO trial recently found that NVP was well tolerated when introduced at full doses (200 mg twice a day) in patients with CD4 cell counts <250 cells/mm<sup>3</sup> receiving RIF [<xref ref-type="bibr" rid="B56">56</xref>]. The use of a dose escalation lead-in phase to avoid toxicity in patients receiving RIF is therefore not recommended.</p><p>Triple nucleoside/tide regimens are less effective than NNRTI-based or PI-based regimens particularly in patients with baseline viral loads >100,000 copies/ml [<xref ref-type="bibr" rid="B64">64</xref>]. However, small uncontrolled studies suggest they may provide an acceptable regimen for TB patients who have not failed an ART regimen [<xref ref-type="bibr" rid="B65">65</xref>,<xref ref-type="bibr" rid="B66">66</xref>] even though the concentrations of abacavir and zidovudine may be reduced by concomitant RIF. This therefore provides an alternative option for those in whom EFV and NVP are contraindicated and integrase inhibitors unavailable.</p></sec><sec><title>Pharmacokinetic interactions with <italic>seco</italic>nd-line ART</title><p>With increasing numbers of patients switching to protease inhibitor (PI)-based second-line ART regimens, defining safe and effective approaches to concurrent TB treatment is an urgent challenge. The pharmacokinetic interactions between rifamycins and PIs are extensive. RIF reduces concentrations of ritonavir-boosted PIs by 75% to 90% [<xref ref-type="bibr" rid="B67">67</xref>]. Conversely, through potent inhibition of CYP3A4 and p-glycoprotein, high-dose ritonavir offsets the effect of RIF-mediated induction such that ‘superboosting’ of lopinavir or saquinavir (lopinavir/ritonavir 400 mg/400 mg or saquinavir/ritonavir 400 mg/400 mg, twice daily) preserves plasma concentrations of the PI [<xref ref-type="bibr" rid="B68">68</xref>-<xref ref-type="bibr" rid="B70">70</xref>]. Adequate plasma concentrations of lopinavir are also achieved in adults by doubling the dose of lopinavir/ritonavir in the tablet formulation (to 800/200 mg twice daily); this is the simplest approach, especially in settings where the separate ritonavir is not available [<xref ref-type="bibr" rid="B71">71</xref>]. Although these approaches are associated with high rates of hepatotoxicity in studies of healthy volunteers, these seem to be much safer in HIV infected patients [<xref ref-type="bibr" rid="B71">71</xref>-<xref ref-type="bibr" rid="B76">76</xref>]. Nevertheless, hepatotoxicity, gastrointestinal side effects and poor tolerability are problematic and treatment discontinuation rates of up to nearly 50% have been reported [<xref ref-type="bibr" rid="B74">74</xref>,<xref ref-type="bibr" rid="B75">75</xref>].</p><p>Rifabutin is an alternative rifamycin to RIF, but data on its use in TB patients receiving ritonavir-boosted PIs are limited. Studies of healthy volunteers show that ritonavir-boosted PIs increase the concentrations of rifabutin approximately fourfold and the concentrations of the active metabolite to an even greater extent. Thus, the dose of rifabutin needs to be reduced. Thrice weekly 150 mg doses of rifabutin in combination with standard doses of lopinavir/ritonavir may be reasonably tolerated [<xref ref-type="bibr" rid="B77">77</xref>,<xref ref-type="bibr" rid="B78">78</xref>]. However, contrary to expectations based on pharmacokinetic data from healthy volunteers, small studies in coinfected patients have found that rifabutin 150 mg used thrice weekly in combination with lopinavir/ritonavir resulted in low rifabutin concentrations [<xref ref-type="bibr" rid="B79">79</xref>-<xref ref-type="bibr" rid="B82">82</xref>]. Such levels would be conducive to acquisition of rifamycin resistance in patients with severe immunosuppression [<xref ref-type="bibr" rid="B79">79</xref>,<xref ref-type="bibr" rid="B83">83</xref>] as has been observed with twice weekly doses [<xref ref-type="bibr" rid="B84">84</xref>]. Thus, recent US national guidelines recommend a daily 150 mg dose of rifabutin for patients on ritonavir-boosted PIs [<xref ref-type="bibr" rid="B85">85</xref>].</p><p>There is extremely limited information about the safety or efficacy using rifabutin with PIs and this may vary between populations due to differential increases in rifabutin concentrations. Severe neutropenia and uveitis occur relatively frequently in patients with increased exposures [<xref ref-type="bibr" rid="B81">81</xref>,<xref ref-type="bibr" rid="B86">86</xref>] and hepatitis, gastrointestinal symptoms, rashes and anemia are also important safety concerns [<xref ref-type="bibr" rid="B87">87</xref>,<xref ref-type="bibr" rid="B88">88</xref>]. While rifabutin is becoming more widely available and affordable, it is not an ideal solution for high burden settings where limited patient monitoring is available and fixed dose drug formulations are preferred. Thus, there is an urgent need for research to define the optimum approaches for the cotreatment of patients with TB who have failed first-line ART, including the use of newer agents.</p></sec><sec><title>Pharmacokinetic interactions with newer ART drugs</title><p>Ritonavir-boosted darunavir has a favorable safety and tolerability compared to lopinavir/ritonavir and promising efficacy, especially in treatment of ART-experienced patients. A pharmacokinetic study in healthy volunteers suggests that it could be used in standard doses with rifabutin 150 mg thrice weekly, but the drug-drug interactions with RIF have not been studied. Integrase inhibitors have potent antiviral activity and are well tolerated, but any future role in ART programs in low-resource settings is at present undefined. However, initial data on use with TB treatment show promise. Pharmacokinetic studies suggest that doubling the dose of raltegravir to 800 mg twice daily compensates for the effect of RIF on overall exposure [<xref ref-type="bibr" rid="B89">89</xref>,<xref ref-type="bibr" rid="B90">90</xref>] and this approach seems to be well tolerated and effective in patients with HIV-associated TB [<xref ref-type="bibr" rid="B91">91</xref>]. However, preliminary results of the REFLATE TB study suggest that such dose adjustment may not even be necessary as virological responses were similar in ART-naive TB patients receiving RIF who were randomized to receive 400 mg or 800 mg of raltegravir twice daily or EFV daily [<xref ref-type="bibr" rid="B92">92</xref>]. Similar to raltegravir, a pharmacokinetic study of dolutegravir in healthy volunteers suggests that the effect of RIF on antiretroviral therapy can be overcome by increasing the daily 50 mg dose of dolutegravir to 50 mg twice daily and that dose adjustment may not be necessary with rifabutin [<xref ref-type="bibr" rid="B93">93</xref>].</p></sec><sec><title>Timing of ART initiation during TB treatment</title><p>The optimum time to start ART in patients with HIV-associated TB is subject to a complex series of competing risks [<xref ref-type="bibr" rid="B94">94</xref>] and must balance the high risk of morbidity and mortality in patients with very low CD4 cell counts and severe disease with the potential occurrence of additive toxicities and immune reconstitution inflammatory syndrome (IRIS). Results of large randomized strategy trials are now available to inform guidelines (Table <xref ref-type="table" rid="T6">6</xref>) [<xref ref-type="bibr" rid="B48">48</xref>,<xref ref-type="bibr" rid="B55">55</xref>,<xref ref-type="bibr" rid="B95">95</xref>-<xref ref-type="bibr" rid="B98">98</xref>]. Patients with baseline CD4 counts of <200 and 200 to 500 cells/μl have improved survival benefit from coadministered ART [<xref ref-type="bibr" rid="B48">48</xref>] and WHO recommends that ART be given to all patients concurrently with TB treatment regardless of the CD4 count. Trial data also demonstrated that mortality was reduced in those with the most severe immunodeficiency (CD4 cell counts <50 cells/μl) if they stated ART within the first 2 weeks of TB treatment [<xref ref-type="bibr" rid="B11">11</xref>]. For patients with less severe immunosuppression (CD4 counts >50 cells/μl), data suggested that ART might be deferred until completion of the intensive phase of TB treatment without compromising survival but reducing the risk of morbidity from TB-IRIS [<xref ref-type="bibr" rid="B55">55</xref>,<xref ref-type="bibr" rid="B96">96</xref>].</p><table-wrap position="float" id="T6"><label>Table 6</label><caption><p>Randomized controlled studies of the timing of starting antiretroviral therapy (ART) during tuberculosis (TB) treatment</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left" valign="bottom"><bold>Study</bold><hr/></th><th colspan="4" align="center" valign="bottom"><bold>Study population</bold><hr/></th><th colspan="2" align="center" valign="bottom"><bold>Methods</bold><hr/></th><th colspan="4" align="center" valign="bottom"><bold>Results</bold><hr/></th></tr><tr><th align="left"> </th><th align="left"><bold>N</bold></th><th align="left"><bold>Location</bold></th><th align="left"><bold>TB</bold></th><th align="left"><bold>Median CD4+ cells/mm</bold><sup> <bold>3 </bold> </sup><bold>(IQR)</bold></th><th align="left"><bold>Timing of ART in weeks ‘earlier’ vs ‘later’</bold></th><th align="left"><bold>Primary endpoint</bold></th><th align="left"><bold>Follow-up in months</bold></th><th align="left"><bold>Primary endpoint ‘earlier’ vs ‘later’</bold><sup> <bold>a</bold> </sup></th><th align="left"><bold>Primary endpoint in CD4 <50 cells/μl</bold><sup> <bold>b</bold> </sup></th><th align="left"><bold>TB immune reconstitution</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">SAPIT [<xref ref-type="bibr" rid="B48">48</xref>] (first analysis)<hr/></td><td align="left" valign="bottom">429<hr/></td><td align="left" valign="bottom">South Africa<hr/></td><td align="left" valign="bottom">Smear-positive pulmonary TB<hr/></td><td align="left" valign="bottom">150 (77 to 254)<hr/></td><td align="left" valign="bottom"><12 vs after end TB treatment<hr/></td><td align="left" valign="bottom">Death<hr/></td><td align="left" valign="bottom">12.1<hr/></td><td align="left" valign="bottom">5.4 vs 12.1 <italic>P</italic> = 0.003<sup>c</sup><hr/></td><td align="left" valign="bottom">Not reported<hr/></td><td align="left" valign="bottom">12.4% vs 3.8% <italic>P</italic> <0.001<hr/></td></tr><tr><td align="left" valign="bottom">SAPIT [<xref ref-type="bibr" rid="B96">96</xref>] (second analysis)<hr/></td><td align="left" valign="bottom">429<hr/></td><td align="left" valign="bottom">South Africa<hr/></td><td align="left" valign="bottom">Smear-positive pulmonary TB<hr/></td><td align="left" valign="bottom">150 (77 to 254)<hr/></td><td align="left" valign="bottom">Within 4 vs 8 to 12<hr/></td><td align="left" valign="bottom">AIDS or death<hr/></td><td align="left" valign="bottom">17.7<hr/></td><td align="left" valign="bottom">6.9 vs 7.8 <italic>P</italic> = 0.73<hr/></td><td align="left" valign="bottom">8.5 vs 26.3<sup>b</sup><italic>P</italic> = 0.06<hr/></td><td align="left" valign="bottom">20.1% vs 7.7% <italic>P</italic> <0.001<hr/></td></tr><tr><td align="left" valign="bottom">CAMELIA [<xref ref-type="bibr" rid="B95">95</xref>]<hr/></td><td align="left" valign="bottom">660<hr/></td><td align="left" valign="bottom">Cambodia<hr/></td><td align="left" valign="bottom">Smear-positive TB<hr/></td><td align="left" valign="bottom">25 (11 to 56)<hr/></td><td align="left" valign="bottom">2 vs 8<hr/></td><td align="left" valign="bottom">Death<hr/></td><td align="left" valign="bottom">25<hr/></td><td align="left" valign="bottom">18% vs 27%, <italic>P</italic> = 0.006<hr/></td><td align="left" valign="bottom">Not reported<sup>d</sup><hr/></td><td align="left" valign="bottom">33.1% vs 13.7% <italic>P</italic> <0.001<hr/></td></tr><tr><td align="left" valign="bottom">STRIDE [<xref ref-type="bibr" rid="B55">55</xref>]<hr/></td><td align="left" valign="bottom">809<hr/></td><td align="left" valign="bottom">Multicontinent<sup>e</sup><hr/></td><td align="left" valign="bottom">Confirmed or presumed pulmonary or extrapulmonary TB<hr/></td><td align="left" valign="bottom">77 (36 to 145)<hr/></td><td align="left" valign="bottom">2 vs 8 to 12<hr/></td><td align="left" valign="bottom">AIDS or death<hr/></td><td align="left" valign="bottom">12<hr/></td><td align="left" valign="bottom">12.9% vs 16.1% <italic>P</italic> = 0.45<hr/></td><td align="left" valign="bottom">15.5% vs 26.6% <italic>P</italic> = 0.02<hr/></td><td align="left" valign="bottom">11% vs 5% <italic>P</italic> = 0.02<hr/></td></tr><tr><td align="left" valign="bottom">TB Meningitis [<xref ref-type="bibr" rid="B97">97</xref>]<hr/></td><td align="left" valign="bottom">253<hr/></td><td align="left" valign="bottom">Vietnam<hr/></td><td align="left" valign="bottom">TB meningitis<hr/></td><td align="left" valign="bottom">39 (18 to 116)<hr/></td><td align="left" valign="bottom">≤1 vs 8<hr/></td><td align="left" valign="bottom">Death<sup>f</sup><hr/></td><td align="left" valign="bottom">12<hr/></td><td align="left" valign="bottom">59.8% vs 55.6% <italic>P</italic> = 0.50<hr/></td><td align="left" valign="bottom">63.3% vs 65.1% <italic>P</italic> = 0.84<hr/></td><td align="left" valign="bottom">Not reported<hr/></td></tr><tr><td align="left">TIME Trial [<xref ref-type="bibr" rid="B98">98</xref>]</td><td align="left">156</td><td align="left">Thailand</td><td align="left">Confirmed or presumed pulmonary or extrapulmonary TB</td><td align="left">43 (37 to 106)</td><td align="left">4 vs 12</td><td align="left">Death</td><td align="left">96 weeks</td><td align="left">7.6% vs 6.5% <italic>P</italic> >0.99</td><td align="left">8.7% vs 13.1% <italic>P</italic> = 0.725</td><td align="left">8.86 vs 5.02 <italic>P</italic> = 0.069</td></tr></tbody></table><table-wrap-foot><p>Footnotes:</p><p><sup>a</sup>Presented either as cumulative incidence of primary endpoint in early vs. later arm (%) or as events per 100 person-years.</p><p><sup>b</sup>Prespecified analysis.</p><p><sup>c</sup>Significant difference in mortality observed in patients with either CD4 counts <200 cells/μl or 200 to 500 cells/μl.</p><p><sup>d</sup>Lower CD4 was not associated with an increased risk for the primary endpoint.</p><p><sup>e</sup>North America, South America, Asia, Africa.</p><p><sup>f</sup>Primary endpoint was all cause mortality at 9 months.</p></table-wrap-foot></table-wrap><p>WHO guidelines reflect these findings, recommending that TB treatment should be started first and followed by ART as soon as possible within the first 8 weeks of treatment but within the first 2 weeks for those with profound immunosuppression (CD4 count <50 cells/μl) [<xref ref-type="bibr" rid="B11">11</xref>]. However, CD4 count measurements may either be unavailable or be inaccurate in some settings. In addition, within different CD4 count categories, there is great diversity in severity of disease and mortality risk. Thus, where feasible, decisions on timing for individual patients might also be further informed by taking into account clinical criteria such as body mass index, Karnofsky score, severity of anemia and extent of TB. Moreover, national guidelines might best be appropriately tailored for operational simplicity. One possible option, for example, might be to start ART in all patients after 2 weeks of TB treatment, accepting lower risk of mortality but higher risk of TB IRIS.</p><p>Patients with HIV-associated TB meningitis represent an important exception. A randomized trial from Viet Nam found no survival benefit from early ART in patients with TB meningitis [<xref ref-type="bibr" rid="B97">97</xref>], reflecting the awful prognosis (mortality approximately 60%) of these patients with advanced disease and the dire consequences of TB-IRIS within the confined space of the CNS [<xref ref-type="bibr" rid="B99">99</xref>]. Further studies are required in different geographical settings to better define appropriate management of these patients.</p></sec><sec><title>Adverse drug reactions and management</title><p>Antituberculosis and antiretroviral drugs have overlapping toxicity profiles that include drug-induced liver injury (DILI), cutaneous reactions, renal impairment, neuropathy and neuropsychiatric adverse effects (Table <xref ref-type="table" rid="T7">7</xref>). These complicate management in a substantial minority of patients.</p><table-wrap position="float" id="T7"><label>Table 7</label><caption><p>Shared side effects of antiretroviral therapy (ART) and antituberculosis drugs</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Adverse effects</bold></th><th align="left"><bold>Antiretroviral drugs</bold></th><th align="left"><bold>Antituberculosis drugs</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Gastrointestinal disturbance and/or pain<hr/></td><td align="left" valign="bottom">AZT, ddI, PIs<hr/></td><td align="left" valign="bottom">RIF, INH, PZA, ethionamide, PAS, clofazamine, linezolid<hr/></td></tr><tr><td align="left" valign="bottom">Liver injury<hr/></td><td align="left" valign="bottom">NVP, EFV, PIs, NRTIs<sup>a</sup><hr/></td><td align="left" valign="bottom">RIF, INH, PZA and many second line drugs including ethionamide, fluoroquinolones, PAS<hr/></td></tr><tr><td align="left" valign="bottom">Peripheral neuropathy<hr/></td><td align="left" valign="bottom">D4T, ddI<hr/></td><td align="left" valign="bottom">INH, ethionamide, terizidone/cycloserine, linezolid<hr/></td></tr><tr><td align="left" valign="bottom">Neuropsychiatric<hr/></td><td align="left" valign="bottom">EFV<hr/></td><td align="left" valign="bottom">Terizidone/cycloserine, ethionamide, fluoroquinolones, INH<hr/></td></tr><tr><td align="left" valign="bottom">Renal impairment<hr/></td><td align="left" valign="bottom">TDF<hr/></td><td align="left" valign="bottom">Aminoglycosides and capreomycin<hr/></td></tr><tr><td align="left" valign="bottom">Rash<hr/></td><td align="left" valign="bottom">NVP, EFV, ABC<hr/></td><td align="left" valign="bottom">Rifampicin, INH, PZA, ethambutol, streptomycin and many second line drugs including fluoroquinolones, PAS, clofazamine<hr/></td></tr><tr><td align="left" valign="bottom">Blood dyscrasias<hr/></td><td align="left" valign="bottom">AZT, 3TC<hr/></td><td align="left" valign="bottom">Linezolid, rifabutin, INH, rifampicin<hr/></td></tr><tr><td align="left" valign="bottom">Cardiac conduction abnormalities<hr/></td><td align="left" valign="bottom">PIs<hr/></td><td align="left" valign="bottom">Bedaquiline, fluoroquinolones, clofazamine<hr/></td></tr><tr><td align="left" valign="bottom">Pancreatitis<hr/></td><td align="left" valign="bottom">D4T, ddI<hr/></td><td align="left" valign="bottom">Linezolid<hr/></td></tr><tr><td align="left">Lactic acidosis</td><td align="left">D4T, ddI</td><td align="left">Linezolid</td></tr></tbody></table><table-wrap-foot><p><italic>3TC</italic> 2',3'-dideoxy-3'-thiacytidine, <italic>ABC</italic> abacavir, <italic>AZT</italic> zidovudine, <italic>D4T</italic> stavudine, <italic>ddI</italic> didanosine, <italic>EFV</italic> efavirenz, <italic>INH</italic> isoniazid, <italic>NRTIs</italic> nucleoside reverse transcriptase inhibitors, <italic>NVP</italic> nevirapine, <italic>PAS</italic> para-aminosalicylic acid, <italic>PIs</italic> protease inhibitors, <italic>PZA</italic> pyrazinamide, <italic>RIF</italic> rifampicin, <italic>TDF</italic> tenofovir.</p><p><sup>a</sup>NRTIs (especially D4T and ddI) can cause steatohepatitis.</p></table-wrap-foot></table-wrap><p>In patients without coinfection, DILI (variably defined as, for example, an elevation of alanine aminotransferase to >3 or >5 times the upper limit of the normal range) occurs in 5% to 33% of those receiving TB treatment [<xref ref-type="bibr" rid="B100">100</xref>] and in 5% to 11% of those receiving currently recommended ART regimens [<xref ref-type="bibr" rid="B101">101</xref>,<xref ref-type="bibr" rid="B102">102</xref>]. HIV infection itself has been identified as a risk factor for DILI in patients receiving TB treatment in some [<xref ref-type="bibr" rid="B103">103</xref>,<xref ref-type="bibr" rid="B104">104</xref>] but not all studies [<xref ref-type="bibr" rid="B105">105</xref>-<xref ref-type="bibr" rid="B108">108</xref>]. Of the currently used ART drugs, NVP is associated with highest risk of DILI; however, EFV and PIs are also recognized causes.</p><p>Concurrent TB treatment in patients receiving NNRTI-based ART has been associated with an increased risk of DILI in some [<xref ref-type="bibr" rid="B109">109</xref>-<xref ref-type="bibr" rid="B111">111</xref>] but not all [<xref ref-type="bibr" rid="B54">54</xref>] studies. In one of these, the absolute risk of severe hepatotoxicity in patients receiving EFV-based ART was low, but the risk associated with concurrent TB treatment exceeded that associated with positive hepatitis B surface antigen status [<xref ref-type="bibr" rid="B109">109</xref>]. Importantly, a randomized trial of NVP-based versus EFV-based ART in patients receiving TB treatment reported more treatment discontinuations related to DILI in the NVP arm (4 vs 0%) [<xref ref-type="bibr" rid="B56">56</xref>].</p><p>Development of DILI significantly complicates management of HIV-associated TB. Elevation of alanine transaminase (ALT) concentrations >3 to 5 times the upper limit of normal especially when accompanied by symptoms or jaundice requires that all potentially hepatotoxic medication is interrupted until derangements of liver function tests resolve. Thereafter, rechallenge of first-line TB medication should be considered followed by ART, although rechallenge is generally not undertaken if there was liver failure. Rechallenge strategies have not been studied in randomized trials in HIV-infected patients. However, in the largest randomized trial of TB without HIV coinfection, approximately 90% of patients were rechallenged with their first-line TB drugs without recurrence [<xref ref-type="bibr" rid="B112">112</xref>]. Risk of recurrence was not related to whether the four first-line TB drugs were reintroduced sequentially or concurrently. Further studies are needed to define the optimum rechallenge strategy in coinfected patients in whom both TB treatment and ART require reintroduction. Until further evidence emerges, the American Thoracic Society recommends that RIF can be reintroduced in coinfected patients once the ALT is less than two times the upper limit of normal followed by reintroduction of INH with monitoring of liver function [<xref ref-type="bibr" rid="B100">100</xref>]. However, they also suggest that pyrazinamide is not reintroduced.</p><p>While some cohort studies have suggested low morbidity and mortality in HIV-infected patients with DILI [<xref ref-type="bibr" rid="B109">109</xref>], mortality is substantial among those requiring hospital admission. In a South African study, mortality was 35% among patients admitted to hospital with DILI during TB treatment, ART or concurrent therapy [<xref ref-type="bibr" rid="B113">113</xref>]. Reasons for these deaths were sepsis and liver failure, although interruption of required TB treatment and ART are likely to have played a role.</p><p>TB treatment is associated with a spectrum of cutaneous adverse reactions including morbiliform rashes, Steven Johnson syndrome and toxic epidermal necrolysis, fixed drug eruption, lichenoid drug eruptions and acute generalized exanthematous pustulosis [<xref ref-type="bibr" rid="B114">114</xref>]. Cotrimoxazole, NVP, and to a lesser extent EFV, can also cause many of the same clinical presentations [<xref ref-type="bibr" rid="B102">102</xref>,<xref ref-type="bibr" rid="B115">115</xref>,<xref ref-type="bibr" rid="B116">116</xref>]. HIV coinfection was associated with a fivefold increased risk of rash or drug fever in one study [<xref ref-type="bibr" rid="B117">117</xref>] but small, non-significant increases in risk in others [<xref ref-type="bibr" rid="B105">105</xref>,<xref ref-type="bibr" rid="B108">108</xref>]. If a clinically significant rash develops, all potentially responsible drugs need to be interrupted and then a carefully monitored rechallenge of first-line TB drugs can be considered once the rash has resolved. In a cohort of mainly HIV-infected patients rechallenged following cutaneous reactions to TB drugs, 50% developed reintroduction reactions but only a small minority were severe [<xref ref-type="bibr" rid="B118">118</xref>].</p><p>Renal dysfunction may be caused via different mechanisms in patients receiving tenofovir, RIF or aminoglycosides (used for MDR-TB). Tenofovir and aminoglycosides may both cause tubular cell toxicity at the level of the proximal renal tubules, whereas RIF infrequently causes a tubulointerstitial nephritis mediated by immune hypersensitivity. Case reports describe renal failure in patients receiving a combination of tenofovir and aminoglycosides, although cohort studies have not confirmed an increased risk [<xref ref-type="bibr" rid="B119">119</xref>]. The combination is best avoided when possible. In patients with significant renal dysfunction, Use of tenofovir should be avoided where possible and dosing of ethambutol, NRTI drugs, some quinolones (ofloxacin and levofloxacin) and certain other second-line antituberculosis drugs (including cycloserine, para-aminosalicylic acid, clofazamine and linezolid) needs to be adjusted<italic>.</italic></p></sec><sec><title>TB Immune reconstitution inflammatory syndrome (IRIS)</title><p>Two major forms of TB immune reconstitution syndrome (TB-IRIS) are recognized and these are called paradoxical TB IRIS and unmasking TB-IRIS and case definitions have been published [<xref ref-type="bibr" rid="B120">120</xref>]. Paradoxical TB-IRIS is an important cause of morbidity in patients known to have HIV-associated TB and occurs within the first weeks of ART [<xref ref-type="bibr" rid="B120">120</xref>,<xref ref-type="bibr" rid="B121">121</xref>]. The typical clinical course of paradoxical TB-IRIS is as follows. Initiation of TB treatment in a patient with HIV infection and newly diagnosed TB results in clinical stabilization or improvement. However, subsequent introduction of ART is accompanied by recurrence or exacerbation of TB symptoms with new or worsening clinical signs of TB that often have a marked inflammatory component [<xref ref-type="bibr" rid="B120">120</xref>,<xref ref-type="bibr" rid="B121">121</xref>].</p><p>While seldom life-threatening, deaths due to paradoxical TB-IRIS have been described. Two major risk factors identified in observational studies [<xref ref-type="bibr" rid="B122">122</xref>-<xref ref-type="bibr" rid="B125">125</xref>] and in clinical trials [<xref ref-type="bibr" rid="B55">55</xref>,<xref ref-type="bibr" rid="B95">95</xref>,<xref ref-type="bibr" rid="B126">126</xref>] are a low CD4 count prior to ART and a shorter interval between starting TB treatment and ART. There is no diagnostic test for TB-IRIS; the diagnosis is based on clinical presentation and exclusion of alternative diagnoses such as bacterial infection or drug resistant TB [<xref ref-type="bibr" rid="B120">120</xref>]. However, drug-resistant TB is not only in the differential diagnosis as an alternative cause of the clinical deterioration but may also be a risk factor for the development of paradoxical TB-IRIS [<xref ref-type="bibr" rid="B127">127</xref>].</p><p>The second major form of TB-IRIS is commonly referred to as ‘unmasking’ TB-IRIS. This occurs when active TB is present but remains undiagnosed at the time of starting ART [<xref ref-type="bibr" rid="B120">120</xref>,<xref ref-type="bibr" rid="B128">128</xref>]. Subsequent immune recovery triggers the overt symptomatic presentation of TB. In a proportion of cases, unusual inflammatory features may also develop and such cases are regarded as having ‘unmasking’ TB-IRIS. Risk of unmasking TB-IRIS is therefore directly related to the efficiency of the pre-ART screening process and the resulting prevalence of undiagnosed disease.</p><p>Both types of TB IRIS have a wide range of clinical features often with involvement of multiple organ systems, reflecting widespread dissemination of <italic>M. tuberculosis</italic> in those with profound immunosuppression. Common features include fever, recurrence of respiratory symptoms with worsening infiltrates on chest radiographs, enlargement of lymph nodes (often with suppuration), formation of tuberculous abscesses and serous effusions [<xref ref-type="bibr" rid="B120">120</xref>,<xref ref-type="bibr" rid="B121">121</xref>]. There are many case reports of unusual and diverse complications, including granulomatous nephritis with renal impairment, parotitis, epididymo-orchitis, granulomatous hepatitis, splenic enlargement and abscess formation, psoas abscess, peritonitis, ascites and intestinal involvement [<xref ref-type="bibr" rid="B120">120</xref>,<xref ref-type="bibr" rid="B121">121</xref>]. Neurological TB-IRIS is particularly severe, manifesting with tuberculomas, tuberculous abscesses, cerebral edema, meningitis and radiculomyelopathy [<xref ref-type="bibr" rid="B99">99</xref>,<xref ref-type="bibr" rid="B129">129</xref>,<xref ref-type="bibr" rid="B130">130</xref>]. Neurological TB IRIS has a much poorer outcome compared to other forms, with a mortality of 13% to 75% [<xref ref-type="bibr" rid="B99">99</xref>,<xref ref-type="bibr" rid="B129">129</xref>,<xref ref-type="bibr" rid="B130">130</xref>].</p><p>In most cases, the onset of paradoxical TB-IRIS is within the first 4 weeks of ART (median 14 days (IQR, 8 to 23) in 1 series [<xref ref-type="bibr" rid="B127">127</xref>]) but can occur within a few days. The proportion of patients affected ranges widely from 0% to over 40% [<xref ref-type="bibr" rid="B120">120</xref>] and this may relate to differences in risk factors and case definitions. In a meta-analysis, the summary risk estimate was 15.7% [<xref ref-type="bibr" rid="B131">131</xref>]. Of these, 3.2% died, representing approximately 1 in 200 patients with HIV-associated TB who start ART. The median duration of TB-IRIS symptoms has been reported to be 2 to 3 months [<xref ref-type="bibr" rid="B124">124</xref>,<xref ref-type="bibr" rid="B125">125</xref>] but a minority of cases have a protracted course which may last for more than 1 year [<xref ref-type="bibr" rid="B120">120</xref>,<xref ref-type="bibr" rid="B124">124</xref>,<xref ref-type="bibr" rid="B132">132</xref>]. Such protracted cases typically have persistent or recurrent suppurative lymphadenitis or abscess formation. However, the majority of cases have a favorable long-term outcome [<xref ref-type="bibr" rid="B133">133</xref>].</p><p>TB-IRIS is not an indication for stopping ART, although this should be considered in life-threatening cases such as those with cerebral edema and depressed level of consciousness or severe respiratory failure. In mild cases, no specific treatment is usually required; the patient should be treated symptomatically and counseled regarding the need to continue ART and TB treatment. Corticosteroids should be considered if symptoms are more significant. In a randomized placebo-controlled trial, prednisone used at a dose of 1.5 mg/kg/day for 2 weeks followed by 0.75 mg/kg/day for 2 weeks was associated with reduced morbidity (duration of hospitalization and need for therapeutic procedures) [<xref ref-type="bibr" rid="B134">134</xref>]. Symptom improvement was more rapid and there was no excess risk of other severe infections [<xref ref-type="bibr" rid="B134">134</xref>]. Although no mortality benefit was demonstrated, patients with immediately life-threatening TB-IRIS were not enrolled in view of ethical considerations. Indeed, most experts recommend steroid therapy for life-threatening TB-IRIS, especially IRIS involving the CNS. A subgroup of patients in this trial (approximately one in five) relapsed after stopping prednisone and required a further and more prolonged course to control symptoms [<xref ref-type="bibr" rid="B134">134</xref>]. Similarly, in other settings, TB-IRIS has relapsed in up to 50% of patients after stopping steroids [<xref ref-type="bibr" rid="B133">133</xref>] and thus the duration of therapy must be tailored according to the clinical response.</p><p>Non-steroidal anti-inflammatory drugs (NSAIDs) have also been used in the treatment of TB IRIS although no clinical trial data exist to support their use. Other forms of immunomodulatory therapy such as thalidomide, azathioprine and tumor necrosis factor α blockers (such as adalumimab) have been used in cases refractory to steroid therapy with anecdotal reports of benefit [<xref ref-type="bibr" rid="B135">135</xref>]. In patients with suppurative lymphadenitis or abscesses, needle aspiration may provide a pus sample to exclude drug-resistant TB as well as bringing symptomatic relief.</p><p>There is no evidence base for pharmacological prevention of TB-IRIS. However, this needs to be considered in view of the recommendation within guidelines for early ART initiation in TB patients with advanced HIV [<xref ref-type="bibr" rid="B11">11</xref>]. Adjunctive immunomodulatory therapies might reduce the risk or severity of TB-IRIS in such patients. A randomized placebo-controlled trial of prednisone for prevention of TB-IRIS in high-risk patients (CD4 counts <100 cells/mm<sup>3</sup> starting ART within 30 days of TB treatment) is underway [<xref ref-type="bibr" rid="B136">136</xref>]. Until results from this trial are available corticosteroids cannot be recommended for prevention of TB IRIS with the exception of patients with TB of the CNS for whom adjunctive steroids form part of the standard of care [<xref ref-type="bibr" rid="B137">137</xref>]. However, in such patients, TB IRIS occurs in approximately 50% of patients with CNS TB starting ART despite receipt of corticosteroids [<xref ref-type="bibr" rid="B99">99</xref>].</p><p>Other agents that have been proposed for prevention of TB IRIS are vitamin D, statins and the C-C chemokine receptor type 5 (CCR5) blocker maraviroc [<xref ref-type="bibr" rid="B135">135</xref>]. Vitamin D has modulating effects on both the adaptive and innate immune responses [<xref ref-type="bibr" rid="B138">138</xref>,<xref ref-type="bibr" rid="B139">139</xref>]. Statins have anti-inflammatory properties and there is precedence for using these agents for autoimmune inflammatory disorders in an experimental model [<xref ref-type="bibr" rid="B140">140</xref>,<xref ref-type="bibr" rid="B141">141</xref>]. However, neither vitamin D nor statins have yet been tested in clinical studies. Maraviroc, however, was shown not to prevent IRIS in a placebo-controlled trial conducted in Mexico and South Africa [<xref ref-type="bibr" rid="B142">142</xref>].</p></sec></sec><sec><title>Management of HIV-associated MDR-TB</title><p>The emergence of MDR-TB and extensively drug resistant TB (XDR-TB) has compounded the HIV-associated TB epidemic in resource-limited settings [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B143">143</xref>]. MDR-TB is caused by strains that are resistant to both rifampicin and isoniazid whereas XDR-TB strains are MDR-TB strains with additional resistance to any quinolone drug and any one of the second-line injectable aminoglycosides (amikacin, capreomycin or kanamycin). Much disease remains undiagnosed due to lack of laboratory capacity. However, increasing implementation of the Xpert MTB/RIF assay now provides the means for rapid screening for RIF resistance, although follow-on testing is then required to further characterize the full drug susceptibility pattern. This can be performed phenotypically through culture-based systems but is very slow. In 2008, WHO approved the use of line probe assays for the rapid molecular detection of drug resistance in smear-positive specimens or culture isolates [<xref ref-type="bibr" rid="B144">144</xref>] and a range of commercially available assays now offer the possibility of much more rapid diagnosis of both MDR-TB and XDR-TB [<xref ref-type="bibr" rid="B145">145</xref>]. However, line-probe assays can only be used where appropriate laboratory facilities and expertise exist as they are highly technically demanding and are well beyond the scope of most resource-limited settings apart from in specialized reference laboratories.</p><p>Worldwide, successful treatment of MDR-TB is achieved in only approximately 50% to 60% of patients [<xref ref-type="bibr" rid="B146">146</xref>,<xref ref-type="bibr" rid="B147">147</xref>], but management is considerably more difficult in resource-limited settings and especially in those with HIV coinfection due to late diagnosis with more frequent extrapulmonary dissemination, high risks of drug cotoxicity and IRIS, copathology and poor adherence with prolonged, toxic regimens. The WHO recommends that patients with confirmed MDR-TB should receive a regimen containing pyrazinamide together with at least four second-line drugs in the intensive phase that are likely to be effective, including a fluoroquinolone (using a later generation agent where possible), a parenteral agent (such as amikacin or kanamycin), ethionamide (or prothionamide) and either cycloserine or p-aminosalicylic acid (PAS) [<xref ref-type="bibr" rid="B13">13</xref>]. An intensive phase of 8 months and a total treatment duration of 20 months is suggested for most patients, but may be modified according to response. A range of other second-line drugs that have limited efficacy may be used for treatment of XDR-TB and treatment regimens should be based upon drug susceptibility testing [<xref ref-type="bibr" rid="B13">13</xref>]. However, evidence to inform best practice is lacking and outcomes are often poor.</p><p>Co-trimoxazole prophylaxis and ART are recommended for all patients with HIV-associated MDR-TB regardless of CD4 count and the timing of ART initiation is similar as for drug-susceptible TB [<xref ref-type="bibr" rid="B11">11</xref>]. Many of the second-line MDR-TB drugs are poorly tolerated and drug discontinuation rates are high as a result of adverse effects. MDR-TB may be a risk factor for TB IRIS in view of slow mycobacterial antigen clearance [<xref ref-type="bibr" rid="B127">127</xref>]. Nutritional depletion and co-morbid conditions may further undermine outcomes.</p><p>Adverse events are frequent in HIV-infected patients receiving MDR treatment, the most common being gastrointestinal symptoms, peripheral neuropathy, hypothyroidism, deafness, psychiatric symptoms and hypokalemia [<xref ref-type="bibr" rid="B148">148</xref>,<xref ref-type="bibr" rid="B149">149</xref>]. In up to 40% of patients these adverse events are severe [<xref ref-type="bibr" rid="B148">148</xref>]. This relates to the inherent toxicity associated with MDR drugs; it does not appear that HIV-infected patients experience a higher incidence of adverse events than HIV-uninfected patients, nor that coadministration with ART increases toxicity [<xref ref-type="bibr" rid="B148">148</xref>,<xref ref-type="bibr" rid="B150">150</xref>,<xref ref-type="bibr" rid="B151">151</xref>].</p><p>Antiretroviral drugs do share common toxicities with second-line antituberculosis drugs, however (Table <xref ref-type="table" rid="T7">7</xref>). Some of the most challenging of these are neuropsychiatric side effects. EFV causes inattention, vivid dreams and dizziness in up to 50% of patients, but in a minority these can be severe with mood disturbance or psychosis. Cycloserine (or terizidone) is a well recognized cause of psychosis, seizures and other CNS side effects although several other drugs such as the quinolones, ethionamide and high dose isoniazid can also cause CNS side effects. If patients develop severe CNS side effects it may be necessary to withdraw all possible culprit drugs with careful sequential reintroduction once resolved. Cycloserine should probably be regarded as the most likely culprit for psychosis and seizures. Antipsychotic or antidepressant medications may be required. EFV should not be routinely avoided because the majority of MDR-TB patients tolerate it well.</p><p>Much research is needed on how to improve treatment for drug-resistant TB. A shortened MDR-TB regimen of 9 months, which was found to be effective and well tolerated in Bangladesh [<xref ref-type="bibr" rid="B152">152</xref>], is now being evaluated in Ethiopia, South Africa and Vietnam and includes patients with HIV-associated TB. In the future, the newly approved agent bedaquiline (TMC-207) as well as two new nitroimidazoles (PA-824 and delaminid (OPC67683) under evaluation) may offer the prospects of improved treatment for MDR-TB [<xref ref-type="bibr" rid="B38">38</xref>]. However, a prolonged timeline is needed to adequately define how to combine existing agents and new drugs in regimens that optimize outcomes and that can be combined with ART in those with HIV-associated TB.</p></sec></sec><sec sec-type="conclusions"><title>Conclusions</title><p>The HIV-associated TB epidemic is a major challenge to international public health, remaining the most important opportunistic infection in people living with HIV globally and accounting for nearly 0.5 million deaths each year. However, over the past 10 years, major progress has been achieved in defining guidelines for the optimum case management with a combination of co-trimoxazole prophylaxis, optimally timed ART, and diagnosis and appropriate supportive care for treatment complications including drug toxicity and IRIS. The major remaining challenges are the management of TB in the increasing proportion of patients receiving PI-containing ART and the management of drug resistant TB. Having defined case management strategies, the ongoing challenge is to further develop effective, comprehensive and sustainable means of delivery through health systems.</p></sec><sec><title>Abbreviations</title><p>ALT: alanine transaminase; ART: antiretroviral treatment; CNS: central nervous system; CYP: cytochrome P450 enzyme; E: ethambutol; EFV: efavirenz; H/INH: isoniazid; IRIS: immune reconstitution inflammatory syndrome; LAM: lipoarabinomannan; MDR-TB: multidrug resistant tuberculosis; NNRTI: non-nucleoside reverse transcriptase inhibitor; NVP: nevirapine; PI: protease inhibitor; PITC: provider initiated counseling and testing; R/RIF: rifampicin; TB: tuberculosis; VCT: voluntary counseling and testing; WHO: World Health Organization; XDR-TB: extensively drug resistant tuberculosis; Z: pyrazinamide.</p></sec><sec><title>Competing interests</title><p>The authors declare they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>The first draft was written by SDL, GM and HMcI. All authors contributed to the development of subsequent and final drafts. All authors approved the final version.</p></sec>
Increased expression of T cell immunoglobulin and mucin domain 3 aggravates brain inflammation via regulation of the function of microglia/macrophages after intracerebral hemorrhage in mice	<sec><title>Background</title><p>Microglia/macrophages are known to play important roles in initiating brain inflammation after spontaneous intracerebral hemorrhage (ICH). T cell immunoglobulin and mucin domain-3 (Tim-3) have been proven to play a critical part in several inflammatory diseases through regulation of both adaptive and innate immune responses. Tim-3 can be expressed by microglia/macrophages and regulates their function in the innate immune response. However, the effect of Tim-3 on inflammatory responses following ICH is unclear.</p></sec><sec><title>Methods</title><p>In this study, we investigated Tim-3 expression, the inflammatory cytokines tumor necrosis factor-α (TNF-α) and interleukin-1β (IL-1β), and brain water content in peri-hematomal brain tissue at 12 hours and at 1, 3, 5, and 7 days post-ICH in wild type (WT) ICH and Tim-3<sup>−/−</sup> ICH mice. The numbers of Tim-3 positive cells,astrocytes, neutrophils and microglia/macrophages were detected using immunofluorescence staining. Cytokines were measured by ELISA. Double immunoflurorescence labeling was performed to identify the cellular source of Tim-3 expression. Mouse neurological deficit scores were assessed through animal behavior.</p></sec><sec><title>Results</title><p>Expression of Tim-3 increased early in mouse peri-hematomal brain tissue after autologous blood injection, peaked at day 1, and was positively correlated with the concentrations of TNF-α, IL-1β, and brain water content. Tim-3 was predominantly expressed in microglia/macrophages. Compared with WT mice, Tim-3<sup>−/−</sup> mice had reduced ICH-induced brain inflammation with decreased TNF-α and IL-1β, cerebral edema and neurological deficit scores. Moreover, Tim-/- inhibited activation of microglia/macrophages. The number of activated microglia/macrophages in Tim-3<sup>−/−</sup> ICH mice was much lower than that in WT ICH mice.</p></sec><sec><title>Conclusions</title><p>Our findings demonstrate that Tim-3 plays an important role in brain inflammation after ICH, and may be a potential treatment target.</p></sec>	<contrib contrib-type="author" id="A1"><name><surname>Xu</surname><given-names>ChangJun</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>xuchangjun828@163.com</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Wang</surname><given-names>Tao</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>wgg2004@163.com</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Cheng</surname><given-names>Si</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>lcchenming@sina.com</email></contrib><contrib contrib-type="author" corresp="yes" id="A4"><name><surname>Liu</surname><given-names>YuGuang</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>liuyuguang3000@163.com</email></contrib>	Journal of Neuroinflammation	<sec><title>Background</title><p>Spontaneous intracerebral hemorrhage (ICH) is a common disease with high mortality and morbidity, accounting for 15 to 20% of all stokes and affecting more than 2 million people worldwide each year [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>]. The mortality rate of ICH is more than 40% and only 20% of survivors can live independently within six months [<xref ref-type="bibr" rid="B3">3</xref>]. But until now, there has been no satisfactory treatment in clinical practice, mainly because the mechanisms of brain damage after ICH are unclear. Increasing researches show that inflammatory response plays an important role in ICH-induced secondary brain damage [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B5">5</xref>]. Brain inflammation after ICH is characterized by accumulation of activated inflammatory cells, such as blood-derived cells (macrophages, leukocytes) and brain resident cells (astrocytes, microglia and mast cells). These reactive cells can release inflammatory mediators, including chemokines, cytokine, protease, prostaglandins and other immunoactive molecules [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref>]. The microglia are the first cells to react to brain damage among all the inflammatory cells. Activated microglia have a similar shape to the blood-derived macrophages, therefore microglia are also called as the brain macrophages. Increasing evidence indicates that microglia/macrophages are activated early following ICH, and release a series of toxic factors, including chemokines, cytokines, reactive oxygen species (ROS), cyclooxygenase-2, protease, heme oxygenase-1 and prostaglandins [<xref ref-type="bibr" rid="B8">8</xref>-<xref ref-type="bibr" rid="B10">10</xref>]. Therefore, microglia/macrophages play important roles in the secondary brain damage. Tim-3 is a new immuno-regulation molecule found in 2002, which is expressed specially in activated Th1 cells [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B12">12</xref>]. Moreover, it has been proven that Tim-3 is also expressed in the cells of the innate immune system, including macrophages, mast cells and dendritic cells [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B14">14</xref>]. Tim-3 expression in microglia/macrophages can be upregulated and induce the production of proinflammatory cytokines, such as tumor necrosis factor-a (TNF-a) and interleukin-1β (IL-1β), which can aggravate inflammation and secondary brain damage [<xref ref-type="bibr" rid="B14">14</xref>]. However, until now, the effect of Tim-3 on inflammatory response following ICH has been unclear. Considering that the microglia/macrophages are the key cells for inducing brain inflammation and secondary brain damage, and that Tim-3 can regulate the function of microglia/macrophages, we hypothesized that Tim-3 possibly took part in ICH-induced inflammation by regulating the function of microglia/macrophages. This experiment was done to prove our hypothesis.</p></sec><sec sec-type="materials\|methods"><title>Materials and methods</title><sec><title>Intracerebral hemorrhage model</title><p>The ICH model of mice was established using the method described previously [<xref ref-type="bibr" rid="B15">15</xref>]. Animals were anesthetized with chloral hydrate (40 mg/kg, intraperitoneal injection). A catheter plugged in the right femoral artery was used to monitor continuous blood pressure and to take blood sampling. Mice were fixed in a stereotactic frame (Stoelting, Kiel, WI, USA). The puncture point was located 1 mm anterior to the bregma and 2.5 mm lateral to midline. A 1-mm cranial hole was drilled with dental bit and a 27-gauge needle was inserted stereotaxically into the right basal ganglia (4 mm deep). Then, 25 μL of autologous blood was injected using a micro-infusion pump at a rate of 2.5 μL/min. After the infusion, the needle was kept in the place for another 10 minutes to prevent blood leakage, and then the needle was pulled out. The cranial hole was sealed with bone wax and the skin was sutured. Control mice were injected with 25 μL 0.9% saline. Then mice were allowed to recover. During operation, rectal temperature of mice was maintained at 37 ± 1°C. Mean arterial blood pressure (MABP) (mmHg), arterial pH, arterial PO<sub>2</sub>, PCO<sub>2</sub>, Hb (g/L), and glucose levels (mg/dL) were monitored and maintained for stability.</p></sec><sec><title>Animals and grouping</title><p>C57BL/6 mice (male, 8 to 10 weeks old, and weight 20 to 25 g) were purchased from the Animal House Center, Medical College of Shangdong University (Jinan, China). Transgenic line Tim-3<sup>−/−</sup> mice (8 to 10 weeks old, weight 20 to 24 g) were purchased from American Jackson Laboratories (Bar Harbor, ME, USA) and were backcrossed to C57BL/6 mice more than eight times. All animals were housed in individual cages under a 12 hr/12 hr light–dark cycle with temperature 21 ± 1°C, humidity 50 to 60% and free access to food and water. All animals care and experimental protocols were approved by theanimal Ethics Committee of the Sandong University. All efforts were made to minimize the animals’ pain and to reduce the number of animals. Different experimental groups were included: 1) For the WT sham group (n = 60), 25 μL of 0.9% saline was injected into the brain of these mice. 2) For the WT ICH group (n = 60), 25 μL of autologous blood was injected into the right basal ganglia. 3) For the Tim-3<sup>−/−</sup> sham group (n = 50),the operation was the same as for the WT sham group. 4) For the Tim-3<sup>−/−</sup> ICH group (n = 50),the operation was the same as for the WT ICH group. A total of 8 mice (5 mice in the WT ICH group and 3 mice in the Tim-3<sup>−/−</sup> ICH group) died in our experiment, and the number of mice supplemented was the same. There were 228 mice in our experiment.</p></sec><sec><title>Tissue preparation</title><p>In the WT sham group and the WT ICH group, 12 mice were anesthetized at each of the following time points: 12 hours and 1, 3, 5, and 7 days after injection. Half of the mice (n = 6) were directly decapitated and their brains were obtained to store at −80°C for use inreal-time RT-PCR, in ELISA and to determine brain water content. The others (n = 6) were used for immunofluorescence. These mice underwent transcardial perfusion with 200 ml of phosphate buffered saline (PBS), followed by 100 ml of 4% paraformaldehyde (PFA) in 0.1 M PBS as described before [<xref ref-type="bibr" rid="B16">16</xref>]. The brains were removed and postfixed for 24 hours in 4% PFA, and then were placed in 30% sucrose until sinking. Coronal brain sections of 10 μm thickness were obtained with a freezing microtome (Leica, Nussloch, Germany) and were kept at −20°C. In the Tim-3<sup>−/−</sup> sham group and Tim-3<sup>−/−</sup> ICH group, 10 mice were killed at each of the following time points: 12 hours and 1, 3, 5, and 7 days after operation, Among this total, 6 mice were used for real-time RT-PCR, for ELISA and to determine brain water content, and the others (n = 4) were used for immunofluorescence.</p></sec><sec><title>Immunofluorescence staining and cell counting</title><p>After being washed in PBS for 10 min, the sections were incubated with 5% bovine serum albumin for 60 minutes in order to block the nonspecific binding, and then were incubated with goat anti-rat Tim-3 primary antibody (1:100; R&D systems, Minneapolis, MN, USA. <ext-link ext-link-type="uri" xlink:href="http://www.rndsystems.com/">http://www.rndsystems.com/</ext-link>) at 4°C all the night. After being washed three times with PBS, the sections were incubated with secondary antibody goat anti-rabbit IgG (1:100; KPL, Maryland, USA. <ext-link ext-link-type="uri" xlink:href="http://www.kpl.com/home.cfm">http://www.kpl.com/home.cfm</ext-link>) for 60 minutes at the room’s temperature. The sections were rinsed 3 × 5 min and were coverslipped with ProLong antifade medium (Molecular Probes, Eugene, OR, USA). The Tim-3 positive cells were visualized using a fluorescent microscope (Olympus BX51, Japan). For each animal, six representative sections of each brain were selected. Tim-3 positive cells were counted blindly in the approximately 40,000 μm <sup>2</sup> of brain tissues around blood clot. In order to further identify the cellular resource of Tim-3 after ICH, double immunofluorescence labeling [<xref ref-type="bibr" rid="B9">9</xref>] was performed by simultaneous incubation of goat anti-rat Tim-3 primary antibody with rat anti-mouse CD11b (1:200, eBioscience San Diego, CA, USA. <ext-link ext-link-type="uri" xlink:href="http://www.ebioscience.com/">http://www.ebioscience.com/</ext-link>) as marker of activated microglia/macrophage, or rabbit anti-GFAP (1:200, Invitrogen, Carlsbad, CA, USA. <ext-link ext-link-type="uri" xlink:href="http://www.invitrogen.com">http://www.invitrogen.com</ext-link>) as marker of astrocyte, or rabbit anti-MPO (1:100, Dako, Denmark. <ext-link ext-link-type="uri" xlink:href="http://www.dako.com">http://www.dako.com</ext-link>) as mark of neutrophils. In each group, the number of CD11b positive cells (activated microglia/macrophage), GFAP positive cells (astrocyte) and MPO positive cells (neutrophils) were counted using the same method as used for Tim-3 positive cells. Ipp6.0 image processing software was utilized to count the number of Tim-3 positive cells.</p></sec><sec><title>Real-time reverse transcription polymerase chain reaction</title><p>According to the methods described previously [<xref ref-type="bibr" rid="B15">15</xref>], frozen mice brains were homogenized, and total RNA was obtained from about 4 × 4 × 4 mm<sup>3</sup> volume of peri-hematomal tissues (blood clot as center under a stereomicroscope) at 12 hours and at 1, 3, 5 and 7 days post-ICH using Trizol reagent (Invitrogen, Carlsbad, CA, USA) in compliance with the manufacture’s instruction. The character of RNA was tested by a spectrophotometer (DU800, Beckman, Palo Alto, CA, USA). The M-MLV Reverse Transcriptase System (Promega, Madison, WI, USA) was performed for reverse transcription. The cDNA was stored at −20°C. Quantitative real-time PCR was fulfilled with a LightCycler (Roche Diagnostics, Mannheim, GM) and with SYBR Green I in SYBR RT-PCR Kit (TaKaRa Biotechnology, Dalian, China) so as to enlarge and detect the expression of Tim-3 mRNA. The transcript amount of the rat β-actin housekeeping gene was quantified as an internal RNA control. Primers were purchased from BioAsia Corp. (Shanghai, China). The primer sequences were as follows: β-actin forward: 5’-GGCATCGTGATGGACTCCG −3’ and β-actin reverse: 5′-GCTGGAAGGTGGACAGCGA −3′; Tim-3 forward: 5′ACTGGTGACCCTCCATAATAACA −3′ and Tim-3 reverse: 5′ATTTTCCTCAGAGCGAATCCT −3′. Experiments were carried out in triplicate for each data point. A threshold cycle value (CT) was calculated by the ΔΔCT method as previously described [<xref ref-type="bibr" rid="B17">17</xref>]. The data were analyzed by using Light Cycler Software 4.0 (Roche Diagnostics).</p></sec><sec><title>Enzyme-linked immunosorbent assay</title><p>IL-1β and TNF-aare two important inflammatory mediators and can represent the severity of brain inflammation after ICH. The concentrations of IL-1β and TNF-a in brain tissues of the peri-hematomal region were detected via the ELISA method according to the manufacturer’s instructions (R&D systems, Minneapolis, MN, USA). Brain tissues were centrifuged at 12000 rps for 20 min and the supernatant was collected for analysis. The detection threshold of this assay was <1 pg/mL.</p></sec><sec><title>Brain water content</title><p>Brain water content indicates the degree of brain edema, which is mainly due to peri-hematomal inflammation and blood–brain barrier breakdown. Brain samples were immediately weighed on an electric analytical balance to get the wet weight, and then dried at 100°C for 24 hours to obtain the dry weight. Brain water content (%) = (wet weight - dry weight)/wet weight of brain tissue × 100.</p></sec><sec><title>Testing of neurological deficit scores</title><p>At 12 hours and at 1, 3, 5 and 7 days after operation, the neurofunctional abnormality of mice was tested and scored in each group. An observer blinded to the identity of the mice evaluated behavior. Three behavioral examinations, including ipsilateral circling, forelimb flexion and beam balance were used as described before [<xref ref-type="bibr" rid="B18">18</xref>]. Each point is graded from 0 to 4. The maximum abnormal score is 12.</p></sec><sec><title>Statistical analysis</title><p>All statistical analyses were performed with SPSS 16.0 for Windows (Chicago, IL, USA). Data are presented as mean ± standard deviation (SD). Multiple group differences were analyzed using one-way or two-way analysis of variance and Student-Newman-Keuls test in post hoc tests. An independent-samples t-test was adopted for comparison of the two groups. The correlation analysis was completed by bivariate. <italic>P <</italic>0.05 was considered as the indication of statistical significance.</p></sec></sec><sec sec-type="results"><title>Results</title><sec><title>The physiological parameters in four groups during production of hematoma</title><p>During production of hematoma, we monitored rectal temperature, MABP, arterial pH, arterial PO<sub>2</sub>, PCO<sub>2</sub>, Hb, and glucose levels in the four groups. The results showed there was no difference among all groups (<italic>P</italic> > 0.05)(Table <xref ref-type="table" rid="T1">1</xref>).</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p><bold>Comparison of physiological parameters</bold><sup><bold>a</bold></sup> (<bold>PP</bold>) <bold>in all groups during production of hematoma</bold></p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"> </th><th align="left"><bold>WT sham group (n = 60)</bold></th><th align="left"><bold>WT ICH group (n = 60)</bold></th><th align="left"><bold>Tim-3</bold><sup> <bold>−/−</bold> </sup><bold>sham group (n = 50)</bold></th><th align="left"><bold>Tim-3</bold><sup> <bold>−/−</bold> </sup><bold>ICH group (n = 50)</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Rectal temperature<hr/></td><td align="left" valign="bottom">36.97 ± 0.54<hr/></td><td align="left" valign="bottom">37.12 ± 0.42<hr/></td><td align="left" valign="bottom">37.06 ± 0.49<hr/></td><td align="left" valign="bottom">37.01 ± 0.39<hr/></td></tr><tr><td align="left" valign="bottom">MABP(mmHg)<hr/></td><td align="left" valign="bottom">100.04 ± 1.86<hr/></td><td align="left" valign="bottom">101.22 ± 1.96<hr/></td><td align="left" valign="bottom">99.75 ± 2.09<hr/></td><td align="left" valign="bottom">100.09 ± 1.66<hr/></td></tr><tr><td align="left" valign="bottom">Arterial pH<hr/></td><td align="left" valign="bottom">7.38 ± 0.04<hr/></td><td align="left" valign="bottom">7.41 ± 0.02<hr/></td><td align="left" valign="bottom">7.40 ± 0.05<hr/></td><td align="left" valign="bottom">7.35 ± 0.02<hr/></td></tr><tr><td align="left" valign="bottom">PO<sub>2</sub>(mmHg)<hr/></td><td align="left" valign="bottom">106.57 ± 8.50<hr/></td><td align="left" valign="bottom">110.21 ± 5.84<hr/></td><td align="left" valign="bottom">109.86 ± 9.71<hr/></td><td align="left" valign="bottom">103.34 ± 5.04<hr/></td></tr><tr><td align="left" valign="bottom">PCO<sub>2</sub>(mmHg)<hr/></td><td align="left" valign="bottom">42.54 ± 4.12<hr/></td><td align="left" valign="bottom">41.58 ± 2.71<hr/></td><td align="left" valign="bottom">41.77 ± 3.07<hr/></td><td align="left" valign="bottom">42.37 ± 3.42<hr/></td></tr><tr><td align="left" valign="bottom">Hb(g/L)<hr/></td><td align="left" valign="bottom">149.17 ± 6.79<hr/></td><td align="left" valign="bottom">147.01 ± 6.86<hr/></td><td align="left" valign="bottom">143.49 ± 8.64<hr/></td><td align="left" valign="bottom">149.06 ± 7.48<hr/></td></tr><tr><td align="left">Glucose (mg/dL)</td><td align="left">135.90 ± 14.76</td><td align="left">137.16 ± 8.47</td><td align="left">134.64 ± 16.38</td><td align="left">137.72 ± 10.08</td></tr></tbody></table><table-wrap-foot><p><sup>a</sup>Data were presented as mean ± standard deviation. The physiological parameters were not differentamong all groups. <italic>P</italic> > 0.05.ICH, intracerebral hemorrhage; Tim-3, T cell immunoglobulin and mucin domain-3; WT, wild type.</p></table-wrap-foot></table-wrap></sec><sec><title>Increase of Tim-3 expression in the peri-hematomal brain tissues</title><p>In order to investigate the expression of Tim-3 in the peri-hematomal brain tissues, we observed the number of Tim-3 positive cells in the peri-hematomal brain tissues at 12 hours and at 1, 3, 5 and 7 days post-ICH. Results indicated that the number of Tim-3 positive cells in the WT ICH group began to increase at 12 hours, peaked at Day 1, and decreased at Day 3 after ICH. There was a significant difference when compared with that of WT sham mice in all time-tested points (<italic>P</italic> < 0.01) (Figure <xref ref-type="fig" rid="F1">1</xref>A,B). Furthermore, we studied Tim-3 mRNA expression in the peri-hematomal brain tissues at 12 hours and at 1, 3, 5 and 7 days post-ICH using a real-time RT-PCR method. The change trend of Tim-3 mRNA was similar to the number of Tim-3 positive cells. The Tim-3 mRNA expression was strikingly upregulated in the peri-hematomal brain tissues after ICH at 12 hours, peaked at Day 1, and descended at Day 3. The Tim-3 mRNA expression was significantly different when compared with that in WT sham group(<italic>P</italic> < 0.01) (Figure <xref ref-type="fig" rid="F1">1</xref>C).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Expression of Tim-</bold><bold>3 positive cells and Tim</bold>-<bold>3 mRNA in peri</bold>-<bold>hematomal brain tissues at time</bold>-<bold>tested points following intracerebral hemorrhage (ICH). A)</bold> and <bold>B)</bold> Immunofluorescence staining showed that the number of Tim-3 positive cells increased at 12 hours in the wild type (WT) ICH group mice (n = 6), with the maximum at Day 1, <italic>P <</italic>0.01 versus the WT sham group (mean ± SD). Scale bar = 20 μm. <bold>C)</bold> Real-time RT-PCR indicated that the expression of Tim-3 mRNA increased at all time-tested points in the ICH group mice (n = 6) and peaked at Day 1. <italic>P <</italic>0.01 versus WT sham group (mean ± SD). ‘Relative fold’ meant the level of mRNA of Tim-3 relative to that of beta-actin.</p></caption><graphic xlink:href="1742-2094-10-141-1"/></fig></sec><sec><title>Preponderant expression of Tim-3 in microglia/macrophages</title><p>To identify the cellular resource of Tim-3 expression, we observed double-immunofluorescence staining. The results revealed that expression of Tim-3 was preponderant in CD11b<sup>+</sup> cells (microglia/macrophages), was lower in MPO<sup>+</sup> cells (neutrophils), and was lowest in GFAP<sup>+</sup> cells (astrocytes) (Figure <xref ref-type="fig" rid="F2">2</xref>).</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>The expression of Tim</bold>-<bold>3 in different brain cells at day 1 after intracerebral hemorrhage (ICH).</bold> Double-immunofluorescence staining displayed that: <bold>A)</bold> A few of Tim-3 positive cells could express in GFAP<sup>+</sup> cells (astrocytes). <bold>B)</bold> A few of Tim-3 positive cells could express in MPO<sup>+</sup> cells (neutrophils), but more than were expressed in GFAP<sup>+</sup> cells (astrocytes). The arrows indicate co-expressed cells of Tim-3 and MPO. <bold>C)</bold> Almost all CD11b<sup>+</sup> cells (microglia/macrophages) were Tim-3 positive cells at day 1 after ICH. The arrows indicate co-expressed cells of Tim-3 and CD11b. Scale bar = 20 μm.</p></caption><graphic xlink:href="1742-2094-10-141-2"/></fig></sec><sec><title>Obvious increase of IL-1β, TNF-a and brain water content in the peri-hematomal brain tissues</title><p>IL-1β and TNF-a are two main proinflammatory cytokines of microglial source, which represent the level of brain inflammation [<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>]. We detected the concentration of IL-1β and TNF-a in the peri-hematomal brain tissues at 12 hours and at 1, 3, 5 and 7 days post-ICH. Both of them elevated obviously in the WT ICH group at all tested time points. There was significant difference between the WT ICH group and the WT sham group (<italic>P</italic> < 0.01). IL-1β and TNF-a increased at 12 hours after ICH, and peaked at Day 1, then decreased at Day 3 (Figure <xref ref-type="fig" rid="F3">3</xref>A,B). We also measured the mice brain water content to judge the degree of brain edema. Brain water content heightened at 12 hours after ICH, peaked at Day 1 and Day 3, and then gradually declined. There was a significant difference between the WT ICH group and the WT sham group (<italic>P</italic> < 0.01) (Figure <xref ref-type="fig" rid="F3">3</xref>C).</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>The change of IL</bold>-<bold>1β</bold>, <bold>TNF</bold>-<bold>a,</bold><bold>brain water content and neurological deficit scores (NDS) after intracerebral hemorrhage (ICH). A)</bold> indicates the change of the concentration of IL-1β in the peri-hematomal brain tissues at 12 hours andat 1, 3, 5 and 7 days post-ICH. <bold>B)</bold> shows the change of the concentration of TNF-a. <bold>C)</bold> displays the change of brain water content. <bold>D)</bold> shows the NDS between the Tim-3<sup>−/−</sup> ICH group and the wild type (WT) ICH group. The WT ICH group (n = 6) versus the WT sham group (n = 6): <italic>P <</italic>0.01; The Tim-3<sup>−/−</sup> ICH group (n = 6) versus the Tim-3<sup>−/−</sup> sham group (n = 6): &&<italic>P</italic> < 0.01; The Tim-3<sup>−/−</sup> ICH group (n = 6) versus the WT ICH group (n = 6): <sup>#</sup><italic>P</italic> < 0.05, <sup>##</sup><italic>P</italic> < 0.01.</p></caption><graphic xlink:href="1742-2094-10-141-3"/></fig></sec><sec><title>Positive correlation of expression of Tim-3 with IL-1β, TNF-a and brain water content</title><p>We analyzed the correlation between expression of Tim-3 and IL-1β, TNF-a and brain water content. The results show that expression of Tim-3 was positively correlated with IL-1β (<italic>r</italic> = 0.618, <italic>P <</italic>0.001), TNF-a (<italic>r</italic> = 0.610, <italic>P <</italic>0.001) and brain water content (<italic>r</italic> = 0.566, <italic>P</italic> = 0.001)(Figure <xref ref-type="fig" rid="F4">4</xref>A,B,C).</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>The correlation between the expression of Tim</bold>-<bold>3 and IL</bold>-<bold>1β</bold>, <bold>TNF</bold>-<bold>a</bold>, <bold>brain water content. A)</bold>, <bold>B)</bold> and <bold>C)</bold> indicate that the expression of Tim-3 was positively correlated with IL-1β (<italic>r</italic> = 0.618, <italic>P <</italic>0.001), TNF-a (<italic>r</italic> = 0.610, <italic>P <</italic>0.001) and brain water content (<italic>r</italic> = 0.566, <italic>P</italic> = 0.001).</p></caption><graphic xlink:href="1742-2094-10-141-4"/></fig></sec><sec><title>The brain inflammatory response, brain edema and neurologic function in the Tim-3<sup>−/−</sup> ICH mice</title><p>To further prove the effects of Tim-3 on brain inflammation, brain edema and neurologic function, we compared the concentration of IL-1β and TNF-a, brain water content and NDS in the WT ICH group with that in the Tim-3<sup>−/−</sup> ICH group. The concentration of both IL-1β and TNF-a in the peri-hematomal brain tissues notably decreased in the Tim-3<sup>−/−</sup> ICH group, and there was significant difference compared to that in the WT ICH group (the Tim-3<sup>−/−</sup> ICH group versus the WT ICH group, IL-1β: 12 h <italic>P <</italic>0.01; 1d <italic>P <</italic>0.05; 3 d <italic>P <</italic>0.05; 5d <italic>P <</italic>0.05; 7d <italic>P <</italic>0.01. TNF-a: 12 h <italic>P <</italic>0.05; 1d <italic>P <</italic>0.05; 3d <italic>P <</italic>0.01; 5d <italic>P <</italic>0.01; 7d <italic>P <</italic>0.01) (Figure <xref ref-type="fig" rid="F3">3</xref>A,B). Brain water content also markedly declined in the Tim-3<sup>−/−</sup> ICH group in all tested time points (the Tim-3<sup>−/−</sup> ICH group versus the WT ICH group, 12 h <italic>P <</italic>0.01; 1d <italic>P <</italic>0.01; 3 d <italic>P <</italic>0.01; 5d <italic>P <</italic>0.01; 7d <italic>P <</italic>0.05.) (Figure <xref ref-type="fig" rid="F3">3</xref>C). At 12 hours and day 1, there was no difference of NDS between the Tim-3<sup>−/−</sup> ICH group and the WT ICH group. At 3 days, the NDS of the Tim-3<sup>−/−</sup> ICH group was lower than that of the WT ICH group, but there was no statistical difference. At 5 and 7 days, there was a significant difference between two groups (<italic>P</italic> < 0.05) (Figure <xref ref-type="fig" rid="F3">3</xref>D), which meant that blockage of expression of Tim-3 could improve the neurological deficit after ICH.</p></sec><sec><title>The change of astrocytes, neutrophils and microglia/macrophages after intracerebral hemorrhage</title><p>We investigated the number of astrocytes, neutrophils and microglia/macrophages in all groups using immunofluorescence staining. The results indicated that the number of astrocytes, neutrophils and microglia/macrophages all increased after ICH. But the peak times of the three cells were different. The number of astrocytes elevated at Day 1, peaked at Day5 and Day 7. The number of neutrophils rose at 12 hours, peaked at Day 3, and then gradually declined. But the number of astrocytes and neutrophils were not different between the WT ICH group and the Tim-3<sup>−/−</sup> ICH group. The number of microglia/macrophages increased at 12 hours and peaked at Day 1. This variation was the same as that of Tim-3, IL-1β and TNF-a. In Tim-3<sup>−/−</sup> ICH mice, the number of microglia/macrophages was obviously less than that of WT ICH mice and there was a significant difference between the two groups (<italic>P</italic> < 0.01) (Figure <xref ref-type="fig" rid="F5">5</xref>).</p><fig id="F5" position="float"><label>Figure 5</label><caption><p><bold>The change of the number of astrocytes</bold>, <bold>neutrophils and microglia</bold>/<bold>macrophages after intracerebral hemorrhage (ICH).</bold><bold>A)</bold> indicates the change in the number of astrocytes in the peri-hematomal brain tissues at 12 hours and at 1, 3, 5 and 7 days post-ICH. <bold>B)</bold> shows the change in the number of neutrophils. <bold>C)</bold> displays the change in the number of microglia/macrophages in the wild type (WT) ICH group (n = 6) versus the WT sham group (n = 6): <italic>P <</italic>0.01; The Tim-3<sup>−/−</sup> ICH group (n = 4) versus the Tim-3<sup>−/−</sup> sham group (n = 4): &&<italic>P <</italic>0.01; the Tim-3<sup>−/−</sup> ICH group (n = 4) versus the WT ICH group (n = 6): <sup>##</sup><italic>P <</italic>0.01.</p></caption><graphic xlink:href="1742-2094-10-141-5"/></fig></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>In the present study we wanted to prove the effect of Tim-3 on brain inflammation after ICH. Three aspects of this research had to be addressed. The first aspect involved the expression of Tim-3 in the peri-hematomal zone after ICH. The second was related to the cellular resource of Tim-3 expression. The third was concerned with the correlation between the expression of Tim-3 and the severity of inflammatory response and brain edema. Furthermore, we examined the change in the release of inflammatory cytokines, brain edema, neurofunctional impairment and inflammatory cells (astrocytes, neutrophils and microglia/macrophages) in Tim-3<sup>−/−</sup> ICH mice.</p><sec><title>The expression of Tim-3 in the peri-hematomal zone after ICH</title><p>Tim-3 is a new immuno-regulation molecule. It has been proved that Tim-3 extensively expresses itself in innate immune cells, including microglia, monocytes, mast cells, dendritic cells and natural killer cells, and plays complex roles in immune regulation and tolerance [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B21">21</xref>-<xref ref-type="bibr" rid="B23">23</xref>]. Tim-3 is involved in several inflammatory diseases, such as experimental autoimmune encephalomyelitis (EAE) [<xref ref-type="bibr" rid="B11">11</xref>], non-obese diabetes [<xref ref-type="bibr" rid="B24">24</xref>] and Coxsackievirus B3-induced myocarditis [<xref ref-type="bibr" rid="B21">21</xref>]. Anderson <italic>et al</italic>. [<xref ref-type="bibr" rid="B14">14</xref>] demonstrated that Tim-3 mRNA levels were much higher in inflamed white matter tissue in patients with multiple sclerosis (MS) and rat MS models, and Tim-3 upregulated on CD11b<sup>+</sup> peripheral monocytes and resident microglia to promote TNF-a secretion. Zhao <italic>et al</italic>. [<xref ref-type="bibr" rid="B25">25</xref>] investigated the expression of Tim-3 in the acute phase of ischemic stroke, and indicated that overexpression of Tim-3 both in brain tissues of ischemia-reperfusion mice and in peripheral blood mononuclear cells of patients with ischemic stroke positively correlated with plasma IL-17 and TNF-α. All these researchers suggested that Tim-3 took parts in inflammatory-immunologic reaction. But it is completely unknown whether or not Tim-3 plays roles in ICH. Now, we have experimentally found that Tim-3 increases in the mouse peri-hematomal brain tissues early after autologous blood injection and progressively increasesand accompanies the development of brain inflammation, suggesting that Tim-3 is also involved in the inflammatory response following ICH.</p></sec><sec><title>The cellular resource of Tim-3 expression</title><p>Previous studies have revealed that brain inflammation after ICH was characterized by the soakage of neutrophils and macrophages from blood and activation of microglia in brain tissues [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B26">26</xref>]. Among all inflammatory cells, microglia are the first non-neuronal cells to react to brain damage [<xref ref-type="bibr" rid="B27">27</xref>]. The shape of activated microglia is the same as the shape of blood-derived macrophages, so it is impossible to discriminate them from infiltrating macrophages. Microglia are activated at 1 hour after ICH, much earlier than neutrophil infiltration, and the latter appears at 4 to 5 h after ICH [<xref ref-type="bibr" rid="B9">9</xref>]. Activated microglia/macrophages can release cytotoxic mediators, such as IL-1β and TNF-a, inducing brain inflammatory reaction and secondary damage [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B28">28</xref>]. Blockage of microglia activation with tuftsin fragment 1–3 can attenuate neuroinflammation and brain damage after ICH in rats [<xref ref-type="bibr" rid="B24">24</xref>,<xref ref-type="bibr" rid="B29">29</xref>]. In this experimental study, we found that the expression of Tim-3 following ICH was preponderant in microglia/macrophages, was lower in neutrophils, and was lowest in astrocytes. Although astrocytes, neutrophils and microglia/macrophages all increased following ICH, but only variation of microglia/macrophages was the same as that of Tim-3, IL-1β and TNF-a. These indicate that Tim-3 can inhibit the activation of microglia/macrophages, but not influence the astrocytes and neutrophils. Tim-3 not only expressed in microglia/macrophages, but also could regulate the function of them.</p></sec><sec><title>The relationship between the expression of Tim-3 and the severity of inflammatory response and brain edema</title><p>IL-1β and TNF-a are two important proinflammatory cytokines, which can drive the inflammatory process and aggravate inflammation [<xref ref-type="bibr" rid="B30">30</xref>,<xref ref-type="bibr" rid="B31">31</xref>]. Although IL-1β and TNF-a are released by many cells, including microglia/macrophages, astrocytes and neurons, the major source of these are activated microglia/macrophages [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>]. Many experimental data have revealed that expression of IL-1β and TNF-a increase after ICH, and are associated with brain edema and brain damage. Increased expression of IL-1β and TNF-a can been detected not only in the central nervous system but also in systemic circulation [<xref ref-type="bibr" rid="B32">32</xref>-<xref ref-type="bibr" rid="B34">34</xref>]. IL-1β and TNF-α can boost inflammatory reaction in the early stage by promoting secretion of other chemotactic factors and adhesion molecules of the vascular endothelium, which can lead to the early infiltration of macrophages and neutrophils to the injury lesion [<xref ref-type="bibr" rid="B35">35</xref>,<xref ref-type="bibr" rid="B36">36</xref>]. The levels of TNF-α and IL-1β can represent the severity of brain inflammatory to a great degree. In our study, we found that TNF-α and IL-1β increased at 12 hours post-ICH, peaked at Day 1 and decreased at Day 3 after ICH. The expression of Tim-3 is positively correlated to levels of TNF-α, IL-1β and brain water content. Above all, we can deduce that augmented Tim-3 expression may promote inflammation and brain edema in the peri-hematomal tissues after ICH by regulating the function of microglia/macrophage. This result was supported by three points below. First of all, expression of Tim-3 increased early in the perilesional tissue and predominantly expressed itself in microglia/macrophages after ICH. The number of microglia/macrophages was much lower in Tim-3<sup>−/−</sup> ICH mice. Furthermore, the secretion of TNF-α and IL-1β elevated in the peri-hematomal tissues following ICH, and microglia/macrophages are the major types of cells to secrete TNF-a and IL-1β. In addition, there was a positive correlation between the expression of Tim-3 and the levels of TNF-α, IL-1β and brain water content. Blocking expression of Tim-3 could not only reduce the levels of TNF-α, IL-1β, brain water content and the number of microglia/macrophages, but also improve neurological function, which also elucidated the effect of Tim-3 on brain inflammatory reaction from another aspect. But the signal pathways of Tim-3 that regulate the function of microglia/macrophages need to be further explored.</p></sec></sec><sec sec-type="conclusions"><title>Conclusions</title><p>Tim-3 plays an important role in the brain inflammation after ICH by regulating the function of microglia/macrophages and may be a potential treatment target.</p></sec><sec><title>Abbreviations</title><p>ELISA: Enzyme-linked immunosorbent assay; ICH: Intracerebral hemorrhage; MABP: Mean arterial blood pressure; NDS: Neurological deficit scores; RT-PCR: Reverse transcription polymerase chain reaction.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>The research was based on the original idea of CJX and YGL. CJX worked on the development of animal model, immunofluorescent staining, behavioral studies, and data analysis and drafted the manuscript. TW worked on the development of animal model, behavioral studies and real-time RT PCR. SC worked on the development of animal model, behavioral studies, ELISA and brain water content. YGL provided the equipment and wrote the manuscript. All authors read and approved the final manuscript.</p></sec>
Neurovascular unit dysfunction with blood-brain barrier hyperpermeability contributes to major depressive disorder: a review of clinical and experimental evidence	<p>About one-third of people with major depressive disorder (MDD) fail at least two antidepressant drug trials at 1 year. Together with clinical and experimental evidence indicating that the pathophysiology of MDD is multifactorial, this observation underscores the importance of elucidating mechanisms beyond monoaminergic dysregulation that can contribute to the genesis and persistence of MDD. Oxidative stress and neuroinflammation are mechanistically linked to the presence of neurovascular dysfunction with blood-brain barrier (BBB) hyperpermeability in selected neurological disorders, such as stroke, epilepsy, multiple sclerosis, traumatic brain injury, and Alzheimer’s disease. In contrast to other major psychiatric disorders, MDD is frequently comorbid with such neurological disorders and constitutes an independent risk factor for morbidity and mortality in disorders characterized by vascular endothelial dysfunction (cardiovascular disease and diabetes mellitus). Oxidative stress and neuroinflammation are implicated in the neurobiology of MDD. More recent evidence links neurovascular dysfunction with BBB hyperpermeability to MDD without neurological comorbidity. We review this emerging literature and present a theoretical integration between these abnormalities to those involving oxidative stress and neuroinflammation in MDD. We discuss our hypothesis that alterations in endothelial nitric oxide levels and endothelial nitric oxide synthase uncoupling are central mechanistic links in this regard. Understanding the contribution of neurovascular dysfunction with BBB hyperpermeability to the pathophysiology of MDD may help to identify novel therapeutic and preventative approaches.</p>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Najjar</surname><given-names>Souhel</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>mna1024231@aol.com</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Pearlman</surname><given-names>Daniel M</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I3">3</xref><email>daniel.m.pearlman@gmail.com</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Devinsky</surname><given-names>Orrin</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>od4@nyu.edu</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Najjar</surname><given-names>Amanda</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>amanda.najjar@nyu.edu</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Zagzag</surname><given-names>David</given-names></name><xref ref-type="aff" rid="I4">4</xref><xref ref-type="aff" rid="I5">5</xref><email>dz4@nyu.edu</email></contrib>	Journal of Neuroinflammation	<sec><title>Background</title><p>Major depressive disorder (MDD) is the second leading global cause of years lived with disability [<xref ref-type="bibr" rid="B1">1</xref>], with about one-third of patients with MDD failing two or more conventional antidepressant drug trials within the first year of treatment [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B3">3</xref>]. Current evidence suggests that the pathophysiology of MDD is multifactorial, involving heterogeneous and inter-related mechanisms that affect genetic, neurotransmitter, immune, oxidative, and inflammatory systems [<xref ref-type="bibr" rid="B4">4</xref>]. Supporting this interpretation, whereas biomarkers for individual abnormalities possess limited predictive validity for MDD, the predictive validity of several composite biomarker assays is particularly high [<xref ref-type="bibr" rid="B5">5</xref>]. For example, one study of 36 patients with MDD showed that a compositive biomarker test—comprising nine individual biomarker assays (α1 antitrypsin, apolipoprotein CIII, myeloperoxidase, soluble tumor necrosis factor α (TNFα) receptor type II, epidermal growth factor, cortisol, brain-derived neurotropic factor, prolactin, and resistin)—had 91.7% sensitivity and 81.3% specificity for MDD [<xref ref-type="bibr" rid="B6">6</xref>]. A follow-up study involving a distinct sample of 34 MDD patients and using the same composite assay, replicated these results with a high degree of precision: 91.1% sensitivity, 81.0% specificity [<xref ref-type="bibr" rid="B6">6</xref>].</p><p>Oxidative stress and neuroinflammation are implicated in the neurobiology of MDD [<xref ref-type="bibr" rid="B7">7</xref>-<xref ref-type="bibr" rid="B14">14</xref>] (recently reviewed by our group [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B15">15</xref>-<xref ref-type="bibr" rid="B19">19</xref>]). Neuropathological studies comparing brain tissue from individuals with MDD to that from non-depressed controls have documented associations between MDD and (a) decreased levels of antioxidants, such as glutathione [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B16">16</xref>] and (b) increased levels of lipid peroxidation end products, such as 4-hydroxy-2-nonenal [<xref ref-type="bibr" rid="B8">8</xref>]. Studies assessing peripheral markers of oxidative stress have reported similar findings, including: (a) altered activity of antioxidant enzymes, such as glutathione peroxidase, catalase, superoxide dismutase 1, (b) increased activity of pro-oxidant enzymes such as, xanthine oxidase, (c) increased activity of inducible nitric oxide synthase (iNOS) in leukocytes, (d) increased levels of superoxide (O<sub>2</sub><sup>-</sup>), and (e) increased levels of 8-hydroxy-2-deoxyguanosine (a marker for oxidative damage to DNA) [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B12">12</xref>]. Evidence deriving from genetic, neuropathological, cerebrospinal fluid, and serum studies in humans with MDD and from animal models of depressive-like behavior and chronic stress reveal numerous neuroinflammatory abnormalities in MDD, including [<xref ref-type="bibr" rid="B4">4</xref>]: (a) microglial activation [<xref ref-type="bibr" rid="B17">17</xref>-<xref ref-type="bibr" rid="B19">19</xref>], (b) astroglial loss and activation [<xref ref-type="bibr" rid="B20">20</xref>,<xref ref-type="bibr" rid="B21">21</xref>], (c) upregulated ratios of T helper 1 (Th1) cells and proinflammatory cytokines [<xref ref-type="bibr" rid="B22">22</xref>-<xref ref-type="bibr" rid="B24">24</xref>], and (d) decreased CD4<sup>+</sup>CD25<sup>+</sup>FOXP3<sup>+</sup> regulatory T (T<sub>Reg</sub>) cell counts [<xref ref-type="bibr" rid="B25">25</xref>]. Both oxidative stress and neuroinflammation may contribute to decreased serotonergic and increased glutamatergic tone, and increased glutamatergic tone may in turn contribute to oxidative stress and neuroinflammation in a positive feedback loop [<xref ref-type="bibr" rid="B4">4</xref>]. In addition, experimental evidence suggests that increased reactive oxygen species (ROS) synthesis (oxidative stress) and neuroinflammation themselves exhibit a bidirectional relationship (Figure <xref ref-type="fig" rid="F1">1</xref>). Indeed, ROS can activate microglia and increase proinflammatory cytokine synthesis—for example, by stimulating transcription factor nuclear factor κB (NFκB)—whereas activated microglia and proinflammatory cytokines can in turn perpetuate oxidative stress [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B26">26</xref>-<xref ref-type="bibr" rid="B28">28</xref>].</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Putative mechanisms involving the synthesis of reactive oxygen species (ROS) and their bidirectional interaction with neuroinflammation in major depressive disorder.</bold> This figure shows potential mechanistic links among ROS, inflammation, and hyperglutamatergia. Abbreviations: BBB, blood-brain barrier; COX2, cyclo-oxygenase 2; CRH, corticotropin-releasing hormone; eNOS, endothelial nitric oxide synthase; iNOS, inducible nitric oxide synthase; MMP, matrix metalloproteinase; NAD(P)H, nicotinamide adenosine dinucleotide phosphate; NMDAR, <italic>N</italic>-methyl-<sc>D</sc>-aspartate receptor; NO, nitric oxide; PLA2, phospholipase A2.</p></caption><graphic xlink:href="1742-2094-10-142-1"/></fig><p>Collectively, data from postmortem neuropathological human studies and <italic>in vivo</italic> neuroimaging human and animal studies provide strong evidence of neurovascular unit dysfunction with blood-brain barrier (BBB) hyperpermeability in association with oxidative stress and neuroinflammation in selected neurological disorders, such as stroke, epilepsy, Alzheimer’s disease, traumatic brain injury, and multiple sclerosis [<xref ref-type="bibr" rid="B29">29</xref>-<xref ref-type="bibr" rid="B43">43</xref>] (Table <xref ref-type="table" rid="T1">1</xref>). In these disorders, BBB breakdown, oxidative stress, and inflammation are thought to impair neuronal function [<xref ref-type="bibr" rid="B44">44</xref>]. MDD, in contrast to other major psychiatric disorders, is frequently comorbid with such neurological disorders as well as disorders characterized by vascular endothelial dysfunction, such as cardiovascular disease and diabetes mellitus [<xref ref-type="bibr" rid="B45">45</xref>-<xref ref-type="bibr" rid="B52">52</xref>]. Whether neurovascular dysfunction with BBB hyperpermeability occurs in primary MDD (without neurological comorbidity), however, remains less clear.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Putative mechanisms of neurovascular dysfunction and blood–brain barrier hyperpermeability in major depressive disorder in the context of established mechanisms in various neurological disorders</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="left" valign="bottom"><bold>Mechanisms</bold><hr/></th><th colspan="4" align="center" valign="bottom"><bold>Major depressive disorder</bold><hr/></th><th colspan="2" align="center" valign="bottom"><bold>Neurological disorders</bold><hr/></th></tr><tr><th align="left"> </th><th align="center"><bold>Human</bold></th><th align="center"><bold>Sources</bold></th><th align="center"><bold>Animal</bold></th><th align="center"><bold>Sources</bold></th><th align="center"><bold>Human</bold></th><th align="center"><bold>Sources</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom"><bold>Oxidative stress</bold><hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">eNOS uncoupling, decreased NO<hr/></td><td align="center" valign="bottom">■<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B53">53</xref>-<xref ref-type="bibr" rid="B59">59</xref>]<hr/></td><td align="center" valign="bottom">■<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B60">60</xref>]<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B61">61</xref>-<xref ref-type="bibr" rid="B63">63</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Increased ROS synthesis<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B14">14</xref>-<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B64">64</xref>-<xref ref-type="bibr" rid="B84">84</xref>]<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B64">64</xref>,<xref ref-type="bibr" rid="B85">85</xref>]<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B42">42</xref>,<xref ref-type="bibr" rid="B63">63</xref>,<xref ref-type="bibr" rid="B86">86</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Cerebral hypoperfusion<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B87">87</xref>-<xref ref-type="bibr" rid="B91">91</xref>]<hr/></td><td align="center" valign="bottom">N/A<hr/></td><td align="center" valign="bottom">…<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B92">92</xref>-<xref ref-type="bibr" rid="B96">96</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">MMP activation<hr/></td><td align="center" valign="bottom">■<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B97">97</xref>]<hr/></td><td align="center" valign="bottom">?<hr/></td><td align="center" valign="bottom">…<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B39">39</xref>,<xref ref-type="bibr" rid="B42">42</xref>,<xref ref-type="bibr" rid="B98">98</xref>-<xref ref-type="bibr" rid="B101">101</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Decreased E-cadherin activity<hr/></td><td align="center" valign="bottom">?<hr/></td><td align="center" valign="bottom">…<hr/></td><td align="center" valign="bottom">?<hr/></td><td align="center" valign="bottom">…<hr/></td><td align="center" valign="bottom">?<sup>a</sup><hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B102">102</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Tight junction alteration<hr/></td><td align="center" valign="bottom">?<hr/></td><td align="center" valign="bottom">…<hr/></td><td align="center" valign="bottom">?<hr/></td><td align="center" valign="bottom">…<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B41">41</xref>,<xref ref-type="bibr" rid="B103">103</xref>-<xref ref-type="bibr" rid="B106">106</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Endothelial cytoskeletal alteration<hr/></td><td align="center" valign="bottom">?<hr/></td><td align="center" valign="bottom">…<hr/></td><td align="center" valign="bottom">?<hr/></td><td align="center" valign="bottom">…<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B31">31</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Increased NMDAR expression<sup>b</sup><hr/></td><td align="center" valign="bottom"><inline-formula><inline-graphic xlink:href="1742-2094-10-142-i2.gif"/></inline-formula><hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B107">107</xref>-<xref ref-type="bibr" rid="B111">111</xref>]<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B40">40</xref>]<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B112">112</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Mitochondrial alterations<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B65">65</xref>,<xref ref-type="bibr" rid="B113">113</xref>-<xref ref-type="bibr" rid="B121">121</xref>]<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B65">65</xref>,<xref ref-type="bibr" rid="B122">122</xref>]<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B123">123</xref>-<xref ref-type="bibr" rid="B125">125</xref>]<hr/></td></tr><tr><td align="left" valign="bottom"><bold>Neuroinflammation</bold><hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Astroglial loss<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B20">20</xref>,<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B126">126</xref>-<xref ref-type="bibr" rid="B133">133</xref>]<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B134">134</xref>-<xref ref-type="bibr" rid="B138">138</xref>]<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B139">139</xref>-<xref ref-type="bibr" rid="B141">141</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Decreased AQP4<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B142">142</xref>]<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B143">143</xref>]<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B144">144</xref>-<xref ref-type="bibr" rid="B146">146</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Microglial activation<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B147">147</xref>,<xref ref-type="bibr" rid="B148">148</xref>]<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B149">149</xref>-<xref ref-type="bibr" rid="B152">152</xref>]<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B42">42</xref>,<xref ref-type="bibr" rid="B98">98</xref>,<xref ref-type="bibr" rid="B153">153</xref>,<xref ref-type="bibr" rid="B154">154</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Proinflammatory cytokines<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B23">23</xref>,<xref ref-type="bibr" rid="B155">155</xref>]<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B156">156</xref>,<xref ref-type="bibr" rid="B157">157</xref>]<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B42">42</xref>,<xref ref-type="bibr" rid="B98">98</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Bradykinin alteration<hr/></td><td align="center" valign="bottom">●<sup>c</sup><hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B158">158</xref>]<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B159">159</xref>]<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B159">159</xref>,<xref ref-type="bibr" rid="B160">160</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Hyperglutamatergia<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B161">161</xref>-<xref ref-type="bibr" rid="B163">163</xref>]<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B164">164</xref>]<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B165">165</xref>-<xref ref-type="bibr" rid="B167">167</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Mast cell activation<hr/></td><td align="center" valign="bottom">●<sup>c</sup><hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B168">168</xref>,<xref ref-type="bibr" rid="B169">169</xref>]<hr/></td><td align="center" valign="bottom">?<hr/></td><td align="center" valign="bottom">…<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B170">170</xref>,<xref ref-type="bibr" rid="B171">171</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Increased ICAM-1 and VCAM-1<hr/></td><td align="center" valign="bottom"><inline-formula><inline-graphic xlink:href="1742-2094-10-142-i3.gif"/></inline-formula><hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B172">172</xref>-<xref ref-type="bibr" rid="B174">174</xref>]<hr/></td><td align="center" valign="bottom">?<hr/></td><td align="center" valign="bottom">…<hr/></td><td align="center" valign="bottom">●<hr/></td><td align="center" valign="bottom">[<xref ref-type="bibr" rid="B175">175</xref>-<xref ref-type="bibr" rid="B178">178</xref>]<hr/></td></tr><tr><td align="left" valign="bottom"><bold>Other mechanisms</bold><hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left">Increased P-glycoprotein activity</td><td align="center">●</td><td align="center">[<xref ref-type="bibr" rid="B179">179</xref>,<xref ref-type="bibr" rid="B180">180</xref>]</td><td align="center">●</td><td align="center">[<xref ref-type="bibr" rid="B181">181</xref>]</td><td align="center">●</td><td align="center">[<xref ref-type="bibr" rid="B179">179</xref>]</td></tr></tbody></table><table-wrap-foot><p>Symbol key: ●, documented in the central nervous system in major depressive disorder; ■, not documented in the central nervous system, but associated with major depressive disorder; ?, insufficient data; <inline-formula><inline-graphic xlink:href="1742-2094-10-142-i1.gif"/></inline-formula>, mixed evidence.</p><p>Abbreviations: <italic>AQP4,</italic> aquaporin 4; <italic>eNOS</italic>, endothelial nitric oxide synthase; <italic>ICAM-1</italic>, intercellular adhesion molecule 1; <italic>NMDAR, N</italic>-methyl-<sc>D</sc>-aspartate receptor; <italic>MMP,</italic> matrix metalloproteinases; <italic>ROS</italic>, reactive oxygen species; <italic>VCAM-1</italic>, vascular cell adhesion molecule 1.</p><p>a. Refers to data that has only been shown in animal models.</p><p>b. Refers to human data in major depressive disorder refers to increased NMDAR expression that was not specific to the endothelium. Human data of NMDAR subunit composition alteration in neurological disorders was shown in cultured human blood–brain barrier endothelial cells. Animal data refer to increased cerebrovascular endothelial NMDAR subunit 1 (NR1) expression upon exposure to oxidative stress (this was not a depressive-like behavior or chronic stress animal model, though this evidence may be relevant to MDD where oxidative stress is documented).</p><p>c. Refers to abnormalities for which only limited data exists.</p></table-wrap-foot></table-wrap><p>Shalev and colleagues have previously reviewed evidence through 2009 linking BBB hyperpermeability to psychiatric disorders generally [<xref ref-type="bibr" rid="B168">168</xref>]. We review emerging clinical and experimental evidence implicating oxidative stress, eNOS uncoupling, and reduced endothelial NO levels in the pathophysiology of peripheral vascular endothelial dysfunction associated with MDD. We present a theoretical integration of human and animal data linking these mechanisms and those involving neuroinflammation to findings suggesting that neurovascular dysfunction can occur in primary MDD. We also discuss putative links between neurovascular dysfunction with BBB hyperpermeability and neuronal signaling abnormalities in MDD.</p><sec><title>Neurovascular unit dysfunction</title><p>The neurovascular unit consists of cerebral microvessels, glial cells (astroglia, microglia, oligodendroglia), and neurons. It is the epicenter of several tightly controlled, dynamic, and complex cellular interactions between glia and neurons, and the coupling of neuronal activity with endothelium-dependent cerebral blood flow [<xref ref-type="bibr" rid="B33">33</xref>]. Evidence of an association between MDD and neurovascular dysfunction is indirect, deriving primarily from studies assessing peripheral vascular endothelial dysfunction in MDD and from epidemiological data associating MDD with vascular disorders.</p><p>One method for evaluating endothelial dysfunction involves measuring the relative uptake ratio (RUR) of blood flow in the brachial artery after hyperemic challenge via dynamic nuclear imaging. RUR is a measure of the vascular dilatory response whereby a lower RUR implies poorer vascular endothelial function. In a prospective cohort involving 23 patients with MDD, 23 with minor depressive disorder, and 277 non-depressed controls, the mean RUR was significantly lower in participants with MDD (unadjusted mean = 3.13, SD = 1.51) or minor depressive disorder (unadjusted mean = 3.38, SD = 1.00) compared with non-depressed controls (unadjusted mean = 4.22, SD = 1.74) (<italic>F</italic> = 6.68, <italic>P</italic> = 0.001) [<xref ref-type="bibr" rid="B182">182</xref>]. This effect remained statistically significant after adjusting for age, sex, socioeconomic factors, medical comorbidity, and medications (<italic>F</italic> = 5.19, <italic>P</italic> = 0.006) [<xref ref-type="bibr" rid="B182">182</xref>]. One study evaluating endothelial proapoptotic activity, defined as the percentage of apoptotic nuclei in human umbilical vein endothelial cells, found a significantly increased percentage of proapoptotic nuclei in participants with MDD compared with non-depressed controls (4.4% vs 2.3%, <italic>P</italic> ≤ 0.001) [<xref ref-type="bibr" rid="B183">183</xref>]. This finding remained statistically significant after adjusting for age and cardiovascular comorbidity.</p><p>Linking vascular endothelial dysfunction to MDD, epidemiological studies reveal a strong and bidirectional association between MDD and medical conditions characterized by vascular endothelial pathology [<xref ref-type="bibr" rid="B184">184</xref>]. A recent meta-analysis involving 16,221 study participants found a significantly increased risk of MDD among individuals with major vascular diseases compared with those without vascular disease: diabetes (odds ratio (OR) 1.51, 95% confidence interval (CI) 1.30 to 1.76, <italic>P</italic> < 0.0005, 15 studies), cardiovascular disease (OR 1.76, 95% CI 1.08 to 1.80, <italic>P</italic> < 0.0005, 10 studies), and stroke (OR 2.11, 95% CI 1.61 to 2.77, <italic>P</italic> < 0.0005, 10 studies) [<xref ref-type="bibr" rid="B45">45</xref>]. The same meta-analysis also found that MDD was more common among individuals with two or more classic risk factors for vascular disease compared with those with one or no risk factors (OR 1.49, 95% CI 1.27 to 1.7, <italic>P</italic> < 0.0005, 18 studies) [<xref ref-type="bibr" rid="B45">45</xref>]. These findings remained robust after statistical adjustments for chronic illness and disability. Results from meta-analyses having assessed the association from the reverse direction, indicate that MDD is not only an independent risk factor for cardiovascular disease (relative risk (RR) 2.69, 95% CI 1.63 to 4.43, <italic>P</italic> < 0<italic>.</italic>001, 11 studies) [<xref ref-type="bibr" rid="B49">49</xref>], but is also associated with a 3-fold increased cardiovascular disease mortality rate (OR 2.61, 95% CI 1.53 to 4.47, <italic>P</italic> = 0.0004) [<xref ref-type="bibr" rid="B48">48</xref>]. Related studies report similar findings [<xref ref-type="bibr" rid="B50">50</xref>-<xref ref-type="bibr" rid="B52">52</xref>].</p></sec><sec><title>Blood–brain barrier unit hyperpermeability</title><p>The BBB consists of the neurovascular endothelium, extracellular matrix basal lamina, and astrocytic end-feet processes. The BBB secures the brain’s immune-privileged status by restricting the entry of peripheral inflammatory mediators (for example, cytokines, antibodies), which can impair neurotransmission [<xref ref-type="bibr" rid="B37">37</xref>,<xref ref-type="bibr" rid="B168">168</xref>,<xref ref-type="bibr" rid="B185">185</xref>,<xref ref-type="bibr" rid="B186">186</xref>]. Neurovascular endothelial cells regulate influx of essential nutrients, efflux of toxic substances, ionic homeostasis of brain interstitial fluid, and prevent brain influx of peripheral neuroactive substances, neurotransmitters, and water-soluble molecules [<xref ref-type="bibr" rid="B185">185</xref>]. Evidence of an association between BBB hyperpermeability and MDD derives mainly from studies having assessed cerebrospinal fluid (CSF)-to-serum ratios of various molecules, as well as evaluations concerning P-glycoprotein.</p><p>Evidence of an elevated CSF-to-serum albumin ratio in some MDD patients is suggestive of mild hyperpermeability of blood-brain and/or blood-CSF barriers [<xref ref-type="bibr" rid="B186">186</xref>,<xref ref-type="bibr" rid="B187">187</xref>]. A cross-sectional study of elderly women without dementia (11 MDD, 3 dysthymia, 70 non-depressed controls) found an elevated mean CSF-to-serum albumin ratio among those with MDD or dysthymia relative to non-depressed controls (7.1 × 10<sup>-3</sup> vs 5.4 × 10<sup>-3</sup>, age-adjusted <italic>P</italic> < 0.015) [<xref ref-type="bibr" rid="B186">186</xref>]. Another study (24 affective disorders, 4,100 age-matched controls) found an increased mean CSF-to-serum albumin ratio among 37.5% of the affective disorder group (9 of 24); this value was 22% to 89% above the upper limit of healthy age-matched controls (8.7 × 10<sup>-3</sup> vs 5.0 × 10<sup>-3</sup>) [<xref ref-type="bibr" rid="B187">187</xref>]. A third study (99 MDD) found that increased CSF-to-serum ratios of albumin and urate were positively associated with EEG slowing (a measure of cerebral dysfunction) and suicidality [<xref ref-type="bibr" rid="B188">188</xref>]. Elevated levels of S100B protein (a marker of glial activation) [<xref ref-type="bibr" rid="B189">189</xref>,<xref ref-type="bibr" rid="B190">190</xref>] and proinflammatory cytokines [<xref ref-type="bibr" rid="B23">23</xref>,<xref ref-type="bibr" rid="B191">191</xref>] in the serum, CSF, and neuropathological specimens from persons with MDD may be related to increased permeability of blood-brain and blood-CSF barriers. Elevated levels of these molecules may reflect their increased synthesis and increased efflux from (a) brain parenchyma into the blood (BBB hyperpermeability) [<xref ref-type="bibr" rid="B168">168</xref>,<xref ref-type="bibr" rid="B184">184</xref>], and (b) blood into the CSF (blood-CSF hyperpermeability).</p><p>Alteration of BBB endothelial expression of P-glycoprotein (a multidrug efflux transporter) is documented in some persons with MDD [<xref ref-type="bibr" rid="B192">192</xref>]. Reduced expression or function of P-glycoprotein may facilitate BBB permeability to neurotoxic substances [<xref ref-type="bibr" rid="B192">192</xref>]. Positron emission tomography (PET) utilizing the [(11)C]-verapamil radioligand for P-glycoprotein in humans with MDD and in Wistar rats exhibiting depressive-like behavior showed that chronic stress exposure and administration of antidepressants inhibited and enhanced P-glycoprotein function, respectively [<xref ref-type="bibr" rid="B179">179</xref>,<xref ref-type="bibr" rid="B181">181</xref>]. A human genetics study (631 MDD, 110 non-depressed controls) revealed a significant association between alteration of the P-glycoprotein encoding gene ATP-binding cassette, subfamily B member 1 (ABCB1) and MDD (<italic>P</italic> = 0.034) [<xref ref-type="bibr" rid="B180">180</xref>].</p></sec><sec><title>Theoretical integration with oxidative and neuroinflammatory mechanisms</title></sec><sec><title>Oxidative stress</title><p>Common ROS include superoxide (O<sub>2</sub><sup>-</sup>), hydroxyl radical (HO<sup>-</sup>), hydrogen peroxide (H<sub>2</sub>O<sub>2</sub><sup>-</sup>), and peroxynitrite (ONOO<sup>-</sup>). ONOO<sup>-</sup> is a highly reactive oxidant generated by the reaction of nitric oxide (NO) with O<sub>2</sub><sup>-</sup>[<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B123">123</xref>]. The brain is particularly susceptible to oxidative stress due to high levels of peroxidizable polyunsaturated fatty acids and transition minerals (reduced form) that induce lipid peroxidation and convert H<sub>2</sub>O<sub>2</sub><sup>-</sup> to HO<sup>-</sup>; additionally, the brain’s oxygen demand is particularly high and the presence of antioxidant defense mechanisms is relatively limited [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B12">12</xref>].</p><p>Although ROS can limit injury and promote recovery at low levels, ROS facilitate oxidative injury at high levels by damaging biological macromolecules, such as lipids, proteins, and DNA [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B12">12</xref>]. We hypothesize that oxidative stress associated with MDD may impair neurovascular function through several mechanisms, with an emphasis on mechanisms that can shift the functional balance between beneficial endothelial nitric oxide synthase (eNOS)-generated NO versus harmful eNOS-generated O<sub>2</sub><sup>-</sup> (Figure <xref ref-type="fig" rid="F2">2</xref> and Table <xref ref-type="table" rid="T1">1</xref>).</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Theoretical integration of the human and animal data linking oxidative stress, eNOS uncoupling, low endothelial NO levels, and neuroinflammation to indirect evidence of functional and structural abnormalities of neurovascular unit in major depressive disorder.</bold> Adapted with permission from Abbott <italic>et al</italic>., [<xref ref-type="bibr" rid="B185">185</xref>]. This figure describes several putative mechanisms involving neuroinflammation, oxidative stress, endothelial nitric oxide synthase uncoupling, and hyperglutamatergia, as well as their relationships to indirect evidence of neurovascular dysfunction in MDD. Neurovascular endothelial lipofuscin granule accumulation is a marker of endothelial oxidative stress, which we recently documented by ultrastructural analysis of cerebral microvasculature in brain biopsy from a patient with chronic refractory MDD [<xref ref-type="bibr" rid="B90">90</xref>]. Abbreviations: AQP4, aquaporin 4; BH<sub>2</sub>: dihydrobiopterin; BH<sub>4</sub>, tetrahydrobiopterin; CRH, corticotropin-releasing hormone; eNOS, endothelial nitric oxide synthase; mGluR, metabotropic glutamate receptor; MDD, major depressive disorder; MMP, matrix metalloproteinase; NAD(P)H, nicotinamide adenosine dinucleotide phosphate; Na<sup>+</sup>/K<sup>+</sup> ATPase, sodium-potassium adenosine triphosphatase; NFκB, nuclear factor κB; NMDAR, <italic>N</italic>-methyl-<sc>d</sc>-aspartate receptor; NO, nitric oxide, ONOO<sup>-</sup>, peroxynitrite; O<sub>2</sub><sup>-</sup>, superoxide; ROS, reactive oxygen species.</p></caption><graphic xlink:href="1742-2094-10-142-2"/></fig><p>NO has been termed ‘Janus faced’ owing to its ability to either protect vascular endothelial cell function in some instances, while impairing it in others [<xref ref-type="bibr" rid="B193">193</xref>]. These differential effects of NO are primarily determined by its cellular source (non-endothelial vs endothelial) and concentration (high vs low). NOS isoforms regulate NO synthesis in the brain. Of these, one is constitutively expressed in endothelial cells and astrocytes (eNOS) [<xref ref-type="bibr" rid="B194">194</xref>,<xref ref-type="bibr" rid="B195">195</xref>] (that is, eNOS), and another is expressed in neurons (neuronal NOS (nNOS)).</p><p>eNOS regulates vascular smooth muscle tone and nNOS modulates neurotransmission. The expression of a third NOS isoform, iNOS, occurs in glial and inflammatory cells and is induced by pathological inflammatory states, such as following trauma [<xref ref-type="bibr" rid="B38">38</xref>]. More recently, a fourth NOS isoform was described, mitochondrial (mtNOS), which is an eNOS-like isoform that is constitutively expressed in the inner mitochondrial membrane [<xref ref-type="bibr" rid="B196">196</xref>,<xref ref-type="bibr" rid="B197">197</xref>]. When combined with O<sub>2</sub><sup>-</sup>, NO produced by non-endothelial cellular sources (as regulated by nNOS, iNOS) can impair the vascular endothelium and disrupt BBB integrity [<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B53">53</xref>]. nNOS activity itself is positively regulated by Ca<sup>2+</sup> influx [<xref ref-type="bibr" rid="B198">198</xref>], whereas iNOS activity is positively regulated by proinflammatory cytokine [<xref ref-type="bibr" rid="B199">199</xref>] and NFκB signaling [<xref ref-type="bibr" rid="B200">200</xref>].</p><p>NO produced by endothelial cells (as regulated by eNOS) increases cellular levels of cyclic guanosine monophosphate, which can increase cerebral blood flow via mechanisms involving endothelium-dependent vasodilation and platelet aggregation inhibition [<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B53">53</xref>,<xref ref-type="bibr" rid="B201">201</xref>]. <italic>In vitro</italic> studies showed that endothelial-derived NO may dilate cerebral vessels by inhibiting the synthesis of 20-hydroxyeicostetranoic acid—an arachidonic acid metabolite that promotes vasoconstriction [<xref ref-type="bibr" rid="B202">202</xref>,<xref ref-type="bibr" rid="B203">203</xref>]. Endothelial-derived NO can also limit endothelial vascular oxidative stress injury by scavenging free radicals [<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B53">53</xref>]. Endothelial eNOS mediates NO synthesis via oxidative conversion of <sc>l</sc>-arginine to <sc>l</sc>-citrulline. Activity of eNOS is modulated by several factors, including endothelial levels of Ca<sup>2+</sup>, arginine (eNOS substrate) [<xref ref-type="bibr" rid="B204">204</xref>], as well as tetrahydrobiopterin (BH<sub>4</sub>) (eNOS cofactor) [<xref ref-type="bibr" rid="B53">53</xref>-<xref ref-type="bibr" rid="B55">55</xref>,<xref ref-type="bibr" rid="B201">201</xref>,<xref ref-type="bibr" rid="B205">205</xref>,<xref ref-type="bibr" rid="B206">206</xref>] (Figure <xref ref-type="fig" rid="F2">2</xref>). Downregulation of eNOS activity can decrease endothelial NO levels, potentially resulting in (a) reduced cerebral blood flow, (b) increased platelet aggregation, which may contribute to the increased risk of cardiovascular disease in MDD, (c) increased oxidative stress, and (d) decreased vascular reactivity [<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B53">53</xref>,<xref ref-type="bibr" rid="B201">201</xref>].</p><p>Under oxidative conditions, such as those associated with MDD [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B15">15</xref>] (Figure <xref ref-type="fig" rid="F1">1</xref>), endothelial levels of BH<sub>4</sub> are decreased due to increased oxidative conversion of BH<sub>4</sub> to dihydrobiopterin (BH<sub>2</sub>). Decreased endothelial levels of BH<sub>4</sub> and increased endothelial levels of BH<sub>2</sub> (which can also reduce BH<sub>4</sub> binding to eNOS) uncouple <sc>l</sc>-arginine oxidation from the electron transfer process and shift the eNOS substrate from <sc>l</sc>-arginine to molecular oxygen (that is, eNOS uncoupling), thereby promoting the synthesis of harmful O<sub>2</sub><sup>-</sup> instead of beneficial NO [<xref ref-type="bibr" rid="B53">53</xref>-<xref ref-type="bibr" rid="B55">55</xref>,<xref ref-type="bibr" rid="B205">205</xref>,<xref ref-type="bibr" rid="B207">207</xref>,<xref ref-type="bibr" rid="B208">208</xref>]. Once formed, O<sub>2</sub><sup>-</sup> reacts with residual NO (still being produced at a lower rate) to form ONOO<sup>- </sup>[<xref ref-type="bibr" rid="B205">205</xref>]. ONOO<sup>-</sup> in turn oxidizes BH<sub>4</sub>, thereby further decreasing its levels in a positive feedback loop [<xref ref-type="bibr" rid="B54">54</xref>,<xref ref-type="bibr" rid="B205">205</xref>] (Figure <xref ref-type="fig" rid="F2">2</xref>).</p><p>Data from <italic>in vitro</italic> animal models of neurological disorders show that upregulation of iNOS and nNOS expression and downregulation of eNOS expression can worsen neuronal injury [<xref ref-type="bibr" rid="B209">209</xref>-<xref ref-type="bibr" rid="B213">213</xref>]. In murine models of ischemic stroke, knocking out iNOS and nNOS decreased the size of infarct while knocking out eNOS expanded infracted zone, compared to wild-type mice [<xref ref-type="bibr" rid="B214">214</xref>,<xref ref-type="bibr" rid="B215">215</xref>]. In animal models of traumatic brain injury, increased levels of endothelial ONOO<sup>-</sup> are associated with BBB breakdown and neurobehavioral deficits [<xref ref-type="bibr" rid="B209">209</xref>]; additionally, treatment with the antioxidant <italic>S</italic>-nitrosoglutathione enhances neural reparative mechanisms and improves neurovascular unit function by decreasing endothelial ONOO<sup>-</sup> synthesis [<xref ref-type="bibr" rid="B209">209</xref>].</p><p>Clinical and experimental studies suggest that eNOS uncoupling can contribute to vascular endothelial dysfunction in both cardiovascular diseases and MDD [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B53">53</xref>-<xref ref-type="bibr" rid="B55">55</xref>,<xref ref-type="bibr" rid="B182">182</xref>,<xref ref-type="bibr" rid="B205">205</xref>,<xref ref-type="bibr" rid="B206">206</xref>,<xref ref-type="bibr" rid="B216">216</xref>]. In cardiovascular diseases, eNOS uncoupling-mediated endothelial dysfunction is thought to result from (a) increased O<sub>2</sub><sup>-</sup> synthesis (through an NAD(P)H oxidase-dependent mechanism), (b) increased ONOO<sup>-</sup> formation, and (c) decreased BH<sub>4</sub> levels [<xref ref-type="bibr" rid="B54">54</xref>,<xref ref-type="bibr" rid="B55">55</xref>,<xref ref-type="bibr" rid="B182">182</xref>,<xref ref-type="bibr" rid="B206">206</xref>]. In MDD, however, the potential contribution of eNOS uncoupling to vascular endothelial dysfunction is inferred from less direct evidence. For example, several clinical studies of persons with MDD have shown significant reductions in eNOS activity and NO levels in platelets and sera, respectively [<xref ref-type="bibr" rid="B53">53</xref>-<xref ref-type="bibr" rid="B57">57</xref>]. In a study of 57 MDD patients randomized to either citalopram (<italic>n</italic> = 36) or placebo (<italic>n</italic> = 21), a 3-month trial of citalopram was associated with a statistically significant increase in serum NO levels compared to placebo (<italic>P</italic> = 0.005) [<xref ref-type="bibr" rid="B58">58</xref>]. Another study involving a 2-month trial of paroxetine reproduced similar results [<xref ref-type="bibr" rid="B59">59</xref>]. Fluoxetine treatment in a chronic stress mouse model restored previously deficient aortic endothelial NO levels [<xref ref-type="bibr" rid="B60">60</xref>], suggesting that eNOS uncoupling may not only occur in MDD, but also that eNOS recoupling may be one of the mechanisms by which antidepressants exert their therapeutic effects [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B14">14</xref>].</p><p>The antidepressant effect of <sc>l</sc>-methylfolate, which can reverse eNOS uncoupling <italic>in vitro</italic> via upregulating BH<sub>4</sub> synthesis [<xref ref-type="bibr" rid="B206">206</xref>], suggests that eNOS uncoupling contributes to the neurobiology of MDD. A randomized controlled trial showed that adding <sc>l</sc>-methylfolate at 15 mg/day, but not at 7.5 mg/day, to a stable regimen of selective serotonin reuptake inhibitors (SSRIs) had superior efficacy to SSRIs plus placebo [<xref ref-type="bibr" rid="B217">217</xref>]. Although the authors attributed BH<sub>4</sub> augmenting the antidepressant effects of SSRIs to direct activation of the rate-limiting enzymes of monoamine synthesis (serotonin, norepinephrine, dopamine), we suggest that these effects may also be related to the ability of BH<sub>4</sub> to reverse eNOS uncoupling.</p><p>Although regionally selective (thalamic nuclei, prefrontal, anterior cingulate, temporal, and occipital cortices) cerebral hypoperfusion abnormalities in MDD have traditionally been attributed to depressed mood states and reduced neuronal activity [<xref ref-type="bibr" rid="B87">87</xref>-<xref ref-type="bibr" rid="B91">91</xref>][<xref ref-type="bibr" rid="B208">208</xref>], these findings may also be related to eNOS uncoupling [<xref ref-type="bibr" rid="B65">65</xref>,<xref ref-type="bibr" rid="B113">113</xref>,<xref ref-type="bibr" rid="B114">114</xref>,<xref ref-type="bibr" rid="B218">218</xref>] (Figure <xref ref-type="fig" rid="F2">2</xref>). Sustained cerebral hypoperfusion can impair endothelial mitochondrial oxidative function, resulting in increased synthesis of endothelial ROS [<xref ref-type="bibr" rid="B219">219</xref>-<xref ref-type="bibr" rid="B222">222</xref>]. ROS can in turn promote eNOS uncoupling, leading to reduced vasodilatory endothelial NO levels and cerebral hypoperfusion in a positive feedback loop [<xref ref-type="bibr" rid="B54">54</xref>,<xref ref-type="bibr" rid="B55">55</xref>,<xref ref-type="bibr" rid="B182">182</xref>,<xref ref-type="bibr" rid="B206">206</xref>]. In addition, SSRIs have been shown to induce vasodilation through eNOS-mediated downregulation of NO [<xref ref-type="bibr" rid="B223">223</xref>]. We recently reported a case of chronic and refractory MDD with moderately severe bifrontal cerebral hypoperfusion (seen via single photon emission tomography (SPECT)) associated with lipofuscin granule accumulation (a marker of oxidative stress [<xref ref-type="bibr" rid="B224">224</xref>-<xref ref-type="bibr" rid="B228">228</xref>]) (Figure <xref ref-type="fig" rid="F2">2</xref>) identified exclusively within the neurovascular unit (predominately within the endothelium) [<xref ref-type="bibr" rid="B90">90</xref>]; restoration of cerebral hypoperfusion in temporal association with intravenous immunoglobulin and minocycline therapy was accompanied with significant improvement of depressive symptoms, after more than 20 years of refractoriness to conventional psychiatric treatments [<xref ref-type="bibr" rid="B90">90</xref>]. We suggest that eNOS uncoupling may occur in MDD primarily as the result of non-heritable factors such as oxidative mechanisms. Indeed, several genetic studies show a non-significant association between eNOS gene polymorphisms and MDD [<xref ref-type="bibr" rid="B229">229</xref>,<xref ref-type="bibr" rid="B230">230</xref>].</p><p>Under oxidative conditions, BBB endothelial cells are not only the source of harmful eNOS uncoupling, but also can be the target of oxidative damage [<xref ref-type="bibr" rid="B39">39</xref>]. In neurological disorders associated with neurovascular dysfunction, oxidative stress can also increase BBB permeability through several mechanisms (Table <xref ref-type="table" rid="T1">1</xref>), which include: (a) activation of metalloproteinase (MMP)-2/9 directly or indirectly through proinflammatory cytokines [<xref ref-type="bibr" rid="B39">39</xref>]; (b) downregulation of endothelial expression of E-cadherin [<xref ref-type="bibr" rid="B38">38</xref>]; (c) alteration of the expression, distribution, and phosphorylation of BBB tight junction proteins (for example, claudin, occluding, ZO proteins) by molecules such as phosphatidylinositol-3-kinase γ [<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B41">41</xref>,<xref ref-type="bibr" rid="B103">103</xref>,<xref ref-type="bibr" rid="B104">104</xref>]; (d) alteration of endothelial cytoskeletal structure; (e) induction of endothelial NMDAR subunit expression such as NMDA receptor subunit 1 (NR1) subunit, leading endothelial excitotoxicity [<xref ref-type="bibr" rid="B40">40</xref>]; and (f) impairment of vascular endothelial mitochondrial oxidative metabolism [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B123">123</xref>]. The relevance of these mechanisms to the neurobiology of MDD, however, remains unclear (Table <xref ref-type="table" rid="T1">1</xref> and Figure <xref ref-type="fig" rid="F2">2</xref>).</p></sec><sec><title>Neuroinflammation</title><p>Neuroinflammation may impair neurovascular function and increase BBB permeability in MDD [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B168">168</xref>] (Figure <xref ref-type="fig" rid="F2">2</xref> and Table <xref ref-type="table" rid="T1">1</xref>). Astroglial cells are an integral part of the neurovascular unit. They are involved in regulating blood flow, BBB permeability, energy metabolism, and neuronal signaling [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B184">184</xref>]. Astroglial loss has been consistently documented in functionally relevant areas (prefrontal and cingulate cortices, amygdala, hippocampus) among persons with MDD [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B142">142</xref>,<xref ref-type="bibr" rid="B168">168</xref>,<xref ref-type="bibr" rid="B231">231</xref>-<xref ref-type="bibr" rid="B236">236</xref>]. Other studies have documented decreased expression of the astroglial end-feet process water channel, aquaporin 4 (AQP4) in the orbitofrontal cortical gray matter (but not white matter) of individuals with MDD relative to non-depressed controls [<xref ref-type="bibr" rid="B142">142</xref>]. Animal models of depressive-like behavior also found decreased AQP4 density in association with oxidative stress [<xref ref-type="bibr" rid="B143">143</xref>]. Decreased AQP4 density may impair critical glial-vascular homeostatic pathways within the neurovascular unit and increase BBB permeability (Figure <xref ref-type="fig" rid="F2">2</xref>). Reduced AQP4 density may also contribute to cerebral perfusion and metabolic abnormalities detected by SPECT and PET imaging in human MDD [<xref ref-type="bibr" rid="B184">184</xref>].</p><p>Microglia provide immune surveillance and regulate developmental synaptic pruning of the brain [<xref ref-type="bibr" rid="B237">237</xref>]. Although transient microglial activation and proliferation (MAP) can limit neuronal injury and enhance recovery (beneficial phenotype), persistent MAP can induce and exacerbate neuronal injury (harmful phenotype) [<xref ref-type="bibr" rid="B238">238</xref>]. Harmful MAP is implicated in the pathophysiology of MDD [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B19">19</xref>], though neuropathological evidence of MAP in the brains of subjects with MDD is inconsistent [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B148">148</xref>,<xref ref-type="bibr" rid="B239">239</xref>]. One neuropathological study found a positive association between suicidality and both MAP density and microglial quinolinic acid expression [<xref ref-type="bibr" rid="B17">17</xref>]. In rats, chronic psychological stress promotes MAP in the prefrontal cortex, amygdala, and hippocampus [<xref ref-type="bibr" rid="B19">19</xref>]. Recent meta-analysis in MDD patients confirmed elevation of serum levels of proinflammatory cytokines, such as interleukin 6 (IL-6) and TNFα [<xref ref-type="bibr" rid="B23">23</xref>,<xref ref-type="bibr" rid="B240">240</xref>]. Multiple <italic>in vitro</italic> studies of various neurological conditions showed that MAP and proinflammatory cytokines could increase BBB permeability [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B38">38</xref>-<xref ref-type="bibr" rid="B40">40</xref>,<xref ref-type="bibr" rid="B168">168</xref>,<xref ref-type="bibr" rid="B184">184</xref>,<xref ref-type="bibr" rid="B241">241</xref>] (Figures <xref ref-type="fig" rid="F1">1</xref> and <xref ref-type="fig" rid="F2">2</xref>) (Table <xref ref-type="table" rid="T1">1</xref>). BBB hyperpermeability may in turn increase crosstalk between innate and adaptive immunity, thereby resulting in further upregulation of MAP and brain cytokine production in a positive feedback loop [<xref ref-type="bibr" rid="B242">242</xref>]. MAP can activate iNOS [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B26">26</xref>,<xref ref-type="bibr" rid="B27">27</xref>], increase ROS synthesis [<xref ref-type="bibr" rid="B28">28</xref>], and promote COX2 expression within the neurovascular unit [<xref ref-type="bibr" rid="B4">4</xref>]; these factors may increase BBB permeability <italic>in vitro </italic>[<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B53">53</xref>]. MAP and proinflammatory cytokines can release and activate matrix metalloproteinases (MMPs) [<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B39">39</xref>,<xref ref-type="bibr" rid="B168">168</xref>], which have been shown <italic>in vitro</italic> to disrupt BBB endothelial tight junction proteins and increase BBB opening [<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B39">39</xref>,<xref ref-type="bibr" rid="B168">168</xref>,<xref ref-type="bibr" rid="B184">184</xref>]. Serum MMP-9 levels have been shown to correlate with depressive symptom severity in humans (as assessed by the Hamilton Depression Scale) [<xref ref-type="bibr" rid="B97">97</xref>]. Highly reproducible <italic>in vitro</italic> data showed that proinflammatory cytokines (TNFα, IL-1β, interferon γ (IFNγ)) can cause a dose-dependent increase in BBB permeability by inducing expression of intercellular adhesion molecule 1 (ICAM-1) on the luminal surface of BBB endothelial cells in animals [<xref ref-type="bibr" rid="B243">243</xref>-<xref ref-type="bibr" rid="B249">249</xref>] and humans [<xref ref-type="bibr" rid="B250">250</xref>,<xref ref-type="bibr" rid="B251">251</xref>]. One neuropathological study found a significant increase in the ICAM-1 expression in the deep white matter of the dorsolateral prefrontal cortex in MDD relative to controls [<xref ref-type="bibr" rid="B172">172</xref>]. Another study showed SSRIs can reduce vascular endothelial expression and serum levels of both ICAM-1 and vascular cell adhesion molecule 1 (VCAM-1) [<xref ref-type="bibr" rid="B173">173</xref>]. Thus, increased BBB endothelial cell expression of adhesion molecules may be one mechanism by which BBB hyperpermeability occurs in MDD [<xref ref-type="bibr" rid="B174">174</xref>,<xref ref-type="bibr" rid="B252">252</xref>] (Figure <xref ref-type="fig" rid="F2">2</xref>). However, contrary to this interpretation, a separate postmortem study has shown decreased expression of VCAM-1 and ICAM-1 in the orbitofrontal cortex in depressed subjects compared with non-depressed controls [<xref ref-type="bibr" rid="B174">174</xref>]. Increased TNFα production occurring after acute myocardial infarction is associated with an increased risk of MDD and BBB endothelial hyperpermeability [<xref ref-type="bibr" rid="B241">241</xref>]. <italic>In vitro</italic> animal studies showed that TNFα could reduce mitochondrial density and impair mitochondrial oxidative metabolism, leading to increased ROS synthesis [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B253">253</xref>]. Several lines of human [<xref ref-type="bibr" rid="B65">65</xref>,<xref ref-type="bibr" rid="B113">113</xref>-<xref ref-type="bibr" rid="B121">121</xref>] and animal [<xref ref-type="bibr" rid="B65">65</xref>,<xref ref-type="bibr" rid="B122">122</xref>] evidence implicate mitochondrial abnormalities in MDD. <italic>In vitro</italic> data mechanistically link mitochondrial abnormalities to oxidative injury-related vascular abnormalities [<xref ref-type="bibr" rid="B219">219</xref>] (Figures <xref ref-type="fig" rid="F1">1</xref> and <xref ref-type="fig" rid="F2">2</xref>). Thus, proinflammatory cytokines may also induce depression and increase BBB permeability by promoting oxidative stress and impairing mitochondrial functions. The relevance of these mechanisms to MDD, however, remains unproven.</p><p>Bradykinin is a polypeptide that mediates inflammation, vasodilation, and increased capillary permeability. Human data of bradykinin alterations in MDD are limited to evidence of functional single nucleotide polymorphisms of the bradykinin receptor B2 gene (BDKRB2) [<xref ref-type="bibr" rid="B158">158</xref>] (Table <xref ref-type="table" rid="T1">1</xref>). LPS-induced depressive-like behavior in mice was associated with upregulation of bradykinin activity and bradykinin B1 receptor expression [<xref ref-type="bibr" rid="B159">159</xref>]; further, selective bradykinin B1 receptor antagonists improved depression-like behavior [<xref ref-type="bibr" rid="B159">159</xref>]. Activation of bradykinin and its inducible B1 and constitutively expressed B2 receptors induces inflammation, promotes oxidative injury, and increases BBB permeability [<xref ref-type="bibr" rid="B160">160</xref>] (Figures <xref ref-type="fig" rid="F1">1</xref> and <xref ref-type="fig" rid="F2">2</xref>). Bradykinin activation can augment the astroglial NFκB pathway-mediated IL-6 production, which may increase BBB permeability [<xref ref-type="bibr" rid="B168">168</xref>,<xref ref-type="bibr" rid="B184">184</xref>]. Bradykinin activation can also stimulate phospholipase A2 activity, which in turn enhances arachidonic acid release and its metabolism, leading to increased malondialdehyde [<xref ref-type="bibr" rid="B12">12</xref>] and NO production [<xref ref-type="bibr" rid="B38">38</xref>] that may increase BBB permeability. Activation of B2 receptor increases endothelial Ca<sup>2+</sup> influx, which can activate pro-oxidant enzymes involved in ROS synthesis [<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B168">168</xref>,<xref ref-type="bibr" rid="B184">184</xref>]. Increased ROS production can increase BBB permeability and its susceptibility to the harmful effects of bradykinin [<xref ref-type="bibr" rid="B12">12</xref>]. <italic>In vitro</italic> human studies showed that inflammation-related upregulation of BBB endothelial bradykinin B1 receptor expression could increase BBB permeability [<xref ref-type="bibr" rid="B160">160</xref>].</p><p>Glutamatergic hyperfunction may contribute to neurovascular dysfunction in MDD (Figure <xref ref-type="fig" rid="F2">2</xref> and Table <xref ref-type="table" rid="T1">1</xref>). Numerous experimental paradigms such as, brain proton magnetic resonance imaging, postmortem brain investigations, and CSF studies, have documented glutamatergic hyperfunction in persons with MDD [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B161">161</xref>,<xref ref-type="bibr" rid="B162">162</xref>]. Neuroinflammation may contribute to hyperglutamatergia in a positive feedback loop through several potential mechanisms, which include: (a) inhibition and reversal of astroglial excitatory amino acid transporter-mediated glutamate reuptake function (this process mediates more than 90% of glutamate uptake [<xref ref-type="bibr" rid="B254">254</xref>]); (b) stimulation of microglial synthesis of quinolinic acid, which can promote synaptosomal glutamate release and increase astroglial glutamate and <sc>d</sc>-serine release; and (c) upregulation of MAP expression of X<sub>c</sub> antiporter system, which increases microglial glutamate release [<xref ref-type="bibr" rid="B4">4</xref>]. Postmortem investigations of <italic>N</italic>-methyl-<sc>d</sc>-aspartate receptors (NMDARs) subunit expression in the brains of MDD subjects compared with those of non-depressed controls show (a) an increase or no change of NR1 subunit expression in the hippocampus [<xref ref-type="bibr" rid="B107">107</xref>-<xref ref-type="bibr" rid="B109">109</xref>], (b) an increase of NR2A and NR2B subunit expression in the hippocampus [<xref ref-type="bibr" rid="B107">107</xref>,<xref ref-type="bibr" rid="B108">108</xref>], (c) a decrease or no change in NR1 subunit expression in the prefrontal cortex [<xref ref-type="bibr" rid="B110">110</xref>,<xref ref-type="bibr" rid="B111">111</xref>], (d) a decrease of NR2A and NR2B subunit expression in the prefrontal cortex [<xref ref-type="bibr" rid="B110">110</xref>], and (e) an increase of NR2A subunit expression in the lateral amygdalae [<xref ref-type="bibr" rid="B255">255</xref>]. Binding of excess glutamate to its dysregulated BBB endothelial ionic NMDARs and metabotropic glutamate receptors (mGluRs) can increase intracellular Ca<sup>2+</sup> level-dependent oxidative stress and BBB permeability via increasing Ca<sup>2+</sup> influx and release from endoplasmic reticulum stores, respectively [<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B40">40</xref>,<xref ref-type="bibr" rid="B159">159</xref>,<xref ref-type="bibr" rid="B256">256</xref>]. Animal data showed that NMDAR activation facilitates free radical production such as ONOO<sup>-</sup>[<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B40">40</xref>,<xref ref-type="bibr" rid="B256">256</xref>] (Figures <xref ref-type="fig" rid="F1">1</xref> and <xref ref-type="fig" rid="F2">2</xref>). Administration of glutamate receptor antagonists has been shown to attenuate NMDAR-induced oxidative stress [<xref ref-type="bibr" rid="B40">40</xref>]. Animal studies showed that oxidative stress in turn can alter cerebral endothelial NMDAR subunit composition and upregulate NR1 subunit expression [<xref ref-type="bibr" rid="B40">40</xref>], thus setting up a positive feedback loop that increases BBB endothelium vulnerability to both glutamate excitotoxicity and oxidative stress [<xref ref-type="bibr" rid="B40">40</xref>]. Alteration of endothelial NMDAR subunit compositions may also reduce cerebral blood flow, as physiologic activation of endothelial NMDAR may activate eNOS and increase endothelial-derived NO [<xref ref-type="bibr" rid="B256">256</xref>]. BBB breakdown may also increase CNS glutamate levels via disruption of endothelial-bound glutamate efflux transporters [<xref ref-type="bibr" rid="B44">44</xref>]; in turn, hyperglutamatergia may heighten BBB susceptibility to the harmful effects of bradykinin. Administration of glutamate receptor antagonists can block bradykinin-induced endothelial Ca<sup>2+</sup> rise [<xref ref-type="bibr" rid="B38">38</xref>]. Thus, BBB hyperpermeability, increased endothelial NMDAR expression, and increased CNS glutamate levels may contribution to neuronal dysfunction in MDD.</p><p>Mast cells are tissue-bound granulated cells most commonly found in the skin and gastrointestinal tract. They, like basophils, contain high levels of histamine and heparin. In the brain, mast cells are particularly abundant in the hypothalamic region. Mast cell activation has been associated with MDD [<xref ref-type="bibr" rid="B169">169</xref>] (Table <xref ref-type="table" rid="T1">1</xref>). Approximately 40% to 70% of persons with mastocytosis, an uncommon and heterogeneous syndrome characterized by increased mast cell density, exhibit depressive symptoms [<xref ref-type="bibr" rid="B257">257</xref>]. Increased corticotropin-releasing hormone (CRH) secretion may contribute to mast cell activation associated with MDD [<xref ref-type="bibr" rid="B168">168</xref>,<xref ref-type="bibr" rid="B170">170</xref>,<xref ref-type="bibr" rid="B171">171</xref>]. Experimental evidence suggests that mast cells can cause inflammation [<xref ref-type="bibr" rid="B170">170</xref>], modulate BBB permeability [<xref ref-type="bibr" rid="B170">170</xref>], and facilitate NMDAR-induced neuronal excitotoxicity [<xref ref-type="bibr" rid="B170">170</xref>] (Figure <xref ref-type="fig" rid="F2">2</xref>). Mast cell activation can release inflammatory substances (for example, IL-6, TNFα, vascular endothelial growth factor) and stimulate vascular endothelial cell adhesion molecule expression [<xref ref-type="bibr" rid="B170">170</xref>]. These molecules can disrupt BBB integrity and enhance inflammatory cell transmigration into the brain [<xref ref-type="bibr" rid="B170">170</xref>].</p></sec><sec><title>Future Directions</title><p>Human and animal studies are needed to evaluate the validity of the BBB dysfunction hypothesis and to explore the mechanistic links between oxidative stress, eNOS uncoupling, and neuroinflammation and neurovascular unit dysfunction with BBB hyperpermeability in MDD. Future postmortem studies investigating the relationship between neurovascular unit dysfunction with BBB hyperpermeability and MDD should focus primarily on the neuroanatomical regions where astroglial loss and MAP have been documented in MDD brains such as anterior mid/cingulate cortex, prefrontal cortex, amygdala, and white matter [<xref ref-type="bibr" rid="B4">4</xref>]. Developing methods with increased sensitivity to detect and quantitate subtle BBB hyperpermeability in MDD are likely to be informative [<xref ref-type="bibr" rid="B37">37</xref>]. These methods might utilize fluorescent dyes in animal models of depressive-like behavior similar to those developed for <italic>in vivo</italic> imaging of specific neurovascular elements in animal models of various neurological disorders associated with neurovascular dysfunction [<xref ref-type="bibr" rid="B43">43</xref>]: sulforhodamine 101 dye, Ca<sup>2+</sup> sensitive dyes, glial fibrillary acidic protein (GFAP), AQP4 (astroglia), CX3C chemokine receptor 1 (CX3CR1) (microglia), dextran-conjugated dyes, alpha SMA-RFPcherry (pericytes), dextran dyes, Tie2 (vasculature) and Thy1 (neurons) [<xref ref-type="bibr" rid="B43">43</xref>]. A promising neuroimaging modality for visualizing MAP in humans with psychiatric illnesses is PET imaging utilizing microglial peripheral benzodiazepine receptor (also known as translocator protein) C11-PK11195 radioligand [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B258">258</xref>-<xref ref-type="bibr" rid="B260">260</xref>]. We suspect that various neurovascular processes particularly those promoting endothelial (and potentially astroglial) eNOS dysfunction may emerge as key targets for cellular and molecular research in MDD. Adequately powered randomized controlled trials investigating the effects of anti-inflammatory agents and antioxidants in MDD [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B90">90</xref>] should also assess their effects on cerebral microvascular endothelial functions (for example, by utilizing techniques that measure peripheral vascular dilatory response [<xref ref-type="bibr" rid="B182">182</xref>] and cerebral perfusion [<xref ref-type="bibr" rid="B90">90</xref>]), as well as the relationship between the extent of endothelial dysfunction and the severity of depressive symptoms.</p></sec></sec><sec sec-type="conclusions"><title>Conclusions</title><p>Neurovascular dysfunction with BBB hyperpermeability may occur in MDD. Cumulative clinical and experimental evidence implicates oxidative stress, eNOS uncoupling, and reduced endothelial NO levels in the pathophysiology of peripheral vascular endothelial dysfunction associated with MDD. Our theoretical integration of the human and animal data links oxidative stress, eNOS uncoupling, low endothelial NO levels, and neuroinflammation to putative neurovascular and BBB abnormalities in MDD. If future studies confirm their relevance to the pathophysiology of MDD, novel agents correcting these abnormalities may prove to be effective treatment strategies.</p></sec><sec><title>Abbreviations</title><p>AQP4: Aquaporin 4; BH2: Dihydrobiopterin; BH4: Tetrahydrobiopterin; CBF: Cerebral blood flow; COX2: Cyclooxygenase 2; CRH: Corticotropin-releasing hormone; CSF: Cerebrospinal fluid; CT: Computed tomography; EEG: Electroencephalogram; eNOS: Endothelial nitric oxide synthase; EAAT: Excitatory amino acid transporter; Fc: Immunoglobulin constant region; H2O2: Hydrogen peroxide; HO-: Hydroxyl radical; ICAM-1: Intercellular adhesion molecule 1; IL: Interleukin; iNOS: Inducible nitric oxide synthase; MAP: Microglial activation and proliferation; MDD: Major depressive disorder; MRI: Magnetic resonance imaging; mGluR: Metabotropic glutamate receptor; MMPs: Matrix metalloproteinases; NAD(P)H: Nicotinamide adenosine dinucleotide phosphate; Na+/K+ ATPase: Sodium-potassium adenosine triphosphates; NFκB: Nuclear factor κB; NMDAR: <italic>N</italic>-methyl-<sc>D</sc>-aspartate receptor; NO: Nitric oxide; ONOO-: Peroxynitrite; O2-: Superoxide; PET: Positron emission tomography; PLA2: Phospholipase A2; RNS: Reactive nitrogen species; ROS: Reactive oxygen species; RUR: Relative uptake ratio; SOD-1: Superoxide dismutase 1; SPECT: Single photon emission computed tomography; SSRI: Selective serotonin reuptake inhibitor; Th: T helper; TNFα: Tumor necrosis factor α; TReg: CD4<sup>+</sup>CD25<sup>+</sup>FOXP3<sup>+</sup> T regulatory; VCAM-1: Vascular cell adhesion molecule 1.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>SN, DMP conceived and designed the research; SN, DMP wrote the manuscript; SN, DMP, AN, OD, DZ, revised the manuscript for important content; SN, DMP, AN, OD, DZ, performed literature searches and gathered data for the review; all authors read and approved the final version of the manuscript for submission.</p></sec>
Activation of microglia induces symptoms of Parkinson’s disease in wild-type, but not in IL-1 knockout mice	<sec><title>Background</title><p>Parkinson’s disease (PD) is an age-related progressive neurodegenerative disorder caused by selective loss of dopaminergic neurons from the substantia nigra (SN) to the striatum. The initial factor that triggers neurodegeneration is unknown; however, inflammation has been demonstrated to be significantly involved in the progression of PD. The present study was designed to investigate the role of the pro-inflammatory cytokine interleukin-1 (IL-1) in the activation of microglia and the decline of motor function using IL-1 knockout (KO) mice.</p></sec><sec><title>Methods</title><p>Lipopolysaccharide (LPS) was stereotaxically injected into the SN of mice brains as a single dose or a daily dose for 5 days (5 mg/2 ml/injection, bilaterally). Animal behavior was assessed with the rotarod test at 2 hr and 8, 15 and 22 days after the final LPS injection.</p></sec><sec><title>Results</title><p>LPS treatment induced the activation of microglia, as demonstrated by production of IL-1β and tumor necrosis factor (TNF) α as well as a change in microglial morphology. The number of cells immunoreactive for 4-hydroxynonenal (4HNE) and nitrotyrosine (NT), which are markers for oxidative insults, increased in the SN, and impairment of motor function was observed after the subacute LPS treatment. Cell death and aggregation of α-synuclein were observed 21 and 30 days after the final LPS injection, respectively. Behavioral deficits were observed in wild-type and TNFα KO mice, but IL-1 KO mice behaved normally. Tyrosine hydroxylase (TH) gene expression was attenuated by LPS treatment in wild-type and TNFα KO mice but not in IL-1 KO mice.</p></sec><sec><title>Conclusions</title><p>The subacute injection of LPS into the SN induces PD-like pathogenesis and symptoms in mice that mimic the progressive changes of PD including the aggregation of α-synuclein. LPS-induced dysfunction of motor performance was accompanied by the reduced gene expression of TH. These findings suggest that activation of microglia by LPS causes functional changes such as dopaminergic neuron attenuation in an IL-1-dependent manner, resulting in PD-like behavioral impairment.</p></sec>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Tanaka</surname><given-names>Sachiko</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>stanaka@pharm.showa-u.ac.jp</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Ishii</surname><given-names>Atsuko</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>atk-ysi-09096843233@softbank.ne.jp</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Ohtaki</surname><given-names>Hirokazu</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>taki@med.showa-u.ac.jp</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Shioda</surname><given-names>Seiji</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>shioda@med.showa-u.ac.jp</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Yoshida</surname><given-names>Takemi</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>yoshida@pharm.showa-u.ac.jp</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Numazawa</surname><given-names>Satoshi</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>numazawa@pharm.showa-u.ac.jp</email></contrib>	Journal of Neuroinflammation	<sec><title>Background</title><p>Lipopolysaccharide (LPS) is a bacterial endotoxin known to stimulate immune responses [<xref ref-type="bibr" rid="B1">1</xref>]. In <italic>in vivo</italic> experiments, LPS activates the microglia, resulting in the release of inflammatory cytokines such as interleukin-1β (IL-1β) and tumor necrosis factor α (TNFα), which contribute to neurodegeneration [<xref ref-type="bibr" rid="B1">1</xref>-<xref ref-type="bibr" rid="B5">5</xref>]. Activation of the microglia has also been observed during the development of neurodegenerative conditions such as Alzheimer’s disease (AD), Parkinson’s disease (PD) and multiple sclerosis [<xref ref-type="bibr" rid="B6">6</xref>-<xref ref-type="bibr" rid="B8">8</xref>]. Several epidemiological studies have shown that non-steroidal anti-inflammatory drugs are associated with a reduced risk of developing PD [<xref ref-type="bibr" rid="B9">9</xref>-<xref ref-type="bibr" rid="B11">11</xref>]. These phenomena suggest that inflammation is implicated in neurodegenerative diseases [<xref ref-type="bibr" rid="B12">12</xref>]. PD is an age-related progressive neurodegenerative disorder caused by the selective loss of dopaminergic neurons from the substantia nigra (SN) to the striatum [<xref ref-type="bibr" rid="B13">13</xref>]. However, the initial factor that triggers neurodegeneration is unknown. PD animal models have been created by exposing animals to chemical toxins such as 1-methyl 4-phenyl 1,2,3,6,-tetrahydropyridine (MPTP) [<xref ref-type="bibr" rid="B14">14</xref>-<xref ref-type="bibr" rid="B16">16</xref>] and 6-hydroxydopamine (6-OHDA) [<xref ref-type="bibr" rid="B17">17</xref>]. These toxins selectively modulate dopaminergic neurons via the uptake by the dopamine transporter. However, these animal models do not encompass all the prevailing pathologies of PD. Therefore, we designed a PD animal model where there is neuroinflammation in the mouse brain. In our previous studies, we demonstrated that subacute administration of LPS (20 μg/2 μL/injection, daily, bilaterally for 5 consecutive days) into the CA1 region of the rat and mouse hippocampus activated the microglia and increased production of IL-1β and TNFα, concomitantly resulting in learning and memory deficits in the animals as assessed using a step-through passive avoidance test [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B18">18</xref>]. These results suggest that inflammation affects neuronal function. Furthermore, IL-1β plays an important role in LPS-induced impairment of learning and memory using IL-1α/β knockout (KO) mice [<xref ref-type="bibr" rid="B18">18</xref>]. In the present study, we modified the regimen and obtained evidence that suggests there are PD-like pathological changes and symptoms in the animal model. In addition, LPS-induced microglial activation causes toxicity in dopaminergic neurons in an IL-1-dependent manner. The results of the present study may lead to a better understanding of the roles of IL-1 in the activation of the microglia and the mechanisms underlying neurodegenerative diseases.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Materials</title><p>The following reagents were obtained from commercial sources: LPS (from <italic>Escherichia coli</italic> serotype 055:B5; L2880, endotoxin level 3000000 EU/mg), monoclonal anti-mouse glial fibrillary acidic protein (GFAP) antibody from Sigma-Aldrich (St Louis, MO), monoclonal goat anti-mouse CD11b antibody from Serotec Ltd (Oxford, UK), goat anti-murine IL-1β antibody from R&D Systems (Minneapolis, MN), polyclonal rabbit anti-Iba1 antibody from Wako Pure Chemical Industries (Osaka, Japan), polyclonal anti-3-nitrotyrosine antibody, Alexa Fluor 546 donkey anti-goat IgG antibody, Alexa Fluor 488 goat anti-rat IgG antibody and Alexa Fluor 488 goat anti-mouse IgG antibody from Molecular Probes (Eugene, OR), monoclonal anti-α-synuclein antibody from Santa Cruz Biotechnology, Inc (Dallas, TX), polyclonal anti-4-hydroxynonenal (4HNE) antibody from ENZO Life Sciences, Inc (Famingdale, NY) and anti-nitrotyrosine (NT) antibody from Merck (Darmstadt, Germany). All other chemicals were of analytical or the highest grade commercially available.</p></sec><sec><title>Animals, surgical operations and LPS treatment</title><p>All animal experiments were conducted in accordance with the Showa University Animal Experiment and Welfare Regulations. The IL-1 KO male mice, carrying a null mutation in both the IL-1α and IL-1β genes, and the TNFα Κ Ο male mice were established by Horai <italic>et al</italic>. [<xref ref-type="bibr" rid="B19">19</xref>] and Tagawa <italic>et al</italic>. [<xref ref-type="bibr" rid="B20">20</xref>], respectively. Male BALB/c mice were purchased from Sankyo Laboratory Service (Tokyo, Japan). The mice (at 9 to 10 weeks old) were anesthetized with sodium pentobarbital (50 mg/kg intraperitoneally) and immobilized in a stereotaxic frame. Two guide cannulas were implanted and used to inject LPS into both sides of the substantia nigra pars reticulata (SNR) (2.92 mm posterior, 1.25 mm lateral and 4.95 mm ventral to the bregma). These cannulas were fixed to the skull with dental cement. Eight days after surgery, the mice were injected bilaterally with either LPS (5 μg) dissolved in 2 μL of phosphate-buffered saline (PBS) or PBS alone under isoflurane anesthesia. Acute treatment with PBS or LPS (5 μg/2 μL/injection, bilaterally) was carried out to study the immunohistochemistry of the microglia. Subacute PBS or LPS injections (5 μg/2 μL/injection, bilaterally) were administered daily for 5 consecutive days under isoflurane anesthesia. These animals were used for behavioral tests, gene expression experiments and to study the immunohistochemistry of the brain section.</p></sec><sec><title>Behavioral analysis: the rotarod test</title><p>Motor coordination was assessed with a rotating rod apparatus (Panlab Harvard Apparatus, Barcelona, Spain). The rod was 3 cm in diameter. The mice were placed on the rod when it was rotating at 4 rpm. The rotation speed was increased from 4 to 40 rpm within 5 min. The latency was recorded for each animal. This is the time (in seconds) before they fall. Each mouse was tested three times and the median time of the three trials was calculated. The results are expressed as the mean ± standard error of the mean (SEM) in each group.</p></sec><sec><title>Immunohistochemistry</title><p>Six hours after acute LPS or PBS treatment, the mice were anesthetized with pentobarbital and were perfused transcardially with saline followed by 0.1 M phosphate buffer (PB, pH 7.2) containing 2% paraformaldehyde [<xref ref-type="bibr" rid="B21">21</xref>]. The brains were removed and immersed for 1 day in 0.1 M PB containing 2% paraformaldehyde and then for 2 days in 0.1 M PB containing 20% sucrose. The brains were then embedded in a mixture of 20% sucrose in 0.1 M PB and Tissue Teck (2:1; Miles Inc, Elkhart, IN), frozen on dry ice, and stored at -80°C until studied. Subsequently, 10 μm sections were cut with a cryostat and mounted onto gelatin-coated slides.</p><p>For staining of IL-1β, CD11b and tyrosine hydroxylase (TH), sections were pre-incubated in 5% normal horse serum. The sections were incubated with goat anti-mouse IL-1β antibody (1:100) and then were rinsed and incubated with Alexa 546-labeled donkey anti-goat IgG antibody (1:400). After that, they were incubated with rat anti-mouse CD11b antibody (1:100) followed by labeling with Alexa 488-labeled goat anti-rat IgG antibody (1:400). Then they were incubated with polyclonal anti-TH antibody (1:1000) followed by labeling with Alexa 350-labeled goat anti-rabbit IgG antibody (1:400).</p><p>For staining of 4HNE and NT, following pre-incubation in 5% normal horse serum sections were incubated with rabbit anti-4HNE antibody (1:500) or rabbit anti-NT antibody (1:500) followed by labeling with Alexa 546-labeled goat anti-rabbit IgG antibody (1:400) and then incubated with 4',6-diamidino-2-phenylindole (DAPI) solution (1:10000) to stain the nuclei.</p><p>For staining of α-synuclein, following pre-incubation in 5% normal goat serum, sections were incubated with monoclonal anti-α-synuclein antibody(1:500) followed by labeling with Alexa 546-labeled goat anti-mouse IgG antibody (1:400). Then they were incubated with polyclonal anti-TH antibody (1:1000), followed by Alexa 488-labeled goat anti-rabbit IgG antibody (1:400) and DAPI solution. Labeling was imaged with a fluorescence microscope (Olympus AX-70; Olympus, Tokyo, Japan).</p><p>For TH immunostaining using diaminobenzidine, the mice were sacrificed 6 hr after the final injection of the subacute treatment. Slice was prepared as presented above. Slice sections were pre-incubated in 5% normal goat serum after endogenous peroxidase blocking by 0.1 M PB containing 0.3% H<sub>2</sub>O<sub>2</sub> and incubated with polyclonal anti-TH antibody (1:1000). The sections were rinsed and incubated with biotinylated horse anti-rabbit IgG (1:1000). Then they were incubated in an avidin-biotin complex solution followed by diaminobenzidine (DAB kit, Vector, Burlingame, CA).</p><p>Fluoro-Jade B staining was used to detect neurodegenerative cells [<xref ref-type="bibr" rid="B22">22</xref>]. Sections were fixed with 4% paraformaldehyde solution for 20 min. After washing twice with PBS and once with water, the sections were immersed in 0.06% KMnO<sub>4</sub> for 15 min and then rinsed three times with purified water. Fluoro-Jade B solution (0.01% Fluoro-Jade B: 0.1% acetic acid = 1:19) was applied for 30 min at room temperature and then rinsed off with purified water, four times. The sections were dried with cold air and immersed three times in xylene for 2 min. One drop of marinol was put on top and a cover was placed on them. Labeling was imaged with a fluorescence microscope (Olympus AX-70; Olympus, Tokyo, Japan).</p></sec><sec><title>Gene expression</title><p>The mice were decapitated 6 hr after the acute PBS or LPS injection or 6 hr after the final PBS or LPS injection of the subacute treatment. The midbrains were dissected according to the method of Glowinski and Iversen [<xref ref-type="bibr" rid="B23">23</xref>] and stored at -80°C until use. Total RNA was extracted using the QIAGEN RNeasy Lipid Tissue Mini kit (QIAGEN, Hilden, Germany). Real-time RT-PCR was carried out using QuantiTect SYBR Green RT-PCR system (QIAGEN). Primers for IL-1β, TNFα, TH and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) were prepared by QuantiTect Primer Assays (QIAGEN). The amplification conditions were 40 cycles at 94°C for 15 s, 55°C for 30 s and 72°C for 30 s. Quantitative data were obtained from the relative standard curve. mRNA expression was normalized using GAPDH as an endogenous control.</p></sec><sec><title>Statistical analysis</title><p>Immunostaining in the substantia nigra pars compact (SNC) was quantified by counting positively stained cells. The number of positively stained cells in four brain sections was counted and averaged (cells/mm<sup>2</sup>). The statistical analysis used the Mann–Whitney test for the immunostaining, behavioral and gene expression data.</p></sec></sec><sec sec-type="results"><title>Results</title><sec><title>Activation of the microglia after lipopolysaccharide treatment</title><p>An experiment was conducted to determine whether the microglia was activated by acute LPS treatment (5 μg/2 μL/injection). There were many TH antibody-labeled dopaminergic neurons in the SNC. Immunohistochemical analysis revealed no IL-1β immunoreactivity in the SNC from mice treated with PBS (Figure <xref ref-type="fig" rid="F1">1</xref>A). In contrast, cells with strong immunoreactivity for IL-1β were observed 6 hr after the LPS injection. To identify the cells expressing IL-1β, a triple-label immunohistochemical study was performed for IL-1β, CD11b and TH. IL-1β immunopositive cells were co-localized with cells immunoreactive to CD11b. In the early stage of activation, LPS induced an increase in IL-1β signals in the microglia, the shape of which remained in the ramified form 6 hr after the acute LPS treatment. The gene expressions of IL-1β and TNFα in the midbrain were still noticeably higher 6 hr after the acute LPS treatment (Figure <xref ref-type="fig" rid="F1">1</xref>B).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Expression of inflammatory cytokine in an acute LPS-treated mouse. (A)</bold> Immunohistochemical analysis of the effect of LPS treatment on the microglia in the SNC. PBS (2 μL/injection) or LPS (5 μg/2 μL/injection) was injected into the SNR bilaterally. These representative photomicrographs are coronal sections of the SNC at 6 hr after PBS (upper row) or acute LPS treatment (lower row). Triple immunofluorescence staining was performed with CD11b (green), IL-1β (red) and TH (blue) antibodies. Immunoreactivity for IL-1β was co-localized with that for CD11b after LPS treatment (yellow in the merged photomicrograph). Scale bars: 20 μm <bold>(B)</bold> Effect of LPS treatment on the gene expression of inflammatory cytokine in the midbrain. The midbrain was obtained 6 hr after the acute LPS treatment. Values are expressed as percentages of those in the PBS-treated group and represented as mean ± SEM. <italic>P</italic> < 0.05 compared with the PBS-treated group. IL, interleukin; LPS, lipopolysaccharide; PBS, phosphate-buffered saline; SEM, standard error of the mean; SNC, substantia nigra pars compact; SNR, substantia nigra pars reticulata; Τ Η , tyrosine hydroxylase.</p></caption><graphic xlink:href="1742-2094-10-143-1"/></fig><p>We demonstrated the long-term activation of the microglia after the subacute treatment with LPS for 5 days (Figure <xref ref-type="fig" rid="F2">2</xref>). In the SN of mice injected with PBS, the majority of the CD11b immunopositive cells exhibited a resting or ramified state (Figure <xref ref-type="fig" rid="F2">2</xref>A, upper rows). After the subacute LPS treatment, the CD11b-immunopositive cells were activated, as indicated by an increased number of cells (Figure <xref ref-type="fig" rid="F2">2</xref>B) as well as a change in their morphology to a round and blunt shape (amoeboid form, Figure <xref ref-type="fig" rid="F2">2</xref>A, lower rows). The CD11b-immunopositive cells had an activated phenotype. These results suggest that the subacute LPS treatment caused a sustained activation of the microglia, resulting in an increase in the number of microglial cells and morphological changes.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Immunohistochemical analysis of the effect of the subacute LPS treatment on CD11b-positive microglia.</bold> PBS (2 μL/injection, bilaterally) or LPS (5 μg/2 μL/injection, bilaterally) was injected into the SNR daily for 5 consecutive days. <bold>(A)</bold> Representative photomicrographs are coronal sections of the SNC 6 hr after the final PBS (upper rows) or LPS treatment (lower rows). Double immunofluorescence staining was performed with a CD11b antibody (green) and DAPI (blue). Scale bars: 50 μm <bold>(B)</bold> Number of CD11b immunopositive cells in the SNC. The number of microglial cells was counted between 200 μm and 300 μm from the injection site. The results are expressed as the mean ± SEM for four mice in each group. DAPI, 4',6-diamidino-2-phenylindole; IL, interleukin; LPS, lipopolysaccharide; PBS, phosphate-buffered saline; SEM, standard error of the mean; SNC, substantia nigra pars compact; SNR, substantia nigra pars reticulata.</p></caption><graphic xlink:href="1742-2094-10-143-2"/></fig></sec><sec><title>Detection of oxidative stress</title><p>4HNE is a product of lipid peroxidation and its increased production is a biomarker of oxidative stress. The number of 4HNE-positive cells in the SNC was significantly increased 6 hr after the final LPS injection (Figure <xref ref-type="fig" rid="F3">3</xref>A,B, upper row). NT is an indicator of cell damage, the activation of inflammation and NO production [<xref ref-type="bibr" rid="B39">39</xref>]. NT-immunopositive cells were seen to increase up to fourfold 6 hr after the final LPS injection compared with the PBS-treated group (Figure <xref ref-type="fig" rid="F3">3</xref>A,B, lower row). These results suggest that oxidative stress was induced in the SNC by the subacute LPS treatment.</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Immunohistochemical analysis of the effect of subacute LPS treatment on oxidative stress markers.</bold> PBS (2 μL/injection, bilaterally) or LPS (5 μg/2 μL/injection, bilaterally) was injected into the SNR daily for 5 consecutive days. <bold>(A</bold>) Representative photomicrographs are coronal sections of the SNC region 6 hr after the final PBS (upper row) or LPS treatment (lower row). Double immunofluorescence staining was performed with 4HNE (red, upper row) or NT (red, lower row) antibody and DAPI (blue). Scale bars: 20 μm <bold>(B)</bold> Numbers of 4HNE and NT immunopositive cells in the SNC. The results are expressed as the mean ± SEM for four mice in each group. 4HNE, 4-hydroxynonenal; DAPI, 4',6-diamidino-2-phenylindole; LPS, lipopolysaccharide; PBS, phosphate-buffered saline; SEM, standard error of the mean; SNC, substantia nigra pars compact; SNR, substantia nigra pars reticulata.</p></caption><graphic xlink:href="1742-2094-10-143-3"/></fig></sec><sec><title>Detection of cell death using Fluoro-Jade B and tyrosine hydroxylase immunostaining</title><p>Cell death was seen in coronal sections of the SN using Fluoro-Jade B staining (Figure <xref ref-type="fig" rid="F4">4</xref>A). Six hr after the final injection, degenerating neurons stained with Fluoro-Jade B were not observed. However, 21 and 30 days after the final injection, Fluoro-Jade B-positive cells were detected in a similar manner as observed in the positive control specimens prepared after a traumatic cortical brain injury. Furthermore, the number of TH antibody-labeled dopaminergic neurons in the SNC decreased 6 hr after the final LPS injection (Figure <xref ref-type="fig" rid="F4">4</xref>B). These results indicate that delayed cell death was induced in the SNC by the subacute LPS treatment.</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>Detection of cell death using Fluoro-Jade B and TH immunostaining.</bold> PBS (2 μL/injection, bilaterally) or LPS (5 μg/2 μL/injection, bilaterally) was injected into the SNR daily for 5 consecutive days. <bold>(A)</bold> The representative photomicrographs of Fluoro-Jade B staining are coronal sections of the SNC 6 hr, 21 days and 30 days after the final PBS (left) or LPS treatment (right). The positive control was prepared after a traumatic brain injury (TBI) in the cortex. <bold>(B)</bold> The representative photographs of TH immunostaining using diaminobenzidine, are coronal sections of the SN 6 hr after the final PBS (left) or LPS treatment (right). LPS, lipopolysaccharide; PBS, phosphate-buffered saline; SN, substantia nigra; SNC, substantia nigra pars compact; SNR, substantia nigra pars reticulata; Τ Η , tyrosine hydroxylase.</p></caption><graphic xlink:href="1742-2094-10-143-4"/></fig></sec><sec><title>α-synuclein expression in lipopolysaccharide-treated mice</title><p>We performed an immunohistochemical analysis to see if there was any α-synuclein protein in the SNC (Figure <xref ref-type="fig" rid="F5">5</xref>). α-synuclein-immunopositive cells were observed 30 days after the final LPS injection (Figure <xref ref-type="fig" rid="F5">5</xref>A). α-synuclein gene expression in the midbrain area was unchanged 6 hr after the final LPS treatment compared with its expression in the PBS-treated group (data not shown). However, 30 days after the final LPS injection, α-synuclein gene expression was significantly increased to 170% of that for the PBS-treated group (Figure <xref ref-type="fig" rid="F5">5</xref>B).</p><fig id="F5" position="float"><label>Figure 5</label><caption><p><bold>Immunohistochemical analysis of the effect of subacute LPS treatment on α-synuclein.</bold> PBS (2 μL/injection, bilaterally) or LPS (5 μg/2 μL/injection, bilaterally) was injected into the SNR daily for 5 consecutive days. <bold>(A)</bold> Representative photomicrographs are coronal sections of the SNC 30 days after the final PBS (upper) or LPS treatment (lower). Triple immunofluorescence staining was performed with α-synuclein (red) antibody, TH (green) antibody and DAPI (blue). Scale bars: 20 μm <bold>(B)</bold> Effect of LPS treatment on the gene expression of α-synuclein in the midbrain 30 days after the final LPS treatment. Values are expressed as percentages of those in the PBS-treated group and represented as mean ± SEM. <italic>P</italic> < 0.05 compared with the PBS-treated group. DAPI, 4',6-diamidino-2-phenylindole; LPS, lipopolysaccharide; PBS, phosphate-buffered saline; SEM, standard error of the mean; SNC, substantia nigra pars compact; SNR, substantia nigra pars reticulata; Τ Η , tyrosine hydroxylase.</p></caption><graphic xlink:href="1742-2094-10-143-5"/></fig></sec><sec><title>Assessment of animal behavior using the rotarod test</title><p>Behavioral tests were performed 2 hr and 8, 15 and 22 days after the final LPS injection. The latency of the mice increased during the trial (Figure <xref ref-type="fig" rid="F6">6</xref>). In wild-type mice, LPS produced a lower latency to fall off, which was significant 15 days and 22 days after the final injection. These results suggest that the LPS treatment resulted in behavioral impairment of the motor function. We used genetically modified animals to determine whether the pro-inflammatory cytokine was involved in the LPS-induced behavioral deficit. The behavioral tests were performed with the IL-1α/β double KO and TNFα KO mice treated either with PBS or LPS. The latency of the IL-1 KO and TNFα KO mice also increased during the trial, like the wild-type mice (Figure <xref ref-type="fig" rid="F6">6</xref>). TNFα KO mice treated with LPS had a significantly lower latency to fall off 2 hr, 8 days and 22 days after the final LPS injection. However, there was no difference between the PBS and LPS treatments for the IL-1 KO mice. These results suggest that the LPS treatment caused impairment of the motor function in the wild-type and TNFα KO mice, but not in the IL-1 KO mice.</p><fig id="F6" position="float"><label>Figure 6</label><caption><p><bold>Effect of subacute LPS treatment on performance in the rotarod test for wild-type, IL-1 KO and TNFα KO mice.</bold> PBS (2 μL/injection, bilaterally, open column) or LPS (5 μg/2 μL/injection, bilaterally, closed column) was injected into the SNR daily for 5 consecutive days. The rotarod test was conducted at 2 hr and 8, 15 and 22 days after the final PBS or LPS treatment. The test was performed three times and latency to fall off was recorded for each trial. The median latency time was calculated for each mouse. The results are expressed as the mean ± SEM in each group. <italic>P</italic> < 0.05 compared with the PBS-treated group. IL, interleukin; KO, knockout; LPS, lipopolysaccharide; PBS, phosphate-buffered saline; SEM, standard error of the mean; TNF, tumor necrosis factor.</p></caption><graphic xlink:href="1742-2094-10-143-6"/></fig></sec><sec><title>Tyrosine hydroxylase gene expression in wild-type, TNFα and IL-1 KO mice</title><p>We analyzed the gene expression of TH, a marker for dopaminergic and noradrenergic neurons, in the wild-type, TNFα and IL-1 KO mice 6 hr after the final LPS injection. The subacute LPS treatment significantly suppressed TH gene expression in the wild-type and TNFα KO mice, but not IL-1 KO mice (Figure <xref ref-type="fig" rid="F7">7</xref>). These results for TH expression are in agreement with those obtained from the behavior tests.</p><fig id="F7" position="float"><label>Figure 7</label><caption><p><bold>Effect of subacute LPS treatment on the gene expression of tyrosine hydroxylase in the midbrain.</bold> PBS (2 μL/injection, bilaterally) or LPS (5 μg/2 μL/injection, bilaterally) was injected into the SNR daily for 5 consecutive days. The midbrain was obtained 6 hr after the final PBS or LPS treatment. Values are expressed as percentages of those in the PBS-treated group and represented as mean ± SEM. <italic>P</italic> < 0.05 compared with the PBS-treated group. IL, interleukin; KO, knockout; LPS, lipopolysaccharide; PBS, phosphate-buffered saline; SEM, standard error of the mean; SNR, substantia nigra pars reticulata; TNF, tumor necrosis factor.</p></caption><graphic xlink:href="1742-2094-10-143-7"/></fig></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>In the present study, we investigated whether <italic>in vivo</italic> activation of the microglia leads to impaired neuronal function and found that LPS-induced functional outcomes in terms of decreased motor performance correlated well with changes in immunohistochemistry and gene expression.</p><p>PD models have been created by exposing animals to toxins such as MPTP [<xref ref-type="bibr" rid="B14">14</xref>-<xref ref-type="bibr" rid="B16">16</xref>] and 6-OHDA [<xref ref-type="bibr" rid="B17">17</xref>]. These toxins have been demonstrated to cause PD-like symptoms with massive losses of dopaminergic neurons in the SN. These toxins are selectively taken up by dopaminergic neurons via the dopamine transporter and cause neuronal cell death without α-synuclein aggregation. In contrast, neuronal cell death is not believed to be the first event in the development of PD. What triggers PD in humans has not been determined. Therefore, these toxins do not induce changes that completely mimic PD-like pathology.</p><p>We chose LPS to induce microglial activation and neuronal inflammation in the SN of mice. In the LPS-induced inflammation model, dopaminergic neurons were not directly injured by the endotoxin, which may cause deterioration of neurons via the microglia. The present study indicated that LPS treatment caused the activation of the microglial cells that produce IL-1β and TNFα as well as changes in microglial morphology. These findings are comparable to those previously reported for our rat and mouse models [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B18">18</xref>], in which LPS injected into the hippocampus induced learning and memory deficits. TNFα and IL-1β are elevated in patients with PD [<xref ref-type="bibr" rid="B24">24</xref>-<xref ref-type="bibr" rid="B26">26</xref>] and animal models [<xref ref-type="bibr" rid="B27">27</xref>,<xref ref-type="bibr" rid="B28">28</xref>]. The activated microglia increases the secretion of pro-inflammatory cytokines and activates pro-oxidant enzymes such as nicotinamide adenine dinucleotide phosphate oxidase and inducible nitric oxide synthase (iNOS), resulting in the production of reactive oxygen species (ROS) and nitric oxide [<xref ref-type="bibr" rid="B29">29</xref>,<xref ref-type="bibr" rid="B30">30</xref>].</p><p>Oxidative stress contributes to the neurodegenerative process in AD and PD [<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B32">32</xref>]. PD is characterized by the selective loss of dopaminergic neurons in the SN, and dopamine has been suggested to be the main endogenous toxin causing oxidative stress in the genesis of PD [<xref ref-type="bibr" rid="B33">33</xref>]. Dopaminergic neurons are vulnerable to oxidative stress. Dopamine is easily oxidized by molecular oxygen to produce ROS such as superoxide anions and hydrogen peroxide and forms aminochrome via dopamine <italic>o</italic>-quinone [<xref ref-type="bibr" rid="B34">34</xref>]. Iron accumulates in the SN of patients with PD [<xref ref-type="bibr" rid="B35">35</xref>]. Auto-oxidation of dopamine is stimulated by iron in patients with PD, which causes the degeneration of dopaminergic neurons by enhancing oxidative stress [<xref ref-type="bibr" rid="B34">34</xref>]. Levels of free and protein-bound 4HNE, a toxic aldehyde produced by the peroxidation of fatty acids, were significantly elevated in the ventricular fluid of patients with AD [<xref ref-type="bibr" rid="B36">36</xref>] and a PD animal model [<xref ref-type="bibr" rid="B37">37</xref>]. 4HNE induces apoptosis with caspase activation and DNA fragmentation [<xref ref-type="bibr" rid="B38">38</xref>]. In our LPS-induced animal model, we detected an increase in the number of cells immunopositive for 4HNE, which could be involved in neuronal cell death. We also observed that LPS increased the number of cells immunopositive for NT, which has been identified as an indicator of cell damage and inflammation as a result of the production of NO and oxidative stress [<xref ref-type="bibr" rid="B39">39</xref>]. LPS stimulates the induction of iNOS and its product NO reacts with superoxide to produce peroxynitrite and eventually nitration of proteins, which might contribute to disturbing protein functions resulting in the pathological processes seen during neurodegeneration.</p><p>α-synuclein is a major component of Lewy bodies, a pathological hallmark of PD, and is involved in neurodegeneration [<xref ref-type="bibr" rid="B40">40</xref>]. It has been reported that LPS stimulates the production of α-synuclein in the SN of rats [<xref ref-type="bibr" rid="B41">41</xref>]. The present study demonstrated that there was α-synuclein gene expression and cell death 22 and 30 days after the final LPS injection, respectively. However, the changes in animal behavior started just after the final LPS injection, suggesting that impairment of the motor function appears before cell death. These results suggest that the subacute injection of LPS into the SNR induces PD-like pathogenesis and symptoms in mice, which mimic the progressive changes of PD including the aggregation of α-synuclein that causes the dysfunction of motor performance. Therefore, a subacute LPS treatment could be a novel regimen to create animal models of PD.</p><p>We did not determine the type of cells that took part in the delayed cell death in this study. It is well known that 80% of neurons in the SNC are dopaminergic. Numbers of TH-immunopositive cells decreased 6 hr after the LPS final injection. Therefore, it is suggested that Fluoro-Jade B-stained cells are dopaminergic neurons in the SNC.</p><p>Several reports attest the involvement of IL-1 in neurodegenerative diseases. Increased iNOS immunoreactivity, which is normally observed after brain ischemia, is diminished in IL-1 KO mice. These mice exhibit markedly reduced neuronal loss and apoptotic cell death when they experience a transient cardiac arrest [<xref ref-type="bibr" rid="B42">42</xref>]. In a spinal cord injury model, the size of the lesion area decreased in IL-1 KO mice compared with wild-type mice [<xref ref-type="bibr" rid="B43">43</xref>]. We previously demonstrated that LPS-induced deficits of learning and memory did not occur for IL-1 KO mice [<xref ref-type="bibr" rid="B18">18</xref>]. These findings indicate that IL-1 plays an important role in neurodegenerative disorders in which a neuronal dysfunction is associated with the activation of the microglia and increased IL-1 expression.</p><p>The present study demonstrated that behavioral deficits do not occur following a subacute LPS treatment in IL-1 KO mice. TH gene expression, the late-limiting enzyme for dopamine synthesis, was attenuated by the LPS treatment in wild-type and TNFα KO mice, but not in IL-1 Κ Ο mice. Impairment of the motor function was accompanied by an alteration of TH gene expression. It appears that increased IL-1 expression in the early stage of PD may trigger microglial activation and may be involved in neurodegeneration. Thus, neuroinflammation including an increase of IL-1 levels and augmentation of its signaling pathway may contribute to the dopaminergic dysfunction. If we had checked dopamine levels, we may have found direct evidence for dopaminergic neuron attenuation and motor dysfunction. We will check dopamine levels in the following study. Our present results suggest that IL-1 plays an important role in mediating LPS-induced functional changes of motor performance. However, the precise mechanism by which microglial activation occurs remains to be determined.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>The findings presented in this study provide evidence that activation of the microglia by LPS causes functional changes such as dopaminergic neuron attenuation in an IL-1-dependent manner, resulting in PD-like behavioral impairment. The LPS-induced animal model for PD could be a useful tool for clarifying mechanisms underlying the neurodegenerative disease and, therefore, should be evaluated further.</p></sec><sec><title>Abbreviations</title><p>4HNE: 4-hydroxynonenal; 6-OHDA: 6-hydroxydioamine; AD: Alzheimer’s disease; DAPI: 4',6-diamidino-2-phenylindole; GAPDH: Glyceraldehyde 3-phosphate dehydrogenase; GFAP: Glial fibrillary acidic protein; IL: Interleukin; iNOS: inducible nitric oxide synthase; KO: Knockout; LPS: Lipopolysaccharide; MPTP: 1-methyl 4-phenyl 1,2,3,6,-tetrahydropyridine; NT: Nitrotyrosine; PB: Phosphate buffer; PBS: Phosphate-buffered saline; PD: Parkinson’s disease; ROS: Reactive oxygen species; RT-PCR: Real-time polymerase chain reaction; SEM: Standard error of the mean; SN: Substantia nigra; SNC: Substantia nigra pars compact; SNR: Substantia nigra pars reticulata; Τ Η : Tyrosine hydroxylase; TNF: Tumor necrosis factor.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>ST designed the study, established the protocols and drafted the manuscript. AI carried out the experiments and the statistical analysis. HO helped to establish the protocols and participated in the data analysis. SS and TY participated in the data analysis and helped to draft the manuscript. SN coordinated the experiments and co-wrote the manuscript. All authors read and approved the final manuscript.</p></sec>
Identification of broadly neutralizing antibody epitopes in the HIV-1 envelope glycoprotein using evolutionary models	<sec><title>Background</title><p>Identification of the epitopes targeted by antibodies that can neutralize diverse HIV-1 strains can provide important clues for the design of a preventative vaccine.</p></sec><sec><title>Methods</title><p>We have developed a computational approach that can identify key amino acids within the HIV-1 envelope glycoprotein that influence sensitivity to broadly cross-neutralizing antibodies. Given a sequence alignment and neutralization titers for a panel of viruses, the method works by fitting a phylogenetic model that allows the amino acid frequencies at each site to depend on neutralization sensitivities. Sites at which viral evolution influences neutralization sensitivity were identified using Bayes factors (BFs) to compare the fit of this model to that of a null model in which sequences evolved independently of antibody sensitivity. Conformational epitopes were identified with a Metropolis algorithm that searched for a cluster of sites with large Bayes factors on the tertiary structure of the viral envelope.</p></sec><sec><title>Results</title><p>We applied our method to ID<sub>50</sub> neutralization data generated from seven HIV-1 subtype C serum samples with neutralization breadth that had been tested against a multi-clade panel of 225 pseudoviruses for which envelope sequences were also available. For each sample, between two and four sites were identified that were strongly associated with neutralization sensitivity (2ln(BF) > 6), a subset of which were experimentally confirmed using site-directed mutagenesis.</p></sec><sec><title>Conclusions</title><p>Our results provide strong support for the use of evolutionary models applied to cross-sectional viral neutralization data to identify the epitopes of serum antibodies that confer neutralization breadth.</p></sec>	<contrib contrib-type="author" equal-contrib="yes" id="A1"><name><surname>Lacerda</surname><given-names>Miguel</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>miguel.lacerda@uct.ac.za</email></contrib><contrib contrib-type="author" equal-contrib="yes" id="A2"><name><surname>Moore</surname><given-names>Penny L</given-names></name><xref ref-type="aff" rid="I3">3</xref><xref ref-type="aff" rid="I4">4</xref><email>pennym@nicd.ac.za</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Ngandu</surname><given-names>Nobubelo K</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>nobubelo.ngandu@uct.ac.za</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Seaman</surname><given-names>Michael</given-names></name><xref ref-type="aff" rid="I6">6</xref><email>mseaman@bidmc.harvard.edu</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Gray</surname><given-names>Elin S</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>e.gray@ecu.edu.au</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Murrell</surname><given-names>Ben</given-names></name><xref ref-type="aff" rid="I7">7</xref><xref ref-type="aff" rid="I8">8</xref><email>murrellb@gmail.com</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Krishnamoorthy</surname><given-names>Mohan</given-names></name><xref ref-type="aff" rid="I9">9</xref><email>mohan.krishnamoorthy11@gmail.com</email></contrib><contrib contrib-type="author" id="A8"><name><surname>Nonyane</surname><given-names>Molati</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>molatin@nicd.ac.za</email></contrib><contrib contrib-type="author" id="A9"><name><surname>Madiga</surname><given-names>Maphuti</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>maphutim@nicd.ac.za</email></contrib><contrib contrib-type="author" id="A10"><name><surname>Wibmer</surname><given-names>Constantinos Kurt</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>kurtw@nicd.ac.za</email></contrib><contrib contrib-type="author" id="A11"><name><surname>Sheward</surname><given-names>Daniel</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>daniel.sheward@uct.ac.za</email></contrib><contrib contrib-type="author" id="A12"><name><surname>Bailer</surname><given-names>Robert T</given-names></name><xref ref-type="aff" rid="I10">10</xref><email>rbailer@mail.nih.gov</email></contrib><contrib contrib-type="author" id="A13"><name><surname>Gao</surname><given-names>Hongmei</given-names></name><xref ref-type="aff" rid="I11">11</xref><email>hongmei.gao@duke.edu</email></contrib><contrib contrib-type="author" id="A14"><name><surname>Greene</surname><given-names>Kelli M</given-names></name><xref ref-type="aff" rid="I11">11</xref><email>kelli.greene@duke.edu</email></contrib><contrib contrib-type="author" id="A15"><name><surname>Karim</surname><given-names>Salim S Abdool</given-names></name><xref ref-type="aff" rid="I12">12</xref><xref ref-type="aff" rid="I13">13</xref><email>karims1@ukzn.ac.za</email></contrib><contrib contrib-type="author" id="A16"><name><surname>Mascola</surname><given-names>John R</given-names></name><xref ref-type="aff" rid="I10">10</xref><email>jmascola@mail.nih.gov</email></contrib><contrib contrib-type="author" id="A17"><name><surname>Korber</surname><given-names>Bette TM</given-names></name><xref ref-type="aff" rid="I9">9</xref><xref ref-type="aff" rid="I14">14</xref><email>btk@lanl.gov</email></contrib><contrib contrib-type="author" id="A18"><name><surname>Montefiori</surname><given-names>David C</given-names></name><xref ref-type="aff" rid="I11">11</xref><email>monte@duke.edu</email></contrib><contrib contrib-type="author" id="A19"><name><surname>Morris</surname><given-names>Lynn</given-names></name><xref ref-type="aff" rid="I3">3</xref><xref ref-type="aff" rid="I4">4</xref><email>lynnm@nicd.ac.za</email></contrib><contrib contrib-type="author" id="A20"><name><surname>Williamson</surname><given-names>Carolyn</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>carolyn.williamson@uct.ac.za</email></contrib><contrib contrib-type="author" corresp="yes" id="A21"><name><surname>Seoighe</surname><given-names>Cathal</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>cathal.seoighe@nuigalway.ie</email></contrib><on-behalf-of>the CAVD-NSDP Consortium<xref ref-type="corresp" rid="d33e209"/></on-behalf-of>	Virology Journal	<sec><title>Background</title><p>A successful HIV-1 vaccine is likely to require the induction of neutralizing antibodies that can prevent infection. HIV-1 entry into host cells is mediated by the HIV-1 envelope glycoprotein, which forms a trimeric structure on the surface of the virus. Each of these envelope “spikes” consists of three identical, non-covalently associated heterodimers of surface gp120 and transmembrane gp41. Antibodies that bind the envelope can be detected within eight days of infection [<xref ref-type="bibr" rid="B1">1</xref>]. However, neutralizing antibodies that specifically bind the trimeric form of the envelope and prevent cell entry are slower to develop and appear about 3–6 months after infection [<xref ref-type="bibr" rid="B2">2</xref>-<xref ref-type="bibr" rid="B7">7</xref>]. Importantly, these early neutralizing antibodies preferentially bind the autologous virus and are therefore strain-specific [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B7">7</xref>-<xref ref-type="bibr" rid="B9">9</xref>]. In contrast, recent studies have revealed that 15-20% of infected people are able to develop serum antibodies that exhibit neutralization of genetically diverse HIV-1 strains [<xref ref-type="bibr" rid="B10">10</xref>-<xref ref-type="bibr" rid="B12">12</xref>]. However, these broadly neutralizing antibodies are generally only produced 2–4 years after infection [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B13">13</xref>]. Although these neutralizing antibodies may not protect against disease progression [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B14">14</xref>], the fact that the host B cell response has the potential to generate such broadly reactive neutralizing antibodies against HIV-1 has led to renewed interest in the development of a preventative vaccine that elicits similar types of antibodies [<xref ref-type="bibr" rid="B3">3</xref>].</p><p>The HIV-1 envelope has evolved an array of mechanisms that hinder binding by neutralizing antibodies. The envelope glycoprotein is genetically variable, conformationally flexible and heavily glycosylated, resulting in either escape from antibody recognition or shielding of neutralization sensitive sites [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B16">16</xref>]. The narrowness of the initial response, together with the plasticity of the envelope protein, allows escape variants to evolve rapidly in the infected individual [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B17">17</xref>]. Nevertheless, fitness constraints preclude complete resistance and certain regions of envelope remain vulnerable to antibody neutralization [<xref ref-type="bibr" rid="B18">18</xref>]. Recently, new highly potent monoclonal antibodies have been isolated that define targets on the HIV envelope. This includes the PG9/PG16 monoclonal antibodies that recognize an epitope involving V2 and V3 created by the trimeric structure [<xref ref-type="bibr" rid="B19">19</xref>], the PGT antibodies that mostly rely on a glycan at position 332 in the C3 region of gp120 [<xref ref-type="bibr" rid="B20">20</xref>], the VRC01 monoclonal antibody that targets the CD4 binding site [<xref ref-type="bibr" rid="B21">21</xref>] and the gp41 membrane proximal external region-specific antibody 10E8 [<xref ref-type="bibr" rid="B22">22</xref>]. These broadly neutralizing monoclonal antibodies greatly expand our understanding of the conserved epitopes on the envelope, which were previously defined by IgG1b12 against the CD4 binding site, 2G12 against the glycan shield in the outer domain and 4E10 and 2 F5 that recognize distinct epitopes in the membrane proximal external region of gp41 [<xref ref-type="bibr" rid="B23">23</xref>-<xref ref-type="bibr" rid="B27">27</xref>].</p><p>Broad serum neutralization could potentially be mediated by a polyclonal set of neutralizing antibodies that accumulate over time and target several distinct regions of envelope [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B28">28</xref>]. Alternatively, the gradual focusing of the B cell response onto functionally conserved regions of envelope could be responsible for the potent neutralization of diverse HIV-1 isolates in some individuals. Although it is likely that both scenarios occur in infected subjects [<xref ref-type="bibr" rid="B29">29</xref>], the latter possibility is supported by the recent identification of monoclonal antibodies whose neutralization breadth matches that of the corresponding serum [<xref ref-type="bibr" rid="B19">19</xref>-<xref ref-type="bibr" rid="B21">21</xref>]. The identification and characterization of epitopes that are targeted by antibodies in broadly neutralizing sera is a key step in the design of immunogens that can potentially induce broad neutralizing antibodies against HIV-1.</p><p>B cell epitope prediction is complicated by the conformation-dependent nature of antigen-antibody binding. Although more than 90% of antibody epitopes are estimated to be conformational in nature, most experimental and computational methods are designed to identify only linear epitopes [<xref ref-type="bibr" rid="B30">30</xref>-<xref ref-type="bibr" rid="B32">32</xref>]. Here, we present a novel computational method that uses cross-sectional neutralization sensitivity and sequence data from a large panel of viruses to predict sites that lie within antibody epitopes. Similar data were analyzed by Gnanakaran et al. [<xref ref-type="bibr" rid="B15">15</xref>], who developed phylogenetically-corrected methods for identifying signature amino acid sites in envelope that were associated with neutralization phenotypes. While their method required discrete neutralization phenotypes, our method permits the use of continuous neutralization measures, which should enhance statistical power. More recently, Georgiev et al. [<xref ref-type="bibr" rid="B33">33</xref>] developed a technique to deconvolute the antibody specificities in polyclonal sera. However, their method does not identify amino acid positions associated with neutralization sensitivity and does not account for the phylogenetic relationships between viruses.</p><p>Our method incorporates neutralization sensitivities directly into a phylogenetic model of molecular evolution. Amino acid sites at which the pattern of evolution correlated with changes in neutralization sensitivity across the phylogenetic tree were identified. We hypothesized that many of the sites that were associated with changes in the neutralization sensitivity of multiple viruses lie within antibody epitopes. In order to identify sites in the alignment that were most likely to influence neutralization sensitivity, we used Bayes factors to compare the fit of a model that allows the amino acid frequency at each site to depend on neutralization sensitivities (epitope model) to that of a model which assumes independence (non-epitope model). A contiguous set of sites on the primary sequence that favored the epitope model provided evidence for a B cell epitope (which could be linear or conformational), while a set of sites with large Bayes factors that were clustered in three-dimensional space provided evidence of a conformational epitope. We used this approach to predict epitopes targeted by the broadly neutralizing antibodies present in the sera of seven HIV-1 subtype C-infected individuals enrolled in the CAPRISA Acute Infection study [<xref ref-type="bibr" rid="B12">12</xref>]. All sites with strong support were then tested experimentally using mutagenesis studies. We found that our model accurately predicted amino acid residues that were targeted by broadly neutralizing serum antibodies, including those present in V2, the glycan shield and the membrane proximal external region.</p></sec><sec sec-type="results"><title>Results</title><p>We have previously identified sera from seven HIV-1 subtype C-infected women in the CAPRISA 002 cohort that showed substantial neutralization breadth at 3 years post-infection against a panel of 42 viruses from subtypes A, B and C [<xref ref-type="bibr" rid="B12">12</xref>]. In order to assess neutralization serotypes more comprehensively (as part of the Neutralization Serotype Discovery Project of the Collaboration for AIDS Vaccine Discovery), these sera were tested against a much larger panel of 225 envelope-pseudotyped viruses, including subtypes A (<italic>n</italic> = 10), B (<italic>n</italic> = 56), C (<italic>n</italic> = 69), D (<italic>n</italic> = 6) and G (<italic>n</italic> = 6), as well as several circulating recombinant forms (CRFs) (<italic>n</italic> = 78). For each pseudovirus, the neutralization titer of a subject’s serum was recorded as the reciprocal of the maximal plasma dilution that could inhibit 50% of viral entry (denoted as ID<sub>50</sub>). The neutralization breadth, as measured by the overall percentage of viruses neutralized (ID<sub>50</sub> > 20) by each serum, ranged from 82% (CAP255) to 94% (CAP206) (Figure <xref ref-type="fig" rid="F1">1</xref>A). Notably, the titers of CAP256 (which neutralized 93% of the panel viruses) were considerably higher than those of any other serum, with an ID<sub>50</sub> exceeding 10, 000 against some viruses (Figure <xref ref-type="fig" rid="F1">1</xref>B). Serum from CAP8, CAP177, CAP206 and CAP255 showed no clear subtype specificity. In contrast, the CAP248, CAP256 and CAP257 sera neutralized subtype C viruses better than the subtype B panel viruses (Wilcoxon tests of median titer, <italic>p</italic> < 0.001). Since our method takes the phylogenetic relationships between the panel viruses into account, the preferential neutralization of certain HIV-1 clades by these serum antibodies did not bias our results.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>ID</bold><sub><bold>50 </bold></sub><bold>titers for each serum sample. (A)</bold> Heat map of ID<sub>50</sub> titers clustered by the viral phylogeny. The percentages below the heat map indicate the percentage of panel viruses that were neutralized (ID<sub>50</sub> > 20) by the antibodies in each serum. <bold>(B)</bold> Histograms of the natural logarithm of ID<sub>50</sub> titer. The distribution for CAP256 was notably different from that of the other sera.</p></caption><graphic xlink:href="1743-422X-10-347-1"/></fig><sec><title>Identification of amino acid residues targeted by the broadly neutralizing serum of CAP256</title><p>We first analyzed the data from CAP256 which showed the highest neutralization titers (Figure <xref ref-type="fig" rid="F1">1</xref>) mainly attributed to antibodies targeting a PG9/PG16-like quaternary (trimer-specific) epitope [<xref ref-type="bibr" rid="B34">34</xref>]. These antibodies were shown to be dependent on the R166 and K169 residues in the V2 region with F159, N160, L165, D167, and K171 also contributing to the epitope, though to a lesser extent. Consequently, these data provided a set of well-characterized true-positive sites that could be used to test whether our method had the capacity to detect key sites within neutralizing antibody epitopes.</p><p>Given a coding sequence alignment and a phylogenetic tree for the virus panel, we used evolutionary models to identify sites in the alignment at which the pattern of evolution along branches of the tree was dependent on the neutralization titer of the virus at the tip of each branch (see Methods for details). Scaled Bayes factors (denoted as <italic>B</italic><sub><italic>k</italic></sub> for HXB2 amino acid position <italic>k</italic>) were used to compare the fit of a model where the equilibrium frequency of the “reference” amino acid at each site depended on neutralization titer to that of a null model of independence. The reference amino acid at each site was defined by identifying a “reference sequence” that was sensitive to the antibodies present in a particular serum sample. Because our model predictions depended on the reference amino acid at each site, several reference sequences were considered for each serum.</p><p>Our model provided striking evidence of an association between neutralization sensitivity and the amino acids present at sites 166 (<italic>B</italic><sub>166</sub> = 25.4) and 169 (<italic>B</italic><sub>169</sub> = 10.7) (see Figure <xref ref-type="fig" rid="F2">2</xref>). Both of these sites are critical for the formation of the quaternary epitope targeted by the serum antibodies of CAP256 [<xref ref-type="bibr" rid="B34">34</xref>]. In particular, we found that viruses with an arginine at position 166 or a lysine at position 169 were significantly more sensitive to CAP256 serum neutralization than viruses with other amino acid residues at these positions, supporting previous studies [<xref ref-type="bibr" rid="B34">34</xref>]. Use of three other reference sequences identified the same 2 residues (Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Figures S1A-C). Experimental confirmation for the role of 166R and 169K in CAP256 neutralization is shown in Table <xref ref-type="table" rid="T1">1</xref> where mutations at both sites resulted in major reductions in neutralization titers, particularly in the CAP45 backbone.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Scaled Bayes factors for CAP256.</bold> Neutralization titers were strongly associated with sites 166 and 169 when the ConC reference sequence was used. These sites have previously been shown to contribute significantly to the epitope targeted by CAP256 antibodies [<xref ref-type="bibr" rid="B12">12</xref>]. Shaded regions indicate the degree of evidence for an association with ID<sub>50</sub>: white indicates no or negligible evidence, light grey indicates moderate evidence and dark grey indicates strong evidence [<xref ref-type="bibr" rid="B55">55</xref>]. Sites that were strongly associated with ID<sub>50</sub> are annotated with their HXB2 position and the amino acid found to be enriched among sensitive (high titer) viruses at that site.</p></caption><graphic xlink:href="1743-422X-10-347-2"/></fig><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>All sites with scaled Bayes factors ≥ 6</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="center"><bold>Serum</bold></th><th align="center"><bold>HXB2 position</bold></th><th align="center"><bold>Reference residue</bold><sup> <bold>‡</bold> </sup></th><th align="center"><bold>Scaled Bayes factor (LFDR)</bold></th><th align="center"><bold>Backbone</bold></th><th align="center"><bold>ID</bold><sub> <bold>50 </bold> </sub><bold>fold effect</bold></th></tr></thead><tbody><tr><td align="center" valign="bottom">CAP256<hr/></td><td align="center" valign="bottom">166<sup>†</sup><hr/></td><td align="center" valign="bottom">Arg<hr/></td><td align="center" valign="bottom">25.4 (0.0001)<hr/></td><td align="center" valign="bottom">CAP45<hr/></td><td align="center" valign="bottom">261.3<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">ConC<hr/></td><td align="center" valign="bottom">22.2<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">169<sup>†</sup><hr/></td><td align="center" valign="bottom">Lys<hr/></td><td align="center" valign="bottom">10.7 (0.083)<hr/></td><td align="center" valign="bottom">CAP45<hr/></td><td align="center" valign="bottom">365.8<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">ConC<hr/></td><td align="center" valign="bottom">15.3<hr/></td></tr><tr><td align="center" valign="bottom">CAP8<hr/></td><td align="center" valign="bottom">24<hr/></td><td align="center" valign="bottom">Met<hr/></td><td align="center" valign="bottom">6.4 (1.000)<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">NT<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">295<hr/></td><td align="center" valign="bottom">Asn<hr/></td><td align="center" valign="bottom">11.9 (1.000)<hr/></td><td align="center" valign="bottom">TRO<hr/></td><td align="center" valign="bottom">0.4<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">316<sup>†</sup><hr/></td><td align="center" valign="bottom">Thr<hr/></td><td align="center" valign="bottom">9.1 (1.000)<hr/></td><td align="center" valign="bottom">ConC<hr/></td><td align="center" valign="bottom">1.8<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">535<hr/></td><td align="center" valign="bottom">Ile<hr/></td><td align="center" valign="bottom">6.0 (1.000)<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">NT<hr/></td></tr><tr><td align="center" valign="bottom">CAP257<hr/></td><td align="center" valign="bottom">166<hr/></td><td align="center" valign="bottom">Arg<hr/></td><td align="center" valign="bottom">6.3 (0.738)<hr/></td><td align="center" valign="bottom">ConC<hr/></td><td align="center" valign="bottom">1.8<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">295<hr/></td><td align="center" valign="bottom">Asn<hr/></td><td align="center" valign="bottom">7.0 (0.665)<hr/></td><td align="center" valign="bottom">Q842<hr/></td><td align="center" valign="bottom">0.1<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">648<hr/></td><td align="center" valign="bottom">Glu<hr/></td><td align="center" valign="bottom">6.1 (0.757)<hr/></td><td align="center" valign="bottom">Du156<hr/></td><td align="center" valign="bottom">1.2<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">RHPA<hr/></td><td align="center" valign="bottom">2.9<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">702<hr/></td><td align="center" valign="bottom">Leu<hr/></td><td align="center" valign="bottom">6.2 (0.747)<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">NT<hr/></td></tr><tr><td align="center" valign="bottom">CAP255<hr/></td><td align="center" valign="bottom">332<hr/></td><td align="center" valign="bottom">Asn<hr/></td><td align="center" valign="bottom">8.0 (0.622)<hr/></td><td align="center" valign="bottom">Q23<hr/></td><td align="center" valign="bottom">4.0<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">TRO<hr/></td><td align="center" valign="bottom">2.9<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">ConC<hr/></td><td align="center" valign="bottom">0.2<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">334<hr/></td><td align="center" valign="bottom">Ser<hr/></td><td align="center" valign="bottom">6.8 (0.750)<hr/></td><td align="center" valign="bottom">TRO<hr/></td><td align="center" valign="bottom">>12.4<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">ConC<hr/></td><td align="center" valign="bottom">0.3<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">351<hr/></td><td align="center" valign="bottom">Glu<hr/></td><td align="center" valign="bottom">6.8 (0.750)<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">NT<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">837<hr/></td><td align="center" valign="bottom">Phe<hr/></td><td align="center" valign="bottom">10.1 (0.366)<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">NT<hr/></td></tr><tr><td align="center" valign="bottom">CAP177<hr/></td><td align="center" valign="bottom">209<hr/></td><td align="center" valign="bottom">Thr<hr/></td><td align="center" valign="bottom">8.8 (0.388)<hr/></td><td align="center" valign="bottom">TRO<hr/></td><td align="center" valign="bottom">0.4<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">332<sup>†</sup><hr/></td><td align="center" valign="bottom">Asn<hr/></td><td align="center" valign="bottom">7.3 (0.573)<hr/></td><td align="center" valign="bottom">ConC<hr/></td><td align="center" valign="bottom">1.7<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">Q23<hr/></td><td align="center" valign="bottom">11.0<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">TRO<hr/></td><td align="center" valign="bottom">>2.8<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">334<sup>†</sup><hr/></td><td align="center" valign="bottom">Ser<hr/></td><td align="center" valign="bottom">7.8 (0.511)<hr/></td><td align="center" valign="bottom">ConC<hr/></td><td align="center" valign="bottom">2.1<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">TRO<hr/></td><td align="center" valign="bottom">0.2<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">683<hr/></td><td align="center" valign="bottom">Lys<hr/></td><td align="center" valign="bottom">6.3 (0.689)<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">NT<hr/></td></tr><tr><td align="center" valign="bottom">CAP206<hr/></td><td align="center" valign="bottom">150<hr/></td><td align="center" valign="bottom">Met<hr/></td><td align="center" valign="bottom">6.7 (0.457)<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">NT<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">655<hr/></td><td align="center" valign="bottom">Lys<hr/></td><td align="center" valign="bottom">7.3 (0.384)<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">NT<hr/></td></tr><tr><td align="center" valign="bottom">CAP248<hr/></td><td align="center" valign="bottom">85<hr/></td><td align="center" valign="bottom">Val<hr/></td><td align="center" valign="bottom">6.5 (1.000)<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">NT<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">340<hr/></td><td align="center" valign="bottom">Glu<hr/></td><td align="center" valign="bottom">6.0 (1.000)<hr/></td><td align="center" valign="bottom">ConC<hr/></td><td align="center" valign="bottom">2.2<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">CAP45<hr/></td><td align="center" valign="bottom">0.6<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">651<hr/></td><td align="center" valign="bottom">Asn<hr/></td><td align="center" valign="bottom">7.4 (1.000)<hr/></td><td align="center" valign="bottom">ConC<hr/></td><td align="center" valign="bottom">2.0<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">CAP45<hr/></td><td align="center" valign="bottom">1.3<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">Du156<hr/></td><td align="center" valign="bottom">2.4<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">659<hr/></td><td align="center" valign="bottom">Asp<hr/></td><td align="center" valign="bottom">8.9 (1.000)<hr/></td><td align="center" valign="bottom">ConC<hr/></td><td align="center" valign="bottom">2.9<hr/></td></tr><tr><td align="center"> </td><td align="center"> </td><td align="center"> </td><td align="center"> </td><td align="center">CAP45</td><td align="center">0.5</td></tr></tbody></table><table-wrap-foot><p><sup>†</sup>Sites with <italic>q</italic> ≤ 1/3 based on the method of Gnanakaran et al. [<xref ref-type="bibr" rid="B15">15</xref>]. <sup>‡</sup>Amino acid found to be significantly enriched among sensitive (high titer) viruses based on our evolutionary model. Fold effect ≥ 2. NT = Not tested.</p></table-wrap-foot></table-wrap><p>We also found weak evidence of an association with neutralization sensitivity at sites 162, 163, 167, 176, 177, 181, 183, 187, 193 and 194 in V2 (2 < <italic>B</italic><sub><italic>k</italic></sub> < 6). The fact that our model assigns moderately large scaled Bayes factors to this cluster of sites is encouraging. However, there were many other sites throughout gp120 and gp41 with scaled Bayes factors in this intermediate region (2 < <italic>B</italic><sub><italic>k</italic></sub> < 6), as might be expected when multiple model comparisons are performed. To account for the fact that multiple model comparisons were carried out, we computed the local false discovery rate (LFDR) associated with the Bayes factor at each site (see Methods). As expected, the large Bayes factors at sites 166 and 169 were highly likely to be true positives with low LFDRs of 0.0001 and 0.083, respectively (Table <xref ref-type="table" rid="T1">1</xref>). The next highest scaled Bayes factor was 5.95 (LFDR = 0.492) at site 2 when the reference sequence contained an arginine at this position. Since this scaled Bayes factor was much smaller than those at positions 166 and 169 and below our significance threshold of 6, we did not regard any other amino acid positions as significantly associated with neutralization titer.</p></sec><sec><title>Prediction of sites targeted by PG9/PG16-like antibodies</title><p>In order to explore the utility of our model, we tested two additional sera predicted to have a similar specificity to CAP256, but with considerably lower titers. Four sites were found to be strongly associated (<italic>B</italic><sub><italic>k</italic></sub> ≥ 6) with neutralization by CAP8 serum antibodies (see the scaled Bayes factors in Table <xref ref-type="table" rid="T1">1</xref>; plots shown in Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Figure S2). The largest scaled Bayes factor of 11.9 was observed at position 295, a potential N-linked glycosylation (PNG) site. However, replacing the asparagine at this position in the TRO backbone did not reduce ID<sub>50</sub> titers. Site 316 in V3 was also found to be significantly associated with neutralization sensitivity (<italic>B</italic><sub>316</sub> = 9.1). Mutation of this residue in the ConC backbone resulted in a large drop in ID<sub>50</sub> titers from 11,000 to 6,000. Mutations in the V3 region are known to modulate neutralization sensitivity of the conserved V2 epitope recognized by PG9/PG16-like antibodies. Therefore, although no sites within the conserved C-strand in V2 were detected, the detection of residue 316 in V3 was consistent with a known neutralizing specificity in the CAP8 serum, which was previously shown to target a PG9/PG16-like trimer-specific epitope [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B35">35</xref>]. Sites 24 (<italic>B</italic><sub>24</sub> = 6.4) in the envelope signal peptide and site 535 (<italic>B</italic><sub>535</sub> = 6.0) in gp41 were also found to be associated with titer, though we do not expect these to contribute to CAP8 antibody neutralization and therefore did not test them experimentally.</p><p>For CAP257 serum, we identified signals at positions 166 (<italic>B</italic><sub>166</sub> = 6.3) and 295 (<italic>B</italic><sub>295</sub> = 7.0) in the V2 and V3 regions of gp120, respectively, and at positions 648 (<italic>B</italic><sub>648</sub> = 6.1) and 702 (<italic>B</italic><sub>702</sub> = 6.2) in gp41. Sites 166, 295 and 648 were tested with site-directed mutagenesis and the substitution of alanine at site 648 in the RHPA backbone was found to reduce ID<sub>50</sub> titers by more than 2 fold (Table <xref ref-type="table" rid="T1">1</xref>). Serum from this individual has previously been shown to contain at least two distinct antibody specificities; one targeting a PG9/PG16-like trimer-specific epitope, followed by a second that overlaps the CD4 binding site (Wibmer et al., submitted). Although our evolutionary model did identify position 166 in the V2 region, this could not be experimentally validated (Table <xref ref-type="table" rid="T1">1</xref> and Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Figure S3). However, introduction of an R166A mutation completely abrogated neutralization by serum from an earlier time point, confirming the existence of a PG9/PG16-like specificity in CAP257 (Wibmer et al., submitted).</p></sec><sec><title>Prediction of sites associated with N332-dependent broadly neutralizing antibodies</title><p>Two CAPRISA sera that targeted another broadly neutralizing antibody epitope were tested with our model. The Bayes factors for CAP255 serum identified three sites in the C3 region of gp120 (sites 332, 334 and 351) (see Figure <xref ref-type="fig" rid="F3">3</xref>; plots for alternative reference sequences are provided in Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Figure S4). Two of these sites in close proximity on the primary sequence were strongly associated (<italic>B</italic><sub><italic>k</italic></sub> ≥ 6) with neutralization titer, namely positions 332 and 334. For site 332, all of the reference sequences contained an asparagine at this position and yielded a large scaled Bayes factor of 8.0, suggesting that viruses with this amino acid were sensitive to CAP255 antibody neutralization. This result was supported by site-directed mutagenesis in the Q23 (4.0 fold) and TRO (2.9 fold) envelope backbones but not in ConC. Our model predictions also showed a large scaled Bayes factor of 6.8 at site 334, which forms part of the same N-linked glycosylation motif, when the reference residue was serine. The involvement of this site was confirmed experimentally in the TRO envelope backbone (>12.4 fold) only. The Q23 reference sequence has a threonine at this position which yields a Bayes factor close to zero, suggesting that this amino acid is not enriched among the sensitive viruses in the panel. This is perhaps surprising, given that a threonine at this position also permits the attachment of an N-linked glycan at site 332. The identification of 332 and 334 as contributing to the CAP255 epitope confirms previous mapping data showing that these antibodies are dependent on the N332 glycan in the C3 region [<xref ref-type="bibr" rid="B12">12</xref>], similar to many of the recently described PGT monoclonal antibodies [<xref ref-type="bibr" rid="B20">20</xref>].</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Scaled Bayes factors for CAP255.</bold> There was strong evidence of an association with ID<sub>50</sub> titer at sites 332, 334 and 351 when the ConC reference sequence was used. The predictions at sites 332 and 334 were tested and validated experimentally (see Table <xref ref-type="table" rid="T1">1</xref>). There were several sites surrounding these three positions in the C3 region that were moderately associated with neutralization sensitivity. The background shading is described in the legend of Figure <xref ref-type="fig" rid="F2">2</xref>.</p></caption><graphic xlink:href="1743-422X-10-347-3"/></fig><p>In addition to sites 332 and 334, we also obtained a large scaled Bayes factor of 6.82 at position 351 when this site contained an isoleucine in the reference sequence for the CAP255 data. Our model also predicted site 837 in gp41. While we do not expect that this position lies within a CAP255 epitope, it is possible that amino acid changes in gp41 contribute to neutralization sensitivity by influencing epitope accessibility through conformational changes to the envelope complex [<xref ref-type="bibr" rid="B36">36</xref>,<xref ref-type="bibr" rid="B37">37</xref>].</p><p>In addition to the three sites in the C3 region with <italic>B</italic><sub><italic>k</italic></sub> ≥ 6, several surrounding sites were weakly associated with CAP255 neutralization sensitivity and had 2 < <italic>B</italic><sub><italic>k</italic></sub> < 6 for at least one reference sequence (see Figure <xref ref-type="fig" rid="F3">3</xref>). Many of these residues were in close proximity on the tertiary structure as illustrated in Figure <xref ref-type="fig" rid="F4">4</xref>A. Indeed, residues in this spatial region were found to have higher posterior probabilities of belonging to a conformational epitope after applying our three-dimensional epitope prediction algorithm (see Figure <xref ref-type="fig" rid="F4">4</xref>B). Site 332 had the highest posterior probability of 0.191 and neighboring residues had slightly lower posterior probabilities.</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>Three-dimensional predictions for CAP255. (A)</bold> Amino acid residues that were weakly (2 ≤ <italic>B</italic><sub><italic>k</italic></sub> < 4), moderately (4 ≤ <italic>B</italic><sub><italic>k</italic></sub> < 6) and strongly (<italic>B</italic><sub><italic>k</italic></sub> ≥ 6) associated with ID<sub>50</sub> titer using the ConC reference sequence are shown with light, intermediate and dark green, respectively. There is evidence for a cluster of sites with moderately large Bayes factors on the three-dimensional surface, as might be expected of a B cell epitope. <bold>(B)</bold> Posterior probabilities of a conformational epitope using the three-dimensional Metropolis algorithm. The surface of the protein (PDB ID: 2B4C) was shaded from dark blue (posterior probability = 0) to red (posterior probability = 1) according to the posterior probability assigned to each amino acid residue. There was evidence for a conformational epitope involving the C3 region (residues in the light blue region have posterior probabilities of approximately 0.2).</p></caption><graphic xlink:href="1743-422X-10-347-4"/></fig><p>It is possible that the association with neutralization sensitivity at some of these spatially-proximal sites might be a consequence of compensatory mutations. To investigate this, we used the Bayesian graphical model of Poon et al. [<xref ref-type="bibr" rid="B38">38</xref>] to tease apart direct and indirect correlations between residue positions and ID<sub>50</sub> titer. A network graph indicating the direct associations (with posterior probabilities > 0.75) of all sites with <italic>B</italic><sub><italic>k</italic></sub> > 4 using the CAP255 data is shown in Figure <xref ref-type="fig" rid="F5">5</xref>. We found that all of these sites were directly associated with ID<sub>50</sub> titer and therefore that none of our predicted associations were likely to be due to compensation for resistance-imparting mutations elsewhere. Resistance to CAP255 serum could therefore be attributed to mutations at several sites that either constitute the binding interface of a single antibody or represent independent targets of multiple antibodies. This was the case for all sites with <italic>B</italic><sub><italic>k</italic></sub> > 6 across all sera (see the supporting information for each serum).</p><fig id="F5" position="float"><label>Figure 5</label><caption><p><bold>Bayesian evolutionary-network model for CAP255 [</bold>[<xref ref-type="bibr" rid="B38">38</xref>]<bold>].</bold> The red node corresponds to ID<sub>50</sub> titer and all other nodes represent sites in the HIV-1 envelope. Nodes with scaled Bayes factors > 6 are shaded in dark green, while nodes with scaled Bayes factors between 2 and 6 are shaded in light green. An edge connecting two nodes indicates that there is a direct association between the two nodes. Edges are labeled with the estimated posterior probability of an interaction between the nodes they connect. Only sites with scaled Bayes factors > 4 or posterior probabilities of an association with such a site > 0.75 are shown. Since all of the sites with Bayes factors > 4 are directly connected to the ID<sub>50</sub> node, none of the predicted associations could be attributed to compensatory mutations.</p></caption><graphic xlink:href="1743-422X-10-347-5"/></fig><p>For CAP177 serum, our model predicted four sites that influenced neutralization sensitivity with strong support (<italic>B</italic><sub><italic>k</italic></sub> ≥ 6). These included sites 209 in the C2 region, 332 and 334 in the C3 region and 683 in the membrane proximal external region of gp41 (Table <xref ref-type="table" rid="T1">1</xref>, Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Figure S5). Asparagine was observed at a significantly higher frequency at position 332 among neutralization-sensitive viruses than among neutralization-resistant viruses, similar to CAP255 (<italic>B</italic><sub>332</sub> = 7.3). This result was experimentally validated in the Q23, TRO and, to a lesser extent, ConC envelope backbones (see Table <xref ref-type="table" rid="T1">1</xref>). Our model also predicted that a serine residue at site 334, which forms part of the same N-linked glycosylation motif, imparts neutralization sensitivity, with a large scaled Bayes factor of 7.8 observed at this position. Replacing the serine residue with an alanine in the ConC backbone resulted in a >2 fold decrease in ID<sub>50</sub> titer, although this same mutation produced a five-fold increase in ID<sub>50</sub> titers with the TRO backbone. Nonetheless, these experimental results confirmed the importance of this site as a determinant of neutralization titer. This is in line with experimental mapping studies in CAP177, which, like CAP255, showed the presence of antibodies recognizing 332-dependent PGT-like epitopes, though these differ subtly from those in CAP255 in their dependence on variable loops [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B35">35</xref>]. There was, however, less evidence of an association with titer at spatially-proximal sites and consequently our three-dimensional epitope prediction method did not identify a similar region of high posterior probabilities for CAP177 as for CAP255. The prediction that position 209 (<italic>B</italic><sub>209</sub> = 8.8) influenced neutralization sensitivity to CAP177 sera could not be experimentally validated in the envelope backbone tested here. We did not attempt to validate site 683 (<italic>B</italic><sub>683</sub> = 6.3).</p><p>Overall, our model was very effective in predicting sites associated with sensitivity to PGT-like antibodies that are dependent on the glycan at position 332 in both CAP255 and CAP177. These specificities appear to be relatively common among broadly neutralizing sera [<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B29">29</xref>,<xref ref-type="bibr" rid="B33">33</xref>,<xref ref-type="bibr" rid="B39">39</xref>], although strain-specific N332-dependent antibodies have also been reported [<xref ref-type="bibr" rid="B40">40</xref>].</p></sec><sec><title>Identification of sites forming part of epitopes in gp41</title><p>For CAP206, our model predicted two amino acid positions that were strongly associated with antibody neutralization across six different reference sequences (see Table <xref ref-type="table" rid="T1">1</xref> and Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Figure S6). One of these predictions, site 150 (<italic>B</italic><sub>150</sub> = 6.7), was in the V1 region, while the other prediction, site 655 (<italic>B</italic><sub>655</sub> = 7.3), was near the membrane proximal external region in gp41. The latter prediction is consistent with available mapping data, showing that neutralization breadth in CAP206 is mediated largely by antibodies targeting this region [<xref ref-type="bibr" rid="B39">39</xref>]. This is further supported by the recent isolation from CAP206 memory B cells of the moderately broad monoclonal antibody, CAP206-CH12, which recognizes an epitope in the membrane proximal external region [<xref ref-type="bibr" rid="B41">41</xref>].</p><p>For the final serum examined, CAP248, mapping studies have shown that the serum antibodies target a quaternary epitope that has not yet been defined, but is not PG9/PG16-like [<xref ref-type="bibr" rid="B12">12</xref>]. We found associations between the amino acids at positions 651 (<italic>B</italic><sub>651</sub> = 7.4) and 659 (<italic>B</italic><sub>659</sub> = 8.9) adjacent to the membrane proximal external region of gp41 and neutralization sensitivity (Table <xref ref-type="table" rid="T1">1</xref> and Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Figure S7). In particular, our model revealed that viruses with an asparagine at position 651 or an aspartic acid at position 659 exhibited greater sensitivity to CAP248 antibody neutralization than other viruses. Interestingly, both sites lie between two gp41 positions that are known to affect antibody binding to epitopes in gp120 [<xref ref-type="bibr" rid="B36">36</xref>,<xref ref-type="bibr" rid="B37">37</xref>]. Both of these predictions were confirmed experimentally (see Table <xref ref-type="table" rid="T1">1</xref>). Our evolutionary model also identified sites 85 (<italic>B</italic><sub>85</sub> = 6.5) and 340 (<italic>B</italic><sub>340</sub> = 6.0) in gp120 as significantly associated with titer. The prediction at position 340 was confirmed by mutagenesis, while site 85 was not tested experimentally.</p></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>We have developed a novel computational approach to identify amino acid residues in HIV-1 envelope glycoproteins that are targeted by serum neutralizing antibodies. The method can be used to identify neutralizing antibody epitopes when neutralization sensitivity to the serum is available for a large panel of sequenced viruses. Such data will become increasingly available as large-scale efforts to investigate HIV-1 neutralization serotypes are undertaken. The method is an extension of the evolutionary model of Lacerda et al. [<xref ref-type="bibr" rid="B42">42</xref>] and allows the distribution of amino acids at each site to depend on neutralization sensitivities at the tips of the branches along a phylogeny. Bayes factors were used to assess the fit of this model at each site relative to that of a model in which the virus evolves independently of the antibody response. Large Bayes factors indicated positions in the alignment where neutralization sensitivity was significantly associated with amino acid composition after accounting for the phylogeny.</p><p>This method was applied to neutralization datasets of seven HIV-1 subtype C serum samples that were previously shown to have neutralization breadth [<xref ref-type="bibr" rid="B12">12</xref>]. In addition, epitope mapping data was available for six of the seven samples [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B34">34</xref>,<xref ref-type="bibr" rid="B39">39</xref>], providing an opportunity to compare our computational method with experimental approaches to epitope mapping. For each serum sample, envelope sequences and neutralization sensitivities at ID<sub>50</sub> were available from a multi-clade panel of 225 pseudoviruses. Our model identified two to four sites per sample and 24 predictions across all sera that were strongly associated with neutralization sensitivity. We were able to confirm ten of the fifteen sites that we tested using site-directed mutagenesis. In many cases, these corresponded to sites that had previously been linked to antibody neutralization in these and other broadly neutralizing sera. This included two positions in the V2 region (166 and 169) that contributed to a trimer-specific PG9/PG16-like epitope [<xref ref-type="bibr" rid="B34">34</xref>] and the glycan at site 332 that is crucial for the epitope targeted by many of the PGT monoclonal antibodies [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B20">20</xref>,<xref ref-type="bibr" rid="B35">35</xref>]. We also identified two novel sites near the membrane proximal external region in gp41 that influenced neutralization sensitivity to the CAP248 sera and another two positions in the same region that are associated with sensitivity to CAP206 and CAP257 antibodies. The CAP257 serum was unusual in that neutralization breadth has been shown to be mediated by at least two distinct specificities, which arose sequentially and then waned (Wibmer et al., submitted). The three-year time point, which was used for this study, fell between two specificities when the titers for each were lower than their peak. Nonetheless, our model found a significant association with titer at position 166, which is consistent with the PG9/PG16-like specificity that developed initially.</p><p>Our model is similar to the one developed by Gnanakaran et al. [<xref ref-type="bibr" rid="B15">15</xref>] who applied the phylogenetically-corrected method of Bhattacharya et al. [<xref ref-type="bibr" rid="B43">43</xref>] to identify signature positions in the HIV-1 envelope that were associated with antibody neutralization. While their contingency table-based approach requires that viruses be classified into discrete neutralization phenotypes, our evolutionary model permits the use of continuous neutralization measures, which is likely to enhance the statistical power to detect an association. In addition, our method naturally accounts for the uncertainty of the unknown ancestral sequences, while their approach treats the reconstructed sequences as observed data. In contrast to our current model implementation, the method of Gnanakaran et al. [<xref ref-type="bibr" rid="B15">15</xref>] does not require the specification of a neutralization sensitive sequence for each serum sample and can be used to identify combinations of amino acids and sites that influence antibody neutralization sensitivity. Although such combinations could be accommodated straightforwardly in our model, it would be computationally expensive to do so and the increase in the number of tests performed could greatly reduce statistical power. Instead, we identified a set of potentially sensitive amino acid residues at each site based on a selection of pseudoviruses (reference sequences) that have previously been shown to be sensitive to the antibodies of a particular serum sample.</p><p>For comparison with our results, we applied the method of Gnanakaran et al. [<xref ref-type="bibr" rid="B15">15</xref>] to our neutralization datasets. Of the 24 associations that we identified across the seven sera (see Table <xref ref-type="table" rid="T1">1</xref>), five were also detected with the method of Gnanakaran et al. [<xref ref-type="bibr" rid="B15">15</xref>] using a <italic>q</italic>-value threshold of 1/3 and three alternative neutralization classifications (see Additional file <xref ref-type="supplementary-material" rid="S2">2</xref>: Table S3). We were able to detect four of our fifteen experimentally-validated associations using this technique. The method of Gnanakaran et al. [<xref ref-type="bibr" rid="B15">15</xref>] yielded a further thirteen predictions with <italic>q</italic> ≤ 1/3 across the seven sera, though we did not attempt to validate these experimentally.</p><p>The effects of experimental amino acid mutations in different backbones were highly variable, as has been observed in numerous previous studies [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B20">20</xref>,<xref ref-type="bibr" rid="B29">29</xref>,<xref ref-type="bibr" rid="B34">34</xref>]. This reflects the extremely complex nature of neutralization escape. Our evolutionary model makes the simplifying assumption that each site evolves independently and, as such, does not account for context-specific effects that may alter substitution rates differentially in distinct viruses. This assumption does not hold <italic>in vitro</italic> where the fold effect of mutagenesis depends not only on the mutated residue, but also on the entire backbone context. In addition to mutations directly within epitopes, neutralization resistance may also be meditated by steric shielding of epitopes and distal changes, which may be compensatory or may drive conformational changes that limit accessibility of antibodies to epitopes. While our model may detect non-causal associations between titer and the amino acid composition at sites that experience such compensatory mutations, it is not expected that these associations will be confirmed by mutagenesis. In order to assess the extent to which this was the case, we implemented the evolutionary network model of Poon et al. [<xref ref-type="bibr" rid="B38">38</xref>]. Across all seven sera, we found that all sites with scaled Bayes factors greater than six directly coevolved with titer, indicating that our predictions corresponded to potential resistance sites and not compensatory mutations.</p><p>To account for multiple testing, we computed the local false discovery rate (LFDR) associated with each of our predictions; that is, the probability that a site is incorrectly found to be associated with titer given the codon data at the site and the prior probability of no association (see Methods). To compute the prior probability of no association, we first estimated the probability of obtaining a positive scaled Bayes factor when no association exists by randomly shuffling the assignment of titers to sequences. For computational reasons, we were only able to perform 100 such permutations. We then made the conservative assumption that a site that is truly associated with titer will always have a positive scaled Bayes factor. Consequently, the LFDR estimates reported in Table <xref ref-type="table" rid="T1">1</xref> should be interpreted as upper bounds on the true LFDRs. As such, many of these estimates are quite large. Indeed, the estimates of the LFDRs for all CAP8 and CAP248 predictions were 1, since the conservative prior probability of no association was 1 for both of these sera. The true LFDRs for these predictions will likely be less than this.</p><p>To identify amino acid positions that collectively provide support for the existence of a conformational epitope, we introduced a model in which all sites within a sphere on the tertiary structure evolved according to the epitope model, while the evolution of all other sites was described by the non-epitope model. A Metropolis algorithm was used to explore the posterior density of the center and radius of the sphere and thereby determine the most likely location and size of a conformational epitope on the envelope structure. This approach predicted a known epitope in the C3 region targeted by the CAP255 neutralizing serum. No three-dimensional clusters with high posterior probability were predicted in CAP177, CAP248 or CAP206. The three-dimensional prediction algorithm did, however, find evidence for an epitope targeted by the CAP256, CAP8 and CAP257 antibodies in the V3 region of gp120 (largest posterior probability of 0.367, 0.654 and 0.361, respectively; see Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Figures S1E, S2E and S3D). A more realistic model would treat a B cell epitope as a patch of surface-exposed amino acids and estimate the location, shape and size of this surface area. Additional structures including the scaffolded V2 structure [<xref ref-type="bibr" rid="B44">44</xref>] and particularly future structures of trimeric forms of the envelope would greatly enhance the power of this approach.</p><p>The unobserved titers at all ancestral nodes represent a missing data problem. Unfortunately, integration over all of these unknown variables is not a computationally feasible option for large phylogenies. Instead, one must resort to an imputation procedure. In the model presented here, the ancestral nodes were all assigned the median value of the observed titers at the tips of the phylogeny. We obtained very similar results when codon evolution along the internal branches was modeled with a standard, titer-independent MG94×HKY85 model. Although the internal branches alone cannot provide evidence of an association with neutralization sensitivity in our current model implementation, they are informative about the parameters of the evolutionary model and may therefore increase the power of our method to detect associations based on evolution along the terminal branches. This further distinguishes our approach from other phylogenetically-corrected methods, such as that of Gnanakaran et al. [<xref ref-type="bibr" rid="B15">15</xref>], that only consider mutational patterns along the terminal branches. An alternative approach is to reconstruct the ancestral titers based on a model of titer evolution. We investigated this strategy using the ancestral reconstruction method of Felsenstein [<xref ref-type="bibr" rid="B45">45</xref>]. Briefly, the ancestral titer at each parent node in the phylogeny was imputed as the average of the observed and estimated titers at its two daughter nodes, weighted by the reciprocal of their branch lengths after accounting for the variability of the titer estimates. We obtained many additional predictions using this procedure, presumably because the inferred ancestral titers provide more information than the median and are treated as if observed without error (see Additional file <xref ref-type="supplementary-material" rid="S2">2</xref>: Table S4 and Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Figure S8). We therefore consider our use of median titers at the ancestral nodes as conservative. Existing models of quantitative phenotype evolution, including the Brownian motion-based model of Felsenstein [<xref ref-type="bibr" rid="B45">45</xref>] and the Ornstein-Uhlenbeck process [<xref ref-type="bibr" rid="B46">46</xref>], assume continuous sample paths and are therefore not appropriate for modeling neutralization titer that undergoes abrupt and discontinuous changes in equilibrium level induced by specific amino acid substitutions. Modeling the evolution of quantitative traits with discontinuous paths is an active area of research [<xref ref-type="bibr" rid="B47">47</xref>]. A more realistic model of titer evolution may enhance the statistical power of our method by providing further evidence of genotype/phenotype associations along internal branches of the phylogeny. Our model can readily accommodate ancestral titers inferred by any means and we are currently investigating alternative methods for inferring these unknown titers.</p><p>To our knowledge, this is the first attempt to identify conformational B cell epitopes on a tertiary structure based on the evolutionary history of a panel of viruses. In principle, our model could be combined with other structure- and mimotope-based methods that assign scores to residues based on their structural properties and peptide-binding affinities.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>Our method is an effective tool for detecting sequence positions that contribute to neutralizing antibody sensitivity, even within quaternary epitopes on the trimeric envelope complex. However, in the absence of information on the host immune responses experienced by each of the viruses in the panel, we cannot determine whether the correlations between amino acid frequencies and neutralization phenotype at these sites are a consequence of selective immune pressure or random genetic drift. Furthermore, sites that influence neutralization sensitivity through insertions and deletions that alter epitope binding affinity or through shifts in glycosylation patterns will not be detected by our approach [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B8">8</xref>]. Nonetheless, our results provide strong support for the use of evolutionary models as a means to identify key residues in complex B cell epitopes. The conformational nature of these epitopes renders them difficult to discover with currently available computational tools.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Ethics statement</title><p>The CAPRISA Acute Infection study was reviewed and approved by the research ethics committees of the University of KwaZulu-Natal (E013/04), the University of Cape Town (025/2004) and the University of the Witwatersrand (MM040202). All participants provided written informed consent for study participation.</p></sec><sec><title>Neutralization and sequence data</title><p>We analyzed neutralization datasets from seven HIV-1 subtype C-infected women from the CAPRISA 002 Acute Infection Study [<xref ref-type="bibr" rid="B48">48</xref>]. These subjects, namely CAP8, CAP177, CAP206, CAP248, CAP255, CAP256 and CAP257, all developed broadly cross-neutralizing antibodies against HIV-1 [<xref ref-type="bibr" rid="B12">12</xref>]. The potency of the serum antibodies collected at 3 years post-infection from each subject was measured against a multiclade panel of 225 envelope-pseudotyped viruses as part of the Neutralization Serotype Discovery Project (NSDP) (see Figure <xref ref-type="fig" rid="F1">1</xref>A). For each pseudovirus, the neutralization titer of a subject’s serum was recorded as the reciprocal of the maximal plasma dilution that could inhibit 50% of viral entry (ID<sub>50</sub>). The data for CAP256 was used to decide on the optimal modeling strategy.</p><p>HIV-1 gp160 sequences from 225 panel viruses were codon aligned with the hidden Markov model implemented in the HIVAlign tool of the Los Alamos National Laboratory (LANL) HIV database (<ext-link ext-link-type="uri" xlink:href="http://www.hiv.lanl.gov/content/sequence/HIV/HIVTools.html">http://www.hiv.lanl.gov/content/sequence/HIV/HIVTools.html</ext-link>). Nucleotide sites at which more than 10% of the sequences contained gaps were removed by deleting the corresponding codon. In doing so, we restricted our analyses to the regions of envelope that are readily aligned [<xref ref-type="bibr" rid="B15">15</xref>]. The resulting alignment contained 818 codons, spanning HXB2 envelope positions 1 to 856, excluding sites 13–16, 31, 137–149, 189–190, 310–311, 354–355, 395–408, 460–462 and 514, and including four insertions relative to HXB2. The HIV-1 gp160 alignment is provided as Additional file <xref ref-type="supplementary-material" rid="S3">3</xref>: Dataset S1. The Genbank accession numbers and neutralization titers for the envelope sequences are provided in Additional file <xref ref-type="supplementary-material" rid="S4">4</xref>: Dataset S2.</p><p>The alignment was screened for recombination using GARD [<xref ref-type="bibr" rid="B49">49</xref>] with a general time reversible nucleotide model and among-site rate variation modeled with a three-category general discrete distribution. A single recombination breakpoint was detected at nucleotide position 1331. Although recombination can be accommodated straightforwardly in our modeling approach by using different phylogenies for each non-recombinant fragment, this would have required the estimation of 447 additional branch length parameters. Rather than attempt this, we conducted a simulation study to assess whether a single recombination breakpoint was likely to substantively affect our results. Briefly, we simulated 10 alignments with a single recombination breakpoint using NetRecodon [<xref ref-type="bibr" rid="B50">50</xref>] and evolved log titers along the phylogeny corresponding to the largest non-recombinant partition of each alignment according to Brownian motion. We applied our method to each of the 10 simulated datasets ignoring recombination and did not observe any false positives, suggesting that our results are robust to recombination at this level.</p><p>A single, maximum likelihood phylogeny for the panel of viruses was inferred with PhyML [<xref ref-type="bibr" rid="B51">51</xref>] by specifying a general time reversible model of nucleotide substitution and an estimated proportion of invariable sites. Site-to-site variation in the evolutionary rate of the variable sites was modeled by a four-category discrete gamma distribution with unit mean. Subtree pruning and regrafting and nearest neighbor interchange were used to search the tree space for the optimal topology, starting with a BioNJ tree and five random trees. The root was identified by including an outgroup of SIV and HIV-1 groups N, O and P reference sequences obtained from the LANL HIV sequence database.</p></sec><sec><title>Evolutionary models</title><p>Our computational approach to identifying sites targeted by broadly neutralizing antibodies is an extension of the method of Lacerda et al. [<xref ref-type="bibr" rid="B42">42</xref>], which was originally developed to predict T cell epitopes by identifying escape sites where viral evolution correlated with the immune type of the host. Here, we did not have information on the host conditions under which each of the panel viruses evolved and therefore could not infer immune escape <italic>per se</italic>. Instead, we identified amino acid positions in HIV-1 envelope that influenced sensitivity to antibody neutralization across a phylogeny. This was achieved by specifying an evolutionary model that allowed the amino acid equilibrium frequencies at each site to depend on neutralization titer. To avoid estimating 19 frequency parameters at each site, we selected a neutralization-sensitive virus, which we refer to as the “reference sequence,” and modeled only the equilibrium frequency of the amino acid present in this sequence at each site. Reference sequences were selected on the basis of high neutralization titers (which varied depending on the potency of each serum sample) and the availability of cloned envelopes for later validation studies. Where possible, the early autologous virus (isolated from the individual from whom the broadly neutralizing serum was obtained) was also used, as most of these viruses are known to be sensitive to the broadly neutralizing antibodies that develop later. The reference sequences used for each serum and their Genbank accession numbers are provided in Additional file <xref ref-type="supplementary-material" rid="S2">2</xref>: Table S1.</p><p>We adapted the evolutionary model of Halpern and Bruno [<xref ref-type="bibr" rid="B52">52</xref>] in which the instantaneous rate of codon substitution <italic>q</italic><sup><italic>k</italic></sup><sub><italic>ij</italic></sub> at site <italic>k</italic> is parameterized as the product of the mutation rate <italic>μ</italic><sub><italic>ij</italic></sub> and the probability of fixation <italic>f</italic><sup><italic>k</italic></sup><sub><italic>ij</italic></sub> relative to that of a neutral mutation; that is, <italic>q</italic><sup><italic>k</italic></sup><sub><italic>ij</italic></sub> ∝ <italic>μ</italic><sub><italic>ij</italic></sub> × <italic>f</italic><sup><italic>k</italic></sup><sub><italic>ij</italic></sub> / (1/<italic>N</italic>) for codons <italic>i</italic> and <italic>j</italic> and effective haploid population size <italic>N</italic>. Assuming a time-reversible model of codon substitution, the authors showed that the fixation probability of a mutation from codon <italic>i</italic> to codon <italic>j</italic> with relative selective advantage 1 + <italic>s</italic>, <italic>s</italic> ≪ 1, can be written as</p><p><disp-formula id="bmcM1"><label>(1)</label><mml:math id="M1" name="1743-422X-10-347-i1" overflow="scroll"><mml:mrow><mml:msubsup><mml:mi mathvariant="italic">f</mml:mi><mml:mi mathvariant="italic">ij</mml:mi><mml:mi>k</mml:mi></mml:msubsup><mml:mo>=</mml:mo><mml:msup><mml:mi>N</mml:mi><mml:mrow><mml:mo>-</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:msup><mml:mfrac><mml:mrow><mml:mtext>ln</mml:mtext><mml:mfenced open="(" close=")"><mml:mfrac><mml:mrow><mml:msubsup><mml:mi mathvariant="italic">ν</mml:mi><mml:mi>j</mml:mi><mml:mi>k</mml:mi></mml:msubsup><mml:msub><mml:mi>μ</mml:mi><mml:mi mathvariant="italic">ji</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:msubsup><mml:mi mathvariant="italic">ν</mml:mi><mml:mi>i</mml:mi><mml:mi>k</mml:mi></mml:msubsup><mml:msub><mml:mi>μ</mml:mi><mml:mi mathvariant="italic">ij</mml:mi></mml:msub></mml:mrow></mml:mfrac></mml:mfenced></mml:mrow><mml:mrow><mml:mn>1</mml:mn><mml:mo>-</mml:mo><mml:mfrac><mml:mrow><mml:msubsup><mml:mi mathvariant="italic">ν</mml:mi><mml:mi>i</mml:mi><mml:mi>k</mml:mi></mml:msubsup><mml:msub><mml:mi>μ</mml:mi><mml:mi mathvariant="italic">ij</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:msubsup><mml:mi mathvariant="italic">ν</mml:mi><mml:mi>j</mml:mi><mml:mi>k</mml:mi></mml:msubsup><mml:msub><mml:mi>μ</mml:mi><mml:mi mathvariant="italic">ji</mml:mi></mml:msub></mml:mrow></mml:mfrac></mml:mrow></mml:mfrac><mml:mo>,</mml:mo></mml:mrow></mml:math></disp-formula></p><p>where <inline-formula><mml:math id="M2" name="1743-422X-10-347-i2" overflow="scroll"><mml:msubsup><mml:mi>ν</mml:mi><mml:mi>j</mml:mi><mml:mi>k</mml:mi></mml:msubsup></mml:math></inline-formula> is the equilibrium frequency of codon <italic>j</italic> at site <italic>k</italic>.</p><p>We parameterized the site-specific equilibrium frequencies as</p><p><disp-formula id="bmcM2"><label>(2)</label><mml:math id="M3" name="1743-422X-10-347-i3" overflow="scroll"><mml:mrow><mml:msubsup><mml:mi>ν</mml:mi><mml:mi>j</mml:mi><mml:mi>k</mml:mi></mml:msubsup><mml:mo>=</mml:mo><mml:mfenced open="{"><mml:mtable columnalign="center"><mml:mtr columnalign="center"><mml:mtd columnalign="center"><mml:msub><mml:mi>γ</mml:mi><mml:mi>k</mml:mi></mml:msub><mml:mspace width="0.75em"/><mml:mo>×</mml:mo><mml:mfrac><mml:msub><mml:mi>π</mml:mi><mml:mi>j</mml:mi></mml:msub><mml:msub><mml:mi>π</mml:mi><mml:mi>Γ</mml:mi></mml:msub></mml:mfrac><mml:mspace width="4.5em"/><mml:mi>if</mml:mi><mml:mspace width="0.25em"/><mml:mi>j</mml:mi><mml:mo>∈</mml:mo><mml:mi>Γ</mml:mi></mml:mtd></mml:mtr><mml:mtr columnalign="center"><mml:mtd columnalign="center"><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>1</mml:mn><mml:mo>-</mml:mo><mml:msub><mml:mi>γ</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mrow></mml:mfenced><mml:mspace width="0.75em"/><mml:mo>×</mml:mo><mml:mfrac><mml:msub><mml:mi>π</mml:mi><mml:mi>j</mml:mi></mml:msub><mml:mrow><mml:mn>1</mml:mn><mml:mo>-</mml:mo><mml:msub><mml:mi>π</mml:mi><mml:mi>Γ</mml:mi></mml:msub></mml:mrow></mml:mfrac><mml:mspace width="1.25em"/><mml:mi>if</mml:mi><mml:mspace width="0.25em"/><mml:mi>j</mml:mi><mml:mo>∉</mml:mo><mml:mi>Γ</mml:mi><mml:mo>,</mml:mo></mml:mtd></mml:mtr></mml:mtable></mml:mfenced></mml:mrow></mml:math></disp-formula></p><p>where <italic>γ</italic><sub><italic>k</italic></sub> is the equilibrium frequency of the reference amino acid at site <italic>k</italic>, <italic>π</italic><sub><italic>j</italic></sub> is the equilibrium frequency of codon <italic>j</italic> in the absence of selection, Γ represents the set of codons that encode the reference amino acid and <italic>π</italic><sub><italic>Γ</italic></sub> = ∑ <sub><italic>i</italic> ∈ <italic>Γ</italic></sub><italic>π</italic><sub><italic>i</italic></sub>. The factor involving the <italic>π</italic><sub><italic>j</italic></sub> terms distributes the reference amino acid frequency among the codons that encode it in such a way as to maintain the codon usage bias observed over the entire alignment.</p><p>We used an HKY85 model for the mutation rate <italic>μ</italic><sub><italic>ij</italic></sub>, with the nucleotide equilibrium frequencies estimated empirically from the full alignment. Because we fitted this model to coding sequences, our estimates will not only reflect the mutational process, but will also capture selection induced through the genetic code. We do not expect that our model results will be sensitive to this misspecification. Ideally, the mutation parameters should be estimated from non-coding nucleotide sequences [<xref ref-type="bibr" rid="B52">52</xref>]. The estimated nucleotide frequencies from the mutation model were used to construct F1 × 4 estimates of the codon frequencies <italic>π</italic><sub><italic>j</italic></sub> expected in the absence of selection.</p><p>Assuming a time-reversible mutation process, substituting (2) into (1) yields <italic>f</italic><sup><italic>k</italic></sup><sub><italic>ij</italic></sub> = 1/<italic>N</italic>, and hence <italic>s</italic> = 0, for all mutations that do not involve a substitution of the reference residue for or by another amino acid. This formulation therefore permits only directional selection that alters the codon equilibrium frequencies <inline-formula><mml:math id="M4" name="1743-422X-10-347-i4" overflow="scroll"><mml:msubsup><mml:mi mathvariant="italic">ν</mml:mi><mml:mi>j</mml:mi><mml:mi>k</mml:mi></mml:msubsup></mml:math></inline-formula> based on the frequency <italic>γ</italic><sub><italic>k</italic></sub> of the reference amino acid at site <italic>k</italic>. Nonsynonymous mutations that do not involve the reference amino acid would have the same fixation probability (1/<italic>N</italic>) as synonymous mutations and would therefore be modeled as selectively neutral. However, nonsynonymous mutations at different sites are likely to evolve under purifying and diversifying selection that alters the fixation probabilities without affecting codon frequencies. To accommodate this in our model, we parameterize the fixation probability of a nonsynonymous mutation at site <italic>k</italic> as <italic>ω</italic><sub><italic>k</italic></sub> × 1/<italic>N</italic>. A value of <italic>ω</italic><sub><italic>k</italic></sub> = 1 indicates that nonsynonymous and synonymous mutations are fixed with the same probability at site <italic>k</italic> in the absence of directional selection. Values of <italic>ω</italic><sub><italic>k</italic></sub> < 1 and <italic>ω</italic><sub><italic>k</italic></sub> > 1 imply purifying and diversifying selection at site <italic>k</italic>, respectively. The codon substitution rates are then defined as</p><p><disp-formula id="bmcM3"><label>(3)</label><mml:math id="M5" name="1743-422X-10-347-i5" overflow="scroll"><mml:mrow><mml:msubsup><mml:mi>q</mml:mi><mml:mi mathvariant="italic">ij</mml:mi><mml:mi>k</mml:mi></mml:msubsup><mml:mo>∝</mml:mo><mml:mfenced open="{"><mml:mtable columnalign="center"><mml:mtr columnalign="center"><mml:mtd columnalign="center"><mml:msub><mml:mi>μ</mml:mi><mml:mi mathvariant="italic">ij</mml:mi></mml:msub></mml:mtd><mml:mtd columnalign="center"/><mml:mtd columnalign="center"><mml:mtext>if</mml:mtext><mml:mspace width="0.2em"/><mml:mtext>synonymous</mml:mtext></mml:mtd></mml:mtr><mml:mtr columnalign="center"><mml:mtd columnalign="center"><mml:msub><mml:mi>μ</mml:mi><mml:mi mathvariant="italic">ij</mml:mi></mml:msub><mml:msub><mml:mi>ω</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mtd><mml:mtd columnalign="center"/><mml:mtd columnalign="center"><mml:mtext>if</mml:mtext><mml:mspace width="0.2em"/><mml:mtext>nonsynonymous</mml:mtext><mml:mo>,</mml:mo><mml:mspace width="0.2em"/><mml:mi>i</mml:mi><mml:mo>∉</mml:mo><mml:mi>Γ</mml:mi><mml:mo>,</mml:mo><mml:mspace width="0.5em"/><mml:mi>j</mml:mi><mml:mo>∉</mml:mo><mml:mi>Γ</mml:mi></mml:mtd></mml:mtr><mml:mtr columnalign="center"><mml:mtd columnalign="center"><mml:msub><mml:mi>μ</mml:mi><mml:mi mathvariant="italic">ij</mml:mi></mml:msub><mml:msub><mml:mi>ω</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mtd><mml:mtd columnalign="center"><mml:mfrac><mml:mrow><mml:mi>In</mml:mi><mml:mfenced open="(" close=")"><mml:mfrac bevelled="true"><mml:mfrac><mml:msub><mml:mi>γ</mml:mi><mml:mi>k</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>1</mml:mn><mml:mo>-</mml:mo><mml:msub><mml:mi>γ</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mrow></mml:mfenced></mml:mfrac><mml:mfrac><mml:msub><mml:mi>π</mml:mi><mml:mi>Γ</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>1</mml:mn><mml:mo>-</mml:mo><mml:msub><mml:mi>π</mml:mi><mml:mi>Γ</mml:mi></mml:msub></mml:mrow></mml:mfenced></mml:mfrac></mml:mfrac></mml:mfenced></mml:mrow><mml:mrow><mml:mn>1</mml:mn><mml:mo>-</mml:mo><mml:mfenced open="(" close=")"><mml:mfrac bevelled="true"><mml:mfrac><mml:msub><mml:mi>π</mml:mi><mml:mi>Γ</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>1</mml:mn><mml:mo>-</mml:mo><mml:msub><mml:mi>π</mml:mi><mml:mi>Γ</mml:mi></mml:msub></mml:mrow></mml:mfenced></mml:mfrac><mml:mfrac><mml:msub><mml:mi>γ</mml:mi><mml:mi>k</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>1</mml:mn><mml:mo>-</mml:mo><mml:msub><mml:mi>γ</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mrow></mml:mfenced></mml:mfrac></mml:mfrac></mml:mfenced></mml:mrow></mml:mfrac></mml:mtd><mml:mtd columnalign="center"><mml:mtext>if</mml:mtext><mml:mspace width="0.2em"/><mml:mtext>nonsynonymous</mml:mtext><mml:mo>,</mml:mo><mml:mspace width="0.2em"/><mml:mi>i</mml:mi><mml:mo>∉</mml:mo><mml:mi>Γ</mml:mi><mml:mo>,</mml:mo><mml:mspace width="0.5em"/><mml:mi>j</mml:mi><mml:mo>∈</mml:mo><mml:mi>Γ</mml:mi></mml:mtd></mml:mtr><mml:mtr columnalign="center"><mml:mtd columnalign="center"><mml:msub><mml:mi>μ</mml:mi><mml:mi mathvariant="italic">ij</mml:mi></mml:msub><mml:msub><mml:mi>ω</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mtd><mml:mtd columnalign="center"><mml:mfrac><mml:mrow><mml:mi>In</mml:mi><mml:mfenced open="(" close=")"><mml:mfrac bevelled="true"><mml:mfrac><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>1</mml:mn><mml:mo>-</mml:mo><mml:msub><mml:mi>γ</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mrow></mml:mfenced><mml:msub><mml:mi>γ</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mfrac><mml:mfrac><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>1</mml:mn><mml:mo>-</mml:mo><mml:msub><mml:mi>π</mml:mi><mml:mi>Γ</mml:mi></mml:msub></mml:mrow></mml:mfenced><mml:msub><mml:mi>π</mml:mi><mml:mi>Γ</mml:mi></mml:msub></mml:mfrac></mml:mfrac></mml:mfenced></mml:mrow><mml:mrow><mml:mn>1</mml:mn><mml:mo>-</mml:mo><mml:mfenced open="(" close=")"><mml:mfrac bevelled="true"><mml:mfrac><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>1</mml:mn><mml:mo>-</mml:mo><mml:msub><mml:mi>π</mml:mi><mml:mi>Γ</mml:mi></mml:msub></mml:mrow></mml:mfenced><mml:msub><mml:mi>π</mml:mi><mml:mi>Γ</mml:mi></mml:msub></mml:mfrac><mml:mfrac><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>1</mml:mn><mml:mo>-</mml:mo><mml:msub><mml:mi>γ</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mrow></mml:mfenced><mml:msub><mml:mi>γ</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mfrac></mml:mfrac></mml:mfenced></mml:mrow></mml:mfrac></mml:mtd><mml:mtd columnalign="center"><mml:mtext>if</mml:mtext><mml:mspace width="0.2em"/><mml:mtext>nonsynonymous</mml:mtext><mml:mo>,</mml:mo><mml:mspace width="0.2em"/><mml:mi>i</mml:mi><mml:mo>∈</mml:mo><mml:mi>Γ</mml:mi><mml:mo>,</mml:mo><mml:mspace width="0.2em"/><mml:mi>j</mml:mi><mml:mo>∉</mml:mo><mml:mi>Γ</mml:mi></mml:mtd></mml:mtr><mml:mtr columnalign="center"><mml:mtd columnalign="center"/><mml:mtd columnalign="center"><mml:mn>0</mml:mn></mml:mtd><mml:mtd columnalign="center"><mml:mtext>if</mml:mtext><mml:mspace width="0.3em"/><mml:mi>i</mml:mi><mml:mspace width="0.2em"/><mml:mtext>and</mml:mtext><mml:mspace width="0.2em"/><mml:mi>j</mml:mi><mml:mspace width="0.2em"/><mml:mtext>differ</mml:mtext><mml:mspace width="0.2em"/><mml:mtext>at</mml:mtext><mml:mspace width="0.2em"/><mml:mo>></mml:mo><mml:mn>1</mml:mn><mml:mspace width="0.2em"/><mml:mtext>nt</mml:mtext><mml:mspace width="0.2em"/><mml:mtext>position</mml:mtext><mml:mo>.</mml:mo></mml:mtd></mml:mtr></mml:mtable></mml:mfenced></mml:mrow></mml:math></disp-formula></p><p>A similar codon model that distinguishes between diversifying and directional selection was recently considered by Murrell et al. [<xref ref-type="bibr" rid="B53">53</xref>].</p><p>To identify codon sites associated with sensitivity to antibodies, we allowed the equilibrium frequency <italic>γ</italic><sub><italic>k</italic></sub> of the reference amino acid to depend on neutralization titers. More specifically, let <italic>γ</italic><sub><italic>0k</italic></sub> be the equilibrium frequency of the reference amino acid at site <italic>k</italic> among sensitive (high ID<sub>50</sub> titer) viruses and let <italic>γ</italic><sub><italic>1k</italic></sub> denote this frequency for resistant (low ID<sub>50</sub> titer) viruses. We set <italic>γ</italic><sub><italic>k</italic></sub> = <italic>p γ</italic><sub><italic>0k</italic></sub> + (1 – <italic>p</italic>) <italic>γ</italic><sub><italic>1k</italic></sub>, where <italic>p</italic> ∈ [0,1] is a measure of neutralization sensitivity that is monotonically increasing in ID<sub>50</sub> neutralization titer. Since the frequency of the reference residue among sensitive viruses was expected to be at least as large as that among resistant viruses, we constrained <italic>γ</italic><sub><italic>0k</italic></sub> ≥ <italic>γ</italic><sub><italic>1k</italic></sub> where equality implied that a site was not associated with antibody neutralization.</p><p>For the model defined in Equation (3), nonsynonymous substitutions toward the reference amino acid occur at a higher rate if the relative equilibrium frequency of the reference residue <italic>γ</italic><sub><italic>k</italic></sub>/(1-<italic>γ</italic><sub><italic>k</italic></sub>) is larger than its expected value in the absence of directional selection, <italic>π</italic><sub><italic>Γ</italic></sub>/(1-<italic>π</italic><sub><italic>Γ</italic></sub>). Substitutions in the opposite direction are favored if the converse is true. Thus, a neutralization-resistant virus (<italic>p</italic> close to 0) would be more likely to substitute the reference amino acid for another residue at sites where <italic>γ</italic><sub><italic>0k</italic></sub> > <italic>γ</italic><sub><italic>1k</italic></sub>. Similarly, substitutions in favor of the reference residue will be more likely at these sites among neutralization-sensitive viruses (<italic>p</italic> close to 1). This pattern of evolution would reduce the prevalence of the reference residue among resistant viruses and increase its frequency among sensitive viruses, consistent with our expectation of sites associated with antibody neutralization.</p><p>The value of <italic>p</italic> along each branch of the phylogeny was determined by the neutralization titer of the sequence at the tip of that branch. This parameter may be specified by selecting appropriate titer thresholds and setting <italic>p</italic> = 0 if the titer is high (sensitive), <italic>p</italic> = 1 if the titer is low (resistant) and setting <italic>p</italic> equal to the proportion of classified viruses that were sensitive when the titer is an intermediate value or unknown. The method of Gnanakaran et al. [<xref ref-type="bibr" rid="B15">15</xref>] classifies viruses as neutralization sensitive or resistant by applying such a threshold. Here, ID<sub>50</sub> titer was treated as a continuous variable and mapped onto the [0,1] scale to obtain values for <italic>p</italic>. Due to the skewness of the ID<sub>50</sub> titer distribution, we defined <italic>p</italic> = [<italic>t</italic> – min(<italic>t</italic>)]/max[<italic>t</italic> – min(<italic>t</italic>)], where <italic>t</italic> is the natural logarithm of titer. When the titer was unknown, such as at all ancestral nodes, we set <italic>p</italic> equal to its median value among all sequences with observed titers. We also considered imputing the unknown titer values using the method of Felsenstein [<xref ref-type="bibr" rid="B45">45</xref>] (see Discussion).</p><p>Although Equation (3) defines a time-reversible model along each branch of the phylogeny, the process is not time reversible when considered over the entire tree. Consequently, the location of the root node and initial codon frequencies must be specified for valid likelihood-based inference. This specification is trivial when the median titer is used to define <italic>p</italic> at all nodes with unknown titers, since any interior node that is not the immediate ancestor of a leaf node can then act as the root [<xref ref-type="bibr" rid="B42">42</xref>]. However, identification of a specific node as the root is required when the imputed titer values differ between ancestral nodes, such as with Felsenstein’s method of reconstruction [<xref ref-type="bibr" rid="B45">45</xref>].</p><p>Given an inferred topology and F1 × 4 codon equilibrium frequencies, all model parameters, except <italic>γ</italic><sub><italic>0k</italic></sub> and <italic>γ</italic><sub><italic>1k</italic></sub>, were estimated by fitting a GY94 × HKY85 model in HyPhy [<xref ref-type="bibr" rid="B54">54</xref>] with <italic>ω</italic><sub><italic>k</italic></sub> drawn from a three-category general discrete distribution. Fixing the set, <italic>θ</italic>, of these parameters at their estimated values, the probability <italic>p</italic>(<italic>x</italic><sub><italic>k</italic></sub>\| <italic>θ</italic>, <italic>γ</italic><sub><italic>0k</italic></sub>, <italic>γ</italic><sub><italic>1k</italic></sub>) of the codon data <italic>x</italic><sub><italic>k</italic></sub> at site <italic>k</italic> can be computed under the model in Equation (3) for any pair (<italic>γ</italic><sub><italic>0k</italic></sub>, <italic>γ</italic><sub><italic>1k</italic></sub>): 0 ≤ <italic>γ</italic><sub><italic>1k</italic></sub> ≤ <italic>γ</italic><sub><italic>0k</italic></sub> ≤ 1 using Felsenstein’s pruning algorithm. To compute the likelihood of our epitope model, we treated <italic>γ</italic><sub><italic>0k</italic></sub> and <italic>γ</italic><sub><italic>1k</italic></sub> as nuisance parameters with a flat joint distribution and integrated them out of the likelihood function using the adaptive numerical integration routine available in the cubature R package. We also computed the likelihood of a non-epitope model that does not allow for an association between an alignment site and neutralization sensitivities by setting <italic>γ</italic><sub><italic>0k</italic></sub> = <italic>γ</italic><sub><italic>1k</italic></sub> = <italic>γ</italic><sub><italic>k</italic></sub> and integrating <italic>γ</italic><sub><italic>k</italic></sub> out of the likelihood function with <italic>γ</italic><sub><italic>k</italic></sub> ~ <italic>U</italic>(0,1).</p><p>The data-based evidence in support of the epitope model relative to the non-epitope model at site <italic>k</italic> was then assessed using the scaled Bayes factor</p><p><disp-formula><mml:math id="M6" name="1743-422X-10-347-i6" overflow="scroll"><mml:mrow><mml:msub><mml:mi>B</mml:mi><mml:mi>k</mml:mi></mml:msub><mml:mo>=</mml:mo><mml:mn>2</mml:mn><mml:mtext>ln</mml:mtext><mml:mfenced open="(" close=")"><mml:mfrac><mml:mrow><mml:msub><mml:mi>L</mml:mi><mml:mi>E</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>θ</mml:mi><mml:mo>\|</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mrow></mml:mfenced></mml:mrow><mml:mrow><mml:msub><mml:mi>L</mml:mi><mml:mi>N</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>θ</mml:mi><mml:mo>\|</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mrow></mml:mfenced></mml:mrow></mml:mfrac></mml:mfenced><mml:mo>,</mml:mo></mml:mrow></mml:math></disp-formula></p><p>where <italic>L</italic><sub><italic>E</italic></sub>(<italic>θ</italic> \| <italic>x</italic><sub><italic>k</italic></sub>) and <italic>L</italic><sub><italic>N</italic></sub>(<italic>θ</italic> \| <italic>x</italic><sub><italic>k</italic></sub>) are the integrated likelihoods under the epitope and non-epitope models, respectively. Kass and Raftery [<xref ref-type="bibr" rid="B55">55</xref>] suggest that a value of 2 < <italic>B</italic><sub><italic>k</italic></sub> < 6 can be interpreted as positive, but weak evidence against the null model, while <italic>B</italic><sub><italic>k</italic></sub> ≥ 6 indicates strong evidence against the null model. We used the latter criterion to identify sites for experimental validation.</p><p>By performing model comparisons at each of 818 amino acid positions per serum, we anticipated that some sites with <italic>B</italic><sub><italic>k</italic></sub> ≥ 6 would be false positives. To address this issue, we computed the local false discovery rate associated with each of our scaled Bayes factors. The local false discovery rate (LFDR) is defined as the probability of the null (non-epitope) model given the codon data <italic>x</italic><sub><italic>k</italic></sub> at site <italic>k</italic>[<xref ref-type="bibr" rid="B56">56</xref>]. For each site identified as significantly associated with neutralization titer, the LFDR indicates the probability that the site has been classified incorrectly, given the data <italic>x</italic><sub><italic>k</italic></sub> and the prior probability <italic>π</italic><sub>0</sub> of the null model. The LFDR for site <italic>k</italic> may be computed by observing that</p><p><disp-formula><mml:math id="M7" name="1743-422X-10-347-i7" overflow="scroll"><mml:mrow><mml:mfrac><mml:mrow><mml:mn>1</mml:mn><mml:mo>-</mml:mo><mml:msub><mml:mtext>LFDR</mml:mtext><mml:mi>k</mml:mi></mml:msub></mml:mrow><mml:msub><mml:mtext>LFDR</mml:mtext><mml:mi>k</mml:mi></mml:msub></mml:mfrac><mml:mo>=</mml:mo><mml:munder accentunder="true"><mml:munder><mml:mfrac><mml:mrow><mml:msub><mml:mi>p</mml:mi><mml:mi>E</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>θ</mml:mi><mml:mo>\|</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mrow></mml:mfenced></mml:mrow><mml:mrow><mml:msub><mml:mi>p</mml:mi><mml:mi>N</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>θ</mml:mi><mml:mo>\|</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mrow></mml:mfenced></mml:mrow></mml:mfrac><mml:mo stretchy="true">⏟</mml:mo></mml:munder><mml:mrow><mml:mtext>posterior</mml:mtext><mml:mspace width="0.25em"/><mml:mtext>odds</mml:mtext></mml:mrow></mml:munder><mml:mo>=</mml:mo><mml:munder accentunder="true"><mml:munder><mml:mfrac><mml:mrow><mml:msub><mml:mi>L</mml:mi><mml:mi>E</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>θ</mml:mi><mml:mo>\|</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mrow></mml:mfenced></mml:mrow><mml:mrow><mml:msub><mml:mi>L</mml:mi><mml:mi>N</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>θ</mml:mi><mml:mo>\|</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mrow></mml:mfenced></mml:mrow></mml:mfrac><mml:mo stretchy="true">⏟</mml:mo></mml:munder><mml:mrow><mml:mtext>Bayes</mml:mtext><mml:mspace width="0.25em"/><mml:mtext>factor</mml:mtext></mml:mrow></mml:munder><mml:mspace width="0.6em"/><mml:mo>⋅</mml:mo><mml:mspace width="0.3em"/><mml:munder accentunder="true"><mml:munder><mml:mfrac><mml:mrow><mml:mn>1</mml:mn><mml:mo>-</mml:mo><mml:msub><mml:mi>π</mml:mi><mml:mn>0</mml:mn></mml:msub></mml:mrow><mml:msub><mml:mi>π</mml:mi><mml:mn>0</mml:mn></mml:msub></mml:mfrac><mml:mo stretchy="true">⏟</mml:mo></mml:munder><mml:mrow><mml:mtext>prior</mml:mtext><mml:mspace width="0.25em"/><mml:mtext>odds</mml:mtext></mml:mrow></mml:munder></mml:mrow></mml:math></disp-formula></p><p>and rearranging terms to obtain</p><p><disp-formula id="bmcM4"><label>(4)</label><mml:math id="M8" name="1743-422X-10-347-i8" overflow="scroll"><mml:mrow><mml:msub><mml:mtext>LFDR</mml:mtext><mml:mi>k</mml:mi></mml:msub><mml:mo>=</mml:mo><mml:mfrac><mml:mfrac><mml:msub><mml:mi>π</mml:mi><mml:mn>0</mml:mn></mml:msub><mml:mrow><mml:mn>1</mml:mn><mml:mo>-</mml:mo><mml:msub><mml:mi>π</mml:mi><mml:mn>0</mml:mn></mml:msub></mml:mrow></mml:mfrac><mml:mrow><mml:mtext>exp</mml:mtext><mml:mfenced open="(" close=")"><mml:mfrac><mml:msub><mml:mi>B</mml:mi><mml:mi>k</mml:mi></mml:msub><mml:mn>2</mml:mn></mml:mfrac></mml:mfenced><mml:mo>+</mml:mo><mml:mfrac><mml:msub><mml:mi>π</mml:mi><mml:mn>0</mml:mn></mml:msub><mml:mrow><mml:mn>1</mml:mn><mml:mo>-</mml:mo><mml:msub><mml:mi>π</mml:mi><mml:mn>0</mml:mn></mml:msub></mml:mrow></mml:mfrac></mml:mrow></mml:mfrac></mml:mrow></mml:math></disp-formula></p><p>Use of Equation (4) required an estimate of <italic>π</italic><sub>0</sub> which we obtained from the following relation</p><p><disp-formula id="bmcM5"><label>(5)</label><mml:math id="M9" name="1743-422X-10-347-i9" overflow="scroll"><mml:mrow><mml:mi>p</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi>B</mml:mi><mml:mi>k</mml:mi></mml:msub><mml:mo>></mml:mo><mml:mn>0</mml:mn></mml:mrow></mml:mfenced><mml:mo>=</mml:mo><mml:msub><mml:mi>p</mml:mi><mml:mi>N</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi>B</mml:mi><mml:mi>k</mml:mi></mml:msub><mml:mo>></mml:mo><mml:mn>0</mml:mn></mml:mrow></mml:mfenced><mml:msub><mml:mi>π</mml:mi><mml:mn>0</mml:mn></mml:msub><mml:mo>+</mml:mo><mml:msub><mml:mi>p</mml:mi><mml:mi>E</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi>B</mml:mi><mml:mi>k</mml:mi></mml:msub><mml:mo>></mml:mo><mml:mn>0</mml:mn></mml:mrow></mml:mfenced><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>1</mml:mn><mml:mo>-</mml:mo><mml:msub><mml:mi>π</mml:mi><mml:mn>0</mml:mn></mml:msub></mml:mrow></mml:mfenced><mml:mo>,</mml:mo></mml:mrow></mml:math></disp-formula></p><p>where <italic>p</italic><sub><italic>N</italic></sub>(<italic>B</italic><sub><italic>k</italic></sub> > 0) and <italic>p</italic><sub><italic>E</italic></sub>(<italic>B</italic><sub><italic>k</italic></sub> > 0) denoted the probability of a positive scaled Bayes factor if the data was generated under the non-epitope and epitope models, respectively. The probability <italic>p</italic>(<italic>B</italic><sub><italic>k</italic></sub> > 0) can be approximated as the proportion of observed scaled Bayes factors greater than zero for any particular serum sample. To approximate <italic>p</italic><sub><italic>N</italic></sub>(<italic>B</italic><sub><italic>k</italic></sub> > 0), we permuted the assignment of ID<sub>50</sub> titers to the pseudoviruses and recorded the proportion of scaled Bayes factors greater than zero. In this respect, we considered only sites at which the frequency of the most prevalent amino acid was <95%, since invariant sites do not provide evidence for either model. The median proportion over 100 such permutations was used as an estimate of <italic>p</italic><sub><italic>N</italic></sub>(<italic>B</italic><sub><italic>k</italic></sub> > 0). (The mean proportion was not substantively different.) These estimates are reported in Additional file <xref ref-type="supplementary-material" rid="S2">2</xref>: Table S2. For a suitable choice of <italic>p</italic><sub><italic>E</italic></sub>(<italic>B</italic><sub><italic>k</italic></sub> > 0), <italic>π</italic><sub>0</sub> may be computed from Equation (5) and used to obtain the LFDR for a given Bayes factor. We set <italic>p</italic><sub><italic>E</italic></sub>(<italic>B</italic><sub><italic>k</italic></sub> > 0) = 1 to obtain an upper bound on the LFDR. Our reported false discovery rates should therefore be regarded as conservative.</p></sec><sec><title>Conformational epitope prediction</title><p>The method described above can identify individual sites associated with neutralization titer. A set of such predictions in close proximity on the primary sequence provides evidence for a linear B cell epitope. Similarly, a set of predictions clustered on the three-dimensional structure provides evidence for a conformational epitope. For any set of amino acids, <italic>C</italic>, the likelihood that these sites all support the epitope model and that all other sites are generated under the non-epitope model may be computed as</p><p><disp-formula><mml:math id="M10" name="1743-422X-10-347-i10" overflow="scroll"><mml:mrow><mml:mi>L</mml:mi><mml:mfenced open="(" close=")"><mml:mi>C</mml:mi></mml:mfenced><mml:mo>=</mml:mo><mml:mstyle displaystyle="true"><mml:munder><mml:mi mathsize="big">∏</mml:mi><mml:mrow><mml:mi>k</mml:mi><mml:mo>∈</mml:mo><mml:mi>C</mml:mi></mml:mrow></mml:munder><mml:mrow><mml:msub><mml:mi>L</mml:mi><mml:mi>E</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>θ</mml:mi><mml:mo>\|</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mrow></mml:mfenced><mml:mstyle displaystyle="true"><mml:munder><mml:mi mathsize="big">∏</mml:mi><mml:mrow><mml:mi>k</mml:mi><mml:mo>∉</mml:mo><mml:mi>C</mml:mi></mml:mrow></mml:munder><mml:mrow><mml:msub><mml:mi>L</mml:mi><mml:mi>N</mml:mi></mml:msub><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>θ</mml:mi><mml:mo stretchy="true">\|</mml:mo><mml:msub><mml:mi>x</mml:mi><mml:mi>k</mml:mi></mml:msub></mml:mrow></mml:mfenced></mml:mrow></mml:mstyle></mml:mrow></mml:mstyle></mml:mrow></mml:math></disp-formula></p><p>We introduced a simple spherical model of an epitope that can identify sets of sites that support the epitope model and are clustered in three-dimensional space. A Metropolis algorithm was used to explore the posterior density of the location and size of the sphere (see [<xref ref-type="bibr" rid="B57">57</xref>] for a general description of the Metropolis algorithm). The posterior probability that any three-dimensional location lies within the conformational epitope was then computed as the proportion of times that the location falls within the sampled spheres.</p><p>We implemented the Metropolis algorithm in Mathematica 8, using the 2B4C Protein Data Bank structure for gp120 and the likelihoods computed for each site with HyPhy [<xref ref-type="bibr" rid="B54">54</xref>]. We assumed a uniform prior on the three-dimensional coordinate vector defining the centre of the sphere and a gamma prior, <italic>π</italic>(<italic>r</italic>), on the radius <italic>r</italic> of the sphere with a shape parameter of 2 Å and a scale parameter of 4 Å (mean of 8 Å). At the outset, a sphere was generated by randomly selecting an amino acid as the centre of the sphere and drawing a random radius from its prior distribution. A new sphere was proposed by independently drawing each of the three coordinates and the radius from a normal distribution centered on the parameter value of the current sphere with a fixed standard deviation of 3 Å. A move from the current sphere <italic>C</italic> with radius <italic>r</italic> to a new sphere <italic>C’</italic> with a radius <italic>r’</italic> located elsewhere in the protein was accepted with probability</p><p><disp-formula><mml:math id="M11" name="1743-422X-10-347-i11" overflow="scroll"><mml:mrow><mml:mi>α</mml:mi><mml:mo>=</mml:mo><mml:mtext>min</mml:mtext><mml:mfenced open="(" close=")"><mml:mrow><mml:mfrac><mml:mrow><mml:mi>L</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>C</mml:mi><mml:mo>'</mml:mo></mml:mrow></mml:mfenced><mml:mi>π</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>r</mml:mi><mml:mo>'</mml:mo></mml:mrow></mml:mfenced></mml:mrow><mml:mrow><mml:mi>L</mml:mi><mml:mfenced open="(" close=")"><mml:mi>C</mml:mi></mml:mfenced><mml:mi>π</mml:mi><mml:mfenced open="(" close=")"><mml:mi>r</mml:mi></mml:mfenced></mml:mrow></mml:mfrac><mml:mo>,</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:mfenced><mml:mo>.</mml:mo></mml:mrow></mml:math></disp-formula></p><p>A subsequent sphere was then proposed as described above and the process repeated until the Markov chain converged to its stationary distribution. We used a burn-in period of 10 000 iterations, after which spheres were sampled from their posterior distribution at every 100th iteration for 500 000 iterations.</p></sec><sec><title>Identification of coevolving sites</title><p>We used a modified version of the evolutionary network model of Poon et al. [<xref ref-type="bibr" rid="B38">38</xref>] to identify co-evolving sites. The evolutionary events that generated the observed sequence alignment were reconstructed by maximum likelihood under an MG94 × GTR model of codon evolution. For each of the terminal branches, the codon sites that were inferred to have undergone nonsynonymous substitutions were encoded with a 1 and other codon sites were encoded with a 0. This produced an augmented dataset with each terminal branch corresponding to an independent observational vector of the substitution events that occurred along that branch at each codon site and the neutralization titer observed at the tip of the branch. These data were used to construct a Bayesian network under the assumption of log-normal titers. In this context, a Bayesian network is a representation of the joint probability distribution over neutralization titer and the substitution events at all codon positions. The network may be visualized as a graph with nodes representing variables (codon sites and neutralization titer) and edges indicating dependencies between the variables. The evidence in support of each relationship is quantified with a posterior probability. Importantly, by considering the complete distribution of the data, a Bayesian graphical model is able to resolve conditional dependencies, where one variable is related to another only through a mutual relationship with a third variable. The model was implemented in HyPhy with an experimental batch file provided by Dr AFY Poon.</p></sec><sec><title>Phylogenetically-corrected Fisher’s exact tests</title><p>We compared the predictions of our evolutionary model to those of the signature detection method of Gnanakaran et al. [<xref ref-type="bibr" rid="B15">15</xref>]. Briefly, this method uses phylogenetically-corrected Fisher’s exact tests to identify significant associations between amino acid changes along the terminal branches of the viral phylogeny and a binary classification of neutralization phenotypes at the tips of the branches. Three neutralization classification schemes were considered; viruses were classified as neutralization resistant if their ID<sub>50</sub> titers were less than the first quartile, less than the median or less than the third quartile. Ancestral sequences were reconstructed by maximum likelihood using a general time reversible model of nucleotide evolution with site-specific rate heterogeneity. For each possible ancestral amino acid at each site, a contingency table was constructed to test whether the proportion of neutralization sensitive viruses was significantly different between extant viruses that contained the ancestral amino acid at the site versus those that did not. Due to the large number of tests performed, <italic>q</italic>-values were used to judge statistical significance while controlling the false positive rate. All sites with at least one <italic>q</italic> ≤ 1/3 (twice as likely to be a true positive as a false positive) were considered as worthy of further investigation. Only the smallest <italic>q</italic>-value obtained for each site is reported in Additional file <xref ref-type="supplementary-material" rid="S2">2</xref>: Table S3.</p></sec><sec><title>Experimental validation</title><p>A subset of the residues predicted to be associated with neutralization sensitivity were tested for their effect on neutralization sensitivity in one to three sensitive envelope backbones per serum sample using site-directed mutagenesis. Residues of interest were mutated to alanine in all cases except for position 169, which was mutated to glutamic acid as described previously [<xref ref-type="bibr" rid="B34">34</xref>]. Mutagenesis was performed using the Stratagene QuickChange II kit (Stratagene) and confirmed by sequence analysis. Envelope-pseudotyped viruses from mutated and parental envelope clones were obtained by co-transfecting the envelope plasmid with pSG3ΔEnv using Fugene transfection reagent (Roche) into 293 T cells as previously described [<xref ref-type="bibr" rid="B7">7</xref>]. Neutralization was measured by a reduction in luciferase gene expression after single round infection of JC53bl-13 cells with envelope-pseudotyped viruses [<xref ref-type="bibr" rid="B7">7</xref>]. Titers were calculated as the reciprocal plasma dilution (ID<sub>50</sub>) causing 50% reduction of relative light units. The effect of mutations on neutralization sensitivity was calculated as a fold change in neutralization titers of the mutant virus compared to the unmutated parental clone. A model prediction was regarded as validated if a mutation at the corresponding site produced at least a twofold reduction in ID<sub>50</sub> titer in at least one backbone.</p></sec><sec><title>Software availability</title><p>The likelihoods of the evolutionary models were computed on a computer cluster using the parallel processing capabilities of HyPhy and the R Language and Environment for Statistical Computing. We are currently working on an online tool to implement this methodology. All computer code is freely available from the corresponding author.</p></sec><sec><title>Availability of supporting data</title><p>The data set supporting the results of this article is provided in Additional file <xref ref-type="supplementary-material" rid="S3">3</xref>: Dataset S1.</p></sec></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>CW and SSAK conceived, implemented and led the CAPRISA 002 Acute Infection study. LM, CW and CS conceived and designed the experiments. PLM, ESG, MN, MM, CKW, DS, MS, RTB and JM performed the experiments. ML, NKN, BM, BTMK and MK analyzed data. SSAK, HG ad KG contributed reagents and materials. ML, PLM, LM, CW and CS wrote the manuscript. All authors read and approved the final manuscript.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional file 1: Figure S1</title><p>Model predictions for CAP256. Scaled Bayes factors using the (A) autologous CAP256, (B) CAP210 and (C) CAP45 reference sequences. (D) Bayesian evolutionary-network model. (E) Posterior probabilities of a conformational epitope using the three-dimensional model with a ConC reference sequence. The posterior probabilities are shaded as described in the legend for Figure <xref ref-type="fig" rid="F4">4</xref>B. There is some evidence of a conformational epitope in the V3 region (posterior probabilities as high as 0.367). <bold>Figure S2.</bold> Model predictions for CAP8. Scaled Bayes factors using the (A) ConC, (B) Q23 and (C) TRO reference sequences. (D) Bayesian evolutionary-network model. (E) Posterior probabilities of a conformational. <bold>Figure S3.</bold> Model predictions for CAP257. Scaled Bayes factors using the (A) ConC and (B) Q842 reference sequences. (C) Bayesian evolutionary-network model. (D) Posterior probabilities of a conformational epitope using the three-dimensional model with a ConC reference sequence. The posterior probabilities are shaded as described in the legend for Figure <xref ref-type="fig" rid="F4">4</xref>B. There is some evidence of a conformational epitope in the V3 region (posterior probabilities as high as 0.361). <bold>Figure S4.</bold> Model predictions for CAP255. Scaled Bayes factors using the (A) autologous CAP255, (B) TRO and (C) Q23 reference sequences. <bold>Figure S5.</bold> Model predictions for CAP177. Scaled Bayes factors using the (A) ConC, (B) Q23 and (C) TRO reference sequences. (D) Bayesian evolutionary-network model. <bold>Figure S6.</bold> Model predictions for CAP206. Scaled Bayes factors using the (A) ZM197, (B) autologous CAP206, (C) CAP45, (D) Q23, (E) COT6 and (F) TRO reference sequences. (G) Bayesian evolutionary-network model. <bold>Figure S7.</bold> Model predictions for CAP248. Scaled Bayes factors using the (A) ConC, (B) CAP45 and (C) DU156 reference sequences. (D) Bayesian evolutionary-network model. <bold>Figure S8.</bold> Scaled Bayes factors for the CAP256 serum obtained after imputing ancestral titers with the median observed titer and titers reconstructed with Felsenstein’s method.</p></caption><media xlink:href="1743-422X-10-347-S1.pdf"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S2"><caption><title>Additional file 2: Table S1</title><p>Reference sequences. <bold>Table S2.</bold> Estimates used to compute LFDRs. <bold>Table S3.</bold> Significant associations obtained with the method of Gnanakaran et al. [15]. <bold>Table S4.</bold> Sites with scaled Bayes factors ≥ 6 using reconstructed titers.</p></caption><media xlink:href="1743-422X-10-347-S2.doc"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S3"><caption><title>Additional file 3: Dataset S1</title><p>HIV-1 gp160 alignment.</p></caption><media xlink:href="1743-422X-10-347-S3.fas"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S4"><caption><title>Additional file 4: Dataset S2</title><p>Genbank accession numbers and neutralization titers for the virus panel.</p></caption><media xlink:href="1743-422X-10-347-S4.xls"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
Molecular investigation of the 7.2 kb RNA of murine cytomegalovirus	<sec><title>Background</title><p>HCMV encodes a stable 5 kb RNA of unknown function that is conserved across cytomegalovirus species. <italic>In vivo</italic> studies of the MCMV orthologue, a 7.2 kb RNA, demonstrated that viruses that do not express the RNA fail to establish efficient persistent replication in the salivary glands of mice. To gain further insight into the function and properties of this conserved locus, we characterized the MCMV intron in finer detail.</p></sec><sec><title>Methods</title><p>We performed multiple analyses to evaluate transcript expression kinetics, identify transcript termini and promoter elements. The half-lives of intron locus RNAs were quantified by measuring RNA levels following actinomycin D treatment in a qRT-PCR-based assay. We also constructed a series of recombinant viruses to evaluate protein coding potential in the locus and test the role of putative promoter elements. These recombinant viruses were tested in both <italic>in vitro</italic> and <italic>in vivo</italic> assays.</p></sec><sec><title>Results</title><p>We show that the 7.2 kb RNA is expressed with late kinetics during productive infection of mouse fibroblasts. The termini of the precursor RNA that is processed to produce the intron were identified and we demonstrate that the m106 open reading frame, which resides on the spliced mRNA derived from precursor processing, can be translated during infection. Mapping the 5′ end of the primary transcript revealed minimal promoter elements located upstream that contribute to transcript expression. Analysis of recombinant viruses with deletions in the putative promoter elements, however, revealed these elements exert only minor effects on intron expression and viral persistence <italic>in vivo</italic>. Low transcriptional output by the putative promoter element(s) is compensated by the long half-life of the 7.2 kb RNA of approximately 28.8 hours. Detailed analysis of viral spread prior to the establishment of persistence also showed that the intron is not likely required for efficient spread to the salivary gland, but rather enhances persistent replication in this tissue site.</p></sec><sec><title>Conclusions</title><p>This data provides a comprehensive transcriptional analysis of the MCMV 7.2 kb intron locus. Our studies indicate that the 7.2 kb RNA is an extremely long-lived RNA, a feature which is likely to be important in its role promoting viral persistence in the salivary gland.</p></sec>	<contrib contrib-type="author" id="A1"><name><surname>Schwarz</surname><given-names>Toni M</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>toni.schwarz@ucdenver.edu</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Volpe</surname><given-names>Lysa-Anne M</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>lysavolpe@gmail.com</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Abraham</surname><given-names>Christopher G</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>christopher.abraham@ucdenver.edu</email></contrib><contrib contrib-type="author" corresp="yes" id="A4"><name><surname>Kulesza</surname><given-names>Caroline A</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>caroline.kulesza@ucdenver.edu</email></contrib>	Virology Journal	<sec><title>Background</title><p>Human cytomegalovirus (HCMV) is a member of the β-herpesvirus subfamily and a widespread human pathogen [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>]. HCMV infections cause life-threatening illness in the immunocompromised, including bone marrow and solid organ transplant recipients, AIDS patients and patients undergoing cancer chemotherapy. In addition, HCMV infection of the fetus is the leading cause of virally induced birth defects, typically presenting as hearing loss and developmental deficits that impact over 40,000 newborns annually in the US [<xref ref-type="bibr" rid="B3">3</xref>-<xref ref-type="bibr" rid="B5">5</xref>].</p><p>HCMV infection of healthy, immunocompetent individuals elicits a robust host immune response that effectively limits virus replication and pathogenesis. Despite this immune response, HCMV has adapted to long-term infection of humans by balancing persistent replication with clearance by the immune response. HCMV persistently replicates in glandular epithelial tissue and eventually establishes a life-long latent infection of the host. Glandular epithelial cells, such as those in the salivary gland, are key sites of HCMV persistent replication <italic>in vivo,</italic> and contribute significantly to viral transmission. Healthy individuals may secrete virus in saliva, breast milk, and urine for long periods of time following primary infection [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B3">3</xref>]. In order for HCMV to successfully persist, it has evolved to replicate in cell types where the full replication cycle elicits little to no cytopathic effect, such as glandular epithelial cells and some types of endothelial cells [<xref ref-type="bibr" rid="B6">6</xref>-<xref ref-type="bibr" rid="B8">8</xref>]. In addition, the ability to persistently replicate in the host likely depends on reduced immune recognition of virus-infected cells at these specialized sites [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. Few viral determinants that mediate cytomegalovirus persistence have been identified and little is known about the specific molecular functions that facilitate persistence. These include virus-encoded micro-RNAs and a conserved, virus-encoded G-protein-coupled receptor [<xref ref-type="bibr" rid="B11">11</xref>-<xref ref-type="bibr" rid="B13">13</xref>]. In addition, we previously identified a long, non-coding RNA (lncRNA), expressed by all cytomegaloviruses, that we showed to also be an important viral determinant of persistence [<xref ref-type="bibr" rid="B14">14</xref>].</p><p>During lytic replication, HCMV expresses a 5 kb lncRNA of unknown function (also referred to as RNA5.0) [<xref ref-type="bibr" rid="B14">14</xref>-<xref ref-type="bibr" rid="B16">16</xref>]. We showed that this RNA is dispensable for replication in cultured cells and is a stable intron produced by the processing of a large precursor transcript expressed from the genomic region flanked by UL105 and UL111A. Orthologous loci are present in every β-herpesvirus genome examined thusfar, although there is little conservation of sequence or RNA size between different CMV species. Each locus shares some common features, including a high AT sequence content (~60%), and the presence of many homopolymeric stretches of A or T residues. The consensus splice donor sequence that defines the 5′ end of the RNA produced from each locus is also well conserved.</p><p>Cytomegaloviruses exhibit strict species specificity and there is no animal model for HCMV infection. Murine cytomegalovirus (MCMV) infection of the mouse is widely used as an outstanding small-animal model of HCMV infection for several reasons. HCMV and MCMV share similar genomic sequence and organization and undergo similar replication cycles [<xref ref-type="bibr" rid="B17">17</xref>]. Like HCMV, MCMV acutely infects multiple tissues in the mouse, persistently replicates in the salivary gland and establishes a lifelong latent infection of the host [<xref ref-type="bibr" rid="B18">18</xref>]. Therefore, MCMV infection of the mouse is an excellent surrogate for the study of pathogenesis <italic>in vivo</italic>. MCMV expresses a 7.2 kb ortholog of the HCMV 5 kb RNA [<xref ref-type="bibr" rid="B19">19</xref>]. We have shown that recombinant MCMV that does not express the 7.2 kb RNA replicates normally in cultured fibroblasts, but is unable to progress from the acute to the persistent phase of infection in mice. We identified a short hairpin sequence near the 3′ end of the intron that is required for accumulation of the RNA during infection. Persistent replication in the salivary gland of the mouse depends on the accurate processing and stable retention of the intron, since recombinants with a mutation in the splice donor site or deletion of the hairpin sequence fail to transition to the persistent phase of replication <italic>in vivo</italic>. The specific molecular function of this conserved RNA is unknown, but we hypothesize that it mediates processes essential for the virus to evade immune surveillance and/or replicate efficiently at sites of viral persistence.</p><p>To gain further insight into the function and properties of this conserved locus, we characterized the MCMV intron in finer detail. We confirmed that the intron locus RNAs are expressed with late gene kinetics. The 7.2 kb intron has an unusually long half-life, whereas the spliced mRNA that results from processing of the intron is metabolized with kinetics similar to most cellular mRNAs. Investigation of the promoter sequence that controls expression of the intron locus revealed a minimal promoter sequence contributing to the transcriptional output of the locus. Although there is no evidence for translation of the intron itself, we discovered that the spliced mRNA encodes a small protein that co-localizes with the RNA within the nuclei of infected cells. Importantly, we show that the RNA is not required for trafficking of virus to the salivary gland <italic>in vivo,</italic> supporting our hypothesis that the 7.2 kb RNA functions to either evade the host response or maintain viral replication at sites of persistence.</p></sec><sec sec-type="results"><title>Results</title><sec><title>The MCMV 7.2 kb intron locus is transcribed with true late kinetics</title><p>To determine the transcription kinetics of the 7.2 kb intron locus during productive MCMV infection, northern blot analysis was performed on total RNA prepared from cells treated with the translation inhibitor cycloheximide (CHX) or the DNA replication inhibitor phosphonoacetic acid (PAA) prior to MCMV infection. CHX pre-treatment of cells inhibits translation of immediate early (IE) genes blocking subsequent transcription of both early and late classes of viral genes. PAA treatment blocks DNA replication, on which expression of late (L) genes is dependent [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>]. Transcription of the intron (Figure  <xref ref-type="fig" rid="F1">1</xref>A) and the spliced mRNA (Figure  <xref ref-type="fig" rid="F1">1</xref>B) was inhibited by both CHX and PAA treatment. This data indicates that the intron locus RNAs are transcribed with the late class of viral genes during productive infection in fibroblasts.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Expression kinetics of intron locus transcripts.</bold> Total RNA was harvested from infected mouse fibroblasts (MOI = 1.0) at the indicated times and analyzed by northern blot analysis using radio-labeled, antisense RNA probes specific for <bold>(A)</bold> the intron or <bold>(B)</bold> exon 2 of the spliced mRNA. To determine the kinetics of intron locus expression, cells were pre-treated with either PAA or CHX for 1 hour prior to infection. Total RNA was prepared at the indicated times.</p></caption><graphic xlink:href="1743-422X-10-348-1"/></fig><p>In high resolution northern blot analyses specific for the intron RNA, we routinely observed a doublet of bands near 7.2 kb: a major species at approximately 8.0 kb and a minor species migrating faster at 7.2 kb (Figure  <xref ref-type="fig" rid="F1">1</xref>A). These observations were made with multiple intron-specific probes (data not shown). We have been unable to ascertain the basis for this difference in size, although we hypothesize it may be due to effects of lariat secondary structure on RNA migration during electrophoresis resulting in slower migration (data not shown). Likewise, we also observed a doublet of closely migrating bands in northern blot analyses of the spliced RNA product of the locus (Figure  <xref ref-type="fig" rid="F1">1</xref>B). We cannot account for the difference in size based on sequencing of the 5′ and 3′ Rapid Amplification of cDNA Ends (RACE) products (see below). It is possible that we did not capture both species in the RACE reactions but we think it is likely the differences in size reflect variability in 3′ end processing and poly-adenylation that we cannot assess (see Discussion).</p></sec><sec><title>Location of transcriptional start sites and RNA processing signals</title><p>While the splice donor and acceptor sites used in the processing of the 7.2 kb RNA from the primary transcript were previously mapped, the termini of the primary transcript were not determined [<xref ref-type="bibr" rid="B19">19</xref>]. To identify the 5′ and 3′ ends of the primary transcript, we cloned and sequenced PCR products generated by RACE (summarized in Figure  <xref ref-type="fig" rid="F2">2</xref>A-B). To identify the 5′ end of the precursor RNA and capture the predicted intron-exon junctions in the 5′ RACE reaction, we used nested PCR primers specific for the predicted second exon located 3′ of the intron (primers 541 and 542 Table  <xref ref-type="table" rid="T1">1</xref>, Figure  <xref ref-type="fig" rid="F2">2</xref>A). Sequencing of cloned RACE products identified two transcriptional start sites located three nucleotides apart at positions 161,738 and 161,735 in the MCMV genome (sequence coordinates based on the MCMV Smith strain, Genbank accession #NC004065). Sequence alignment of the 5′-RACE products to MCMV genomic sequence also confirmed the splice donor (SD) and splice acceptor (SA) sequences at nucleotide positions 161,622 and 154,366, respectively, as previously annotated [<xref ref-type="bibr" rid="B19">19</xref>]. We also performed primer extension analysis to confirm the transcriptional start sites identified by 5′-RACE. We observed two primer extension products of 101 and 104 nucleotides in length, consistent with the location of the 5′ ends of the spliced RNA as defined by RACE (primer 497 Table  <xref ref-type="table" rid="T1">1</xref>, Figure  <xref ref-type="fig" rid="F2">2</xref>C).</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Mapping of the intron locus. (A)</bold> Diagram of the genomic region encompassing the primary transcript of the 7.2 kb intron, illustrating the transcriptional start sites (TSS) and the splice donor (SD) and splice acceptor sites (SA). The location of primers used for primer extension or RLM-RACE are indicated. <bold>(B)</bold> Diagram of the spliced mRNA and the 5′ and 3′ ends that were identified by RLM-RACE using total RNA harvested from infected mouse fibroblasts at 48 hours post infection (hpi). Putative TATA box directly upstream of the transcriptional start sites is indicated. Also shown are the start and stop codons of the m106 ORF encoded on the second exon of the mRNA as well as the poly-adenylation and cleavage sites used in processing this mRNA. <bold>(C)</bold> Primer extension analysis was performed on total RNA harvested from either mock-infected (M) or MCMV (WT) infected mouse fibroblasts at 48 hpi. Radiolabeled primers were used to validate the 5′RLM-RACE products (primer 497) and confirm the known 7.2 kb splice donor site (primer 50) as a control. <bold>(D)</bold> Northern blot analysis of total RNA (T) harvested from infected mouse fibroblasts and fractionated for either polyadenylated (A+) or non-polyadenylated (A-) RNA. The blot was hybridized with a radiolabeled, antisense RNA probe specific for exon 2.</p></caption><graphic xlink:href="1743-422X-10-348-2"/></fig><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Primer sequences</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="center"/><col align="center"/><col align="left"/></colgroup><thead valign="top"><tr><th align="center"><bold>Target</bold></th><th align="center"><bold>Purpose</bold></th><th align="left"><bold>Sequence 5′ to 3′</bold></th></tr></thead><tbody valign="top"><tr><td rowspan="3" align="center" valign="bottom">7.2 kb Intron (a and a’)<hr/></td><td rowspan="3" align="center" valign="bottom">qPCR<hr/></td><td align="left" valign="bottom">Fwd: GAGTCAGTTCTAACCCATCACG<hr/></td></tr><tr><td align="left" valign="bottom">Rev: AGCTCGAAAGTTGAACGGG<hr/></td></tr><tr><td align="left" valign="bottom">Probe: ACGAACGGGTAAAACGGGTAAGGG<hr/></td></tr><tr><td rowspan="3" align="center" valign="bottom">Exon2 (b and b’)<hr/></td><td rowspan="3" align="center" valign="bottom">qPCR<hr/></td><td align="left" valign="bottom">Fwd: CCACTACCTCTCGATGACAAC<hr/></td></tr><tr><td align="left" valign="bottom">Rev: AGCGAATTCTAGCGTTACCG<hr/></td></tr><tr><td align="left" valign="bottom">Probe: CGGAGCCTGCGACTTGTCTGC<hr/></td></tr><tr><td rowspan="3" align="center" valign="bottom">Spliced mRNA (c and c’)<hr/></td><td rowspan="3" align="center" valign="bottom">qPCR<hr/></td><td align="left" valign="bottom">Fwd: TTATCACACCTGAGCGAACG<hr/></td></tr><tr><td align="left" valign="bottom">Rev: GCAGAGTTCGATGTGTCCG<hr/></td></tr><tr><td align="left" valign="bottom">Probe: AGGATGCGAGATGGCGACGG<hr/></td></tr><tr><td rowspan="3" align="center" valign="bottom">M54<hr/></td><td rowspan="3" align="center" valign="bottom">qPCR<hr/></td><td align="left" valign="bottom">Fwd: AACATATCCCTGCCGATCTTG<hr/></td></tr><tr><td align="left" valign="bottom">Rev: CAACGCTTTCTACGGTTTCAC<hr/></td></tr><tr><td align="left" valign="bottom">Probe: ATGCTCCCGTGTCTCCCCATC<hr/></td></tr><tr><td rowspan="3" align="center" valign="bottom">GAPDH<hr/></td><td rowspan="3" align="center" valign="bottom">qPCR<hr/></td><td align="left" valign="bottom">Fwd: GTGGAGTCATACTGGAACATGTAG<hr/></td></tr><tr><td align="left" valign="bottom">Rev: AATGGTGAAGGTCGGTGTG<hr/></td></tr><tr><td align="left" valign="bottom">Probe: TGCAAATGGCAGCCCTGGTG<hr/></td></tr><tr><td rowspan="3" align="center" valign="bottom">Actin B<hr/></td><td rowspan="3" align="center" valign="bottom">qPCR<hr/></td><td align="left" valign="bottom">Fwd: CTTGATCTTCATGGTGCTAGGAG<hr/></td></tr><tr><td align="left" valign="bottom">Rev: CGTTGACATCCGTAAAGACCT<hr/></td></tr><tr><td align="left" valign="bottom">Probe: ACCATGTACCCAGGCATTGCTGA<hr/></td></tr><tr><td rowspan="3" align="center" valign="bottom">18 srRNA<hr/></td><td rowspan="3" align="center" valign="bottom">qPCR<hr/></td><td align="left" valign="bottom">Fwd: GTTGATTAAGTCCCTGCCCTT<hr/></td></tr><tr><td align="left" valign="bottom">Rev: ATAGTCAAGTTCGACCGTCTTC<hr/></td></tr><tr><td align="left" valign="bottom">Probe: ACCGATTGGATGGTTTAGTGAGGCC<hr/></td></tr><tr><td align="center" valign="bottom">Exon1<hr/></td><td align="center" valign="bottom">Primer extension<hr/></td><td align="left" valign="bottom">497:GGCCTTCGGGACGCCGTCACCTCCGCCGCCGC<hr/></td></tr><tr><td align="center" valign="bottom">Exon2<hr/></td><td align="center" valign="bottom">3′RACE<hr/></td><td align="left" valign="bottom">541: GATCGTTGTCGTCTCTGTCGTGTT<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">3′RACE<hr/></td><td align="left" valign="bottom">542: TGTCATCGAGAGGTAGTGGAGGAT<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">5′RACE<hr/></td><td align="left" valign="bottom">571: ATCCTCCACTACCTCTCGATGACA<hr/></td></tr><tr><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">5′RACE<hr/></td><td align="left" valign="bottom">572: AACACGACAGAGACGACAACGATC<hr/></td></tr><tr><td rowspan="3" align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">Northern<hr/></td><td align="left" valign="bottom">253: GTCGACATGGCGACGGCGAGCCAGCAA<hr/></td></tr><tr><td align="center" valign="bottom">Blot<hr/></td><td align="left" valign="bottom">263: GCGGCCGCGTCTACCACCTCGACCACGATT<hr/></td></tr><tr><td align="center" valign="bottom">Probe<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td rowspan="3" align="center" valign="bottom">7.2 kb Intron<hr/></td><td align="center" valign="bottom">Northern<hr/></td><td align="left" valign="bottom">262: CTCCAATCGGCCTAGGAATCCTGGCTAGGT<hr/></td></tr><tr><td align="center" valign="bottom">Blot<hr/></td><td align="left" valign="bottom">263: AGCAACACGATGCTCTGTGTCGTCGGTCGG<hr/></td></tr><tr><td align="center" valign="bottom">Probe<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td rowspan="2" align="center" valign="bottom">M112/113<hr/></td><td rowspan="2" align="center" valign="bottom">pGL3 Construct<hr/></td><td align="left" valign="bottom">459: ACGAAGGTCTTTTCACCGGT<hr/></td></tr><tr><td align="left" valign="bottom">435: ACCATCTGCTAGGCGGGTCC<hr/></td></tr><tr><td rowspan="2" align="center" valign="bottom">7.2PR1<hr/></td><td rowspan="2" align="center" valign="bottom">pGL3 Construct<hr/></td><td align="left" valign="bottom">28: AGATAGCGCGGCGTCCGTCG<hr/></td></tr><tr><td align="left" valign="bottom">349: CTGAGAGCTCCGGGCCTTCGG<hr/></td></tr><tr><td rowspan="2" align="center">7.2PR2</td><td rowspan="2" align="center">pGL3 Construct</td><td align="left" valign="bottom">133: AAAAGAAAGTCCGTGACCGGGTCG<hr/></td></tr><tr><td align="left">349: CTGAGAGCTCCGGGCCTTCGG</td></tr></tbody></table></table-wrap><p>A single 3′ end was identified at nucleotide position 153,872 by sequencing of 3′-RACE products (Figure  <xref ref-type="fig" rid="F2">2</xref>A). This end is located downstream of a putative polyadenylation signal at position 153,898 (Figure  <xref ref-type="fig" rid="F2">2</xref>B). We also examined the polyadenylation status of the spliced RNA by northern blot analysis of oligo(dT)-selected RNA prepared from MCMV-infected cells. The majority of the spliced mRNA from the intron locus was detected in the poly A + fraction of RNA (Figure  <xref ref-type="fig" rid="F2">2</xref>D). 18S rRNA can only be detected in the non-polyadenylated fraction demonstrating that our fractionation protocol efficiently captured polyadenylated mRNA only (Figure  <xref ref-type="fig" rid="F2">2</xref>D lanes A<sup>+</sup> and A<sup>-</sup>). Taken together, our data suggest that a large precursor RNA is transcribed from the intron locus at late times of infection and processed to yield a single 7.2 kb stable intron and a spliced poly-adenylated mRNA consisting of two exons.</p></sec><sec><title>The m106 open reading frame on the spliced mRNA is translated during infection</title><p>The second exon of the spliced mRNA processed from the primary transcript that produces the 7.2 kb RNA spans a previously annotated open reading frame (ORF) called m106 [<xref ref-type="bibr" rid="B17">17</xref>]. Positional orthologues of m106 have been identified in all cytomegaloviruses, including the UL106 ORF of HCMV, yet there is little sequence homology among the group [<xref ref-type="bibr" rid="B20">20</xref>-<xref ref-type="bibr" rid="B23">23</xref>]. In general, UL106 orthologues score poorly with algorithms designed to predict the potential of an ORF to encode a protein [<xref ref-type="bibr" rid="B24">24</xref>,<xref ref-type="bibr" rid="B25">25</xref>]. To determine if m106 protein is translated during MCMV replication we constructed two recombinant viruses engineered to express m106 as a GFP fusion at the carboxy-terminus (Figure  <xref ref-type="fig" rid="F3">3</xref>A). The first recombinant virus expresses the m106-GFP fusion from the wild-type MCMV genome (MCMV:m106GFP). The second recombinant virus expressing the m106-GFP fusion also contains a five-nucleotide substitution at the splice donor site that defines the 5′ end of the 7.2 kb RNA (MCMV<italic>del</italic>SD:m106GFP) [<xref ref-type="bibr" rid="B19">19</xref>]. This substitution prevents processing of the intron from the primary transcript and we predicted that it would also prevent translation of m106-GFP protein. Both recombinant viruses replicate with wild-type kinetics in multi-step growth analysis in mouse fibroblasts (Figure  <xref ref-type="fig" rid="F3">3</xref>B). Immunoblotting for the m106-GFP fusion protein with antibody specific for GFP only detected protein expression during MCMV:m106GFP infection and not MCMV<italic>del</italic>SD:m106GFP infection, indicating that splicing of the mRNA is necessary for translation of m106 (Figure  <xref ref-type="fig" rid="F3">3</xref>C). Furthermore, this data indicates that cryptic transcriptional initiation does not appear to occur within the unspliced transcript produced by MCMV<italic>del</italic>SD:m106GFP at any significant level.</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>The m106 open reading frame is translated during MCMV infection. (A)</bold> Diagram of the GFP cassette insertion within the MCMV Smith BAC clone. <bold>(B)</bold> Analysis of recombinant virus replication in mouse fibroblasts. Cells were infected at a multiplicity of 0.05 PFU/cell and viral supernatants were collected daily and titrated. Graph represents two biological replicates. <bold>(C)</bold> Western blot analysis of protein lysates prepared from mock-, MCMV:m106GFP (WT-GFP), or MCMV<italic>del</italic>SD:m106GFP (SDM-GFP) infected cells at 48 h p.i. m106-GFP expression was detected using a polyclonal anti-GFP antibody. Asterisk denotes m106-GFP. <bold>(D)</bold> Mouse fibroblasts were infected with MCMV:m106-GFP (MOI = 0.05). Cells were fixed at 72 h p.i. and m106-GFP protein expression and 7.2 kb intron production was detected by by combined immunofluorescence assay and FISH.</p></caption><graphic xlink:href="1743-422X-10-348-3"/></fig></sec><sec><title>m106 and the 7.2 kb intron localize to the nucleus of infected fibroblasts</title><p>Although we previously demonstrated in fractionation studies that the HCMV 5 kb intron localizes to the nuclear compartment of infected cells, the specific sub-nuclear localization of the RNA was not examined [<xref ref-type="bibr" rid="B14">14</xref>]. Fluorescent in situ hybridization (FISH) was used to visualize the 7.2 kb intron in infected mouse fibroblasts. The FISH staining revealed an even, granular distribution of the 7.2 kb intron throughout the nuclear compartment of infected fibroblasts (Figure  <xref ref-type="fig" rid="F3">3</xref>D). Co-staining for the 7.2 kb RNA and m106-GFP revealed that the RNA and m106 protein are found co-localizing in the nucleus late during infection (Figure  <xref ref-type="fig" rid="F3">3</xref>D). We also observed some m106-GFP protein was localized to the cytoplasm of infected cells.</p></sec><sec><title>The MCMV 7.2 kb intron is highly stable</title><p>The 7.2 kb RNA accumulates to high levels during infection as detected by northern blot analysis, suggesting it is unusually stable for an intron [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B26">26</xref>,<xref ref-type="bibr" rid="B27">27</xref>]. To quantify transcript stability, we measured RNA decay rates of intron-locus transcripts during MCMV infection. RNA half-lives were quantified by measuring RNA abundance by quantitative RT-PCR at several time points after treatment of infected cells with Actinomycin D. This compound inhibits RNA Polymerase II by intercalating between GC residues thereby blocking processivity of the enzyme. Actinomycin D treatment effectively arrests transcription of RNA pol II-dependent RNAs and allows us to measure relative decay rates. Using this strategy, we calculated the half-life of the 7.2 kb intron to be ~28.8 hours (Figure  <xref ref-type="fig" rid="F4">4</xref>B). In general, the half-life of low-stability RNAs is typically less than 2 hours whereas long-lived RNAs with high stability possess a half-life greater than 12 hours [<xref ref-type="bibr" rid="B28">28</xref>]. The half-life of the spliced mRNA derived from processing of the intron was measured using two different primer probe sets: b/b’ targets the second exon and c/c’ spans the splice junction (Table  <xref ref-type="table" rid="T1">1</xref>, Figure  <xref ref-type="fig" rid="F4">4</xref>A). We measured a half-life of ~6.8 hours using the primer-probe set targeting the second exon (Figure  <xref ref-type="fig" rid="F4">4</xref>C), while a half-life of ~7.8 hours was measured using the primer-probe set that spans the splice junction (Figure  <xref ref-type="fig" rid="F4">4</xref>D). This difference may be accounted for by differences in primer-probe efficiency, processing of the primary transcript, or location of the primer-probe sets in relation to protective secondary structures within the mRNA from degradation machinery. Both half-lives, however, are consistent with the average half-life for a protein coding RNA. As a control, we determined the half-life of GAPDH mRNA to be ~15 hours, similar to published estimates (Figure  <xref ref-type="fig" rid="F4">4</xref>E) [<xref ref-type="bibr" rid="B28">28</xref>]. Our data demonstrates that the MCMV 7.2 kb intron is, in fact, unusually stable for an intron RNA.</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>Half-life analysis of the intron locus transcripts.</bold> Mouse fibroblasts were infected with MCMV (MOI = 1.0). At 30 hours post infection, cells were treated with 4 ug/ml of Actinomycin D. Total RNA was harvested over the indicated time course. Intron, mRNA, and 18 srRNA transcript levels were quantified by qRT-PCR. Intron and spliced mRNA transcripts were normalized to 18 srRNA. The relative quantitative values at time zero hours were adjusted to 100% and transcript remaining was compared relative to time zero. The fitted curve was modeled by one phase decay using a non-linear regression analysis on four biological replicates for each time point. The half-life (t1/2) shown for each transcript is the best-fit value. Bars represent the mean and error bars represent the standard error of the mean (SEM). <bold>(A)</bold> Schematic representation of primer probe sets used for qRT-PCR analysis. SD = splice donor sequence, SA = splice acceptor sequence. Half-life decay curves for the <bold>(B)</bold> 7.2 kb intron using primer probe set a/a’, <bold>(C)</bold> the second exon of the mRNA (b/b’), <bold>(D)</bold> the spliced mRNA using primer probe set c/c’, and <bold>(E)</bold> GAPDH.</p></caption><graphic xlink:href="1743-422X-10-348-4"/></fig></sec><sec><title>Analysis of minimal promoter sequences</title><p>Little data is available regarding the sequence elements driving late transcriptional units of cytomegaloviruses. Therefore, the DNA sequence within the vicinity of the TSSs of the intron locus was examined for transcriptional regulatory sequences. A putative TATA box element was identified 24 nucleotides upstream of the transcriptional start site, as well as an additional TATA box located 127 nucleotides upstream (Figure  <xref ref-type="fig" rid="F2">2</xref>B, Figure  <xref ref-type="fig" rid="F5">5</xref>A). In order to examine the transcriptional activity of these putative minimal promoter elements and the nucleotide sequence surrounding them, DNA sequences from the region between the intron splice donor site and the M112/113 locus were cloned into a luciferase reporter plasmid to quantify transcriptional promoter activity (7.2PR1 and 7.2PR2). The M112/113 locus is located upstream of the intron locus and on the opposite strand. The transcriptional start site and promoter elements controlling the M112/113 locus have been defined [<xref ref-type="bibr" rid="B29">29</xref>,<xref ref-type="bibr" rid="B30">30</xref>]. Therefore, the M112/113 early-late promoter was cloned into the reporter vector in both sense and antisense orientations (M112PR and aM112PR). To serve as a control, only the M112/113 promoter sequence in the sense orientation induced luciferase activity similar to the control (Figure  <xref ref-type="fig" rid="F5">5</xref>A). Surprisingly, no appreciable induction of luciferase activity was observed for 7.2PR1 or 7.2PR2 relative to the promoterless control vector (Figure  <xref ref-type="fig" rid="F5">5</xref>B). We also found that MCMV co-infection of reporter-tranfected cells did not induce luciferase activity from the minimal 7.2PR1 promoter construct (Figure  <xref ref-type="fig" rid="F5">5</xref>C).</p><fig id="F5" position="float"><label>Figure 5</label><caption><p><bold>Analysis of transcriptional activity of putative intron locus promoter elements. (A)</bold> Diagram of genomic location of the viral sequences cloned into pGL3-reporter plasmids. <bold>(B)</bold> pGL3-reporter constructs were co-transfected into mouse fibroblasts and luciferase induction was assayed 24 h p.i. following either mock co-infection, <bold>(C)</bold> UV inactivated co-infection, or MCMV co-infection (MOI = 0.05).</p></caption><graphic xlink:href="1743-422X-10-348-5"/></fig><p>To examine the contribution of the putative minimal promoter sequences to transcription of the intron locus RNAs in the context of virus infection, three recombinant viruses were made with deletions in this region (Figure  <xref ref-type="fig" rid="F6">6</xref>A). We constructed recombinants with (1) a 20 bp deletion spanning the TSS-proximal TATA box, (2) a 100 bp deletion including the proximal TATA box and upstream sequence, and (3) a 135 bp deletion spanning both the proximal and distal TATA boxes in the putative minimal promoter sequence. All three recombinant viruses replicated similar to wild-type MCMV in multi-step growth analysis (Figure  <xref ref-type="fig" rid="F6">6</xref>B). We quantified RNA expression in cells infected with our panel of recombinant viruses, including previously characterized recombinants with mutations that result in a failure to express significant levels of the intron (MCMV<italic>del</italic>HP and MCMV<italic>del</italic>SD) [<xref ref-type="bibr" rid="B19">19</xref>]. The MCMV<italic>del</italic>HP contains a 28 bp deletion spanning a predicted stem loop structure at the 3′end of the intron that is hypothesized to confer stability. Without this stem loop structure, processing of the primary transcript still occurs since the mRNA is detected by qRT-PCR and northern blot analysis, but accumulation of the intron is significantly reduced (Figure  <xref ref-type="fig" rid="F6">6</xref>C). While the splice donor site mutation impacts processing of the precursor transcript and is not expected to affect transcriptional output of the promoter, we predicted that MCMV<italic>del</italic>135 would reduce overall transcript production from the locus. Measured reductions in levels of the intron were only significant for cells infected with the MCMV<italic>del</italic>135 mutant in addition to the MCMV<italic>del</italic>SD and MCMV<italic>del</italic>HP recombinant viruses (Figure  <xref ref-type="fig" rid="F6">6</xref>C). In cells infected with MCMV<italic>del</italic>135, intron abundance was reduced approximately 5-fold while levels of the mature mRNA transcript were reduced by 10-fold (Figure  <xref ref-type="fig" rid="F6">6</xref>C,D, and E). Despite different predicted consequences of the mutations, the reduction of the spliced mRNA transcript abundance is similar between the MCMV<italic>del</italic>SD and the MCMV<italic>del</italic>135.</p><fig id="F6" position="float"><label>Figure 6</label><caption><p><bold>Deletion mutations in putative viral promoter elements reveal reduction in transcriptional output in cell culture and decreased recovery of infectious virus </bold><bold><italic>in vivo</italic></bold><bold>. (A)</bold> Diagram of the genomic location of intron locus promoter region deletions. <bold>(B)</bold> Multi-step growth analysis of replication of all three recombinant viruses in comparison to WT MCMV. Mouse fibroblasts were infected (MOI = 0.05) and culture supernatants were collected every 24 hours and titrated by plaque assay. Graph represents three biological replicates. <bold>(C</bold>, <bold>D</bold>, <bold>and </bold><bold>E)</bold> Quantification of intron locus RNAs in cells infected with promoter mutant viruses. Mouse fibroblasts were infected (MOI = 1.0) and total RNA was harvested 48 hours post infection. Intron locus transcript levels are quantified relative to WT MCMV transcript levels by qRT-PCR. The primer probes sets used are the same as shown in Figure  <xref ref-type="fig" rid="F4">4</xref>A. Graphs represent three biological replicates. <bold>(F </bold><bold>and </bold><bold>G)</bold> Three-month-old female BALB/c mice were infected with an i.p. dose of 5 x 10<sup>5</sup> PFU with the indicated viruses. At 14 days post infection, animals were euthanized and tissues collected for analysis of infectious virus yield <bold>(F)</bold> and viral genome number <bold>(G)</bold>. <bold>(F)</bold> Salivary gland homogenates were analyzed by plaque assay on mouse fibroblasts. <italic>p</italic> values represent the Student’s T Test result between WT MCMV infected cells or mice and cells or mice infected with the given recombinant viral mutant for each transcript analyzed (<italic>p</italic> < 0.05 <italic>p</italic> < 0.01 <italic>p</italic> < 0.001 *<italic>p</italic> < 0.0001). WT MCMV = WT; MCMV<italic>del</italic>20 = 20DEL; MCMV<italic>del</italic>100 = 100DEL; MCMV<italic>del135 =</italic> 135DEL<italic>;</italic> MCMV<italic>del</italic>HP = HPM; MCMV<italic>del</italic>SD = SDM.</p></caption><graphic xlink:href="1743-422X-10-348-6"/></fig><p>To determine if reductions in intron and mRNA expression have an effect on the establishment of persistence <italic>in vivo</italic>, mice were inoculated with a subset of our panel of recombinant viruses and viral yields measured in the salivary gland at 14 days post-infection (Figure  <xref ref-type="fig" rid="F6">6</xref>F). We observed a slight reduction in viral yield in the salivary glands of mice infected with MCMV<italic>del</italic>20 and a ten-fold reduction of viral yield in mice infected with MCMV<italic>del</italic>135. Viral genome quantification corroborated the measure of infectious virus within the salivary gland (Figure  <xref ref-type="fig" rid="F6">6</xref>G). However, despite 5–10 fold reductions of intron and mRNA production, neither promoter deletion mutant fully attenuated persistent replication to the levels observed in mice infected with MCMV<italic>del</italic>SD.</p></sec><sec><title>Intron locus products do not influence dissemination to the salivary gland</title><p>Although we have shown that recombinant viruses that fail to express the intron replicate poorly in the salivary glands, it was unclear if this attenuation was caused by a lack of dissemination to or a failure to replicate within the salivary gland [<xref ref-type="bibr" rid="B19">19</xref>]. To examine whether the intron is required for dissemination to the salivary gland, mice were inoculated with wild-type MCMV or MCMV<italic>del</italic>SD and viral yields in various tissues were measured at 4, 6, 8, and 14 days post infection (Figure  <xref ref-type="fig" rid="F7">7</xref>). MCMV<italic>del</italic>SD replicated to similar levels as wild-type MCMV until 6 days post infection in all tissues examined. At 8 days post infection, levels of the splice donor mutant virus were significantly reduced in the liver, kidney, and spleen but were unchanged in comparison to wild-type MCMV within the salivary gland and lung. By 14 days post infection, replication of the splice donor mutant virus was severely attenuated in all organs assayed and infectious virus was below the limit of detection by plaque assay. Interestingly, the relative number of MCMV<italic>del</italic>SD genomes was reduced 100-fold in salivary glands at 14 days post infection relative to wild-type MCMV, suggesting that the virus was effectively cleared from this tissue (Figure  <xref ref-type="fig" rid="F8">8</xref>). This data indicates that the intron does not influence viral dissemination to the salivary gland over a time course of infection but may function to promote viral persistence in the glandular epithelial tissue.</p><fig id="F7" position="float"><label>Figure 7</label><caption><p><bold>Intron locus RNAs are dispensable for virus dissemination to the salivary glands during acute infection.</bold> Three-month-old female BALB/c mice were inoculated i.p. with 5x10<sup>5</sup> PFU of WT MCMV or MCMV<italic>del</italic>SD. At the indicated days post infection, organs were harvested from three mice per infection group to quantify infectious virus by plaque assay. Bars represent the mean and error bars represent the standard error of the mean (SEM). The dashed line indicates the limit of detection.</p></caption><graphic xlink:href="1743-422X-10-348-7"/></fig><fig id="F8" position="float"><label>Figure 8</label><caption><p><bold>Detection of viral genomes is significantly reduced in salivary glands during persistence.</bold> Three-month-old female BALB/c mice were inoculated i.p. with 5x10<sup>5</sup> PFU of WT MCMV (WT) or MCMV<italic>del</italic>SD (SDM). DNA was harvested from the indicated organs of three mice per infection group at 14 days post infection and viral genomes were quantified by qPCR using a primer probe set specific for the M54 MCMV gene and normalized to the beta-actin cellular gene. Bars represent the mean and error bars represent the standard error of the mean (SEM). <italic>p</italic> values represent the Student’s T Test result between WT MCMV and MCMV<italic>del</italic>SD infected mice (<italic>p</italic> < 0.05 <italic>p</italic> < 0.01 <italic>p</italic> < 0.001 ***<italic>p</italic> < 0.0001).</p></caption><graphic xlink:href="1743-422X-10-348-8"/></fig></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>Non-coding RNAs are known to be expressed by nearly all herpesviruses that infect humans, yet the function of these RNAs in viral replication and pathogenesis has been elusive [<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B31">31</xref>-<xref ref-type="bibr" rid="B34">34</xref>]. In our current study, we report further characterization of a lncRNA expressed by MCMV, the 7.2 kb RNA. This RNA is the ortholog of the 5.0 kb RNA of HCMV and we have previously shown that it is a virulence factor that promotes viral persistence <italic>in vivo</italic>. We showed that during productive infection in fibroblasts, the intron locus RNAs are transcribed with true late gene kinetics. These RNAs are derived from a common precursor RNA and are produced as the result of splicing of two exons that flank the intron to create a smaller mRNA and the 7.2 kb intron. We observed a doublet of bands with probes specific for the intron in northern blot analysis at 8.0 and 7.2 kb. While it is formally possible that these represent two different species of intron, we did not detect evidence of alternative splicing of a larger intron from the precursor RNA by sequencing of 5′-RACE products. We think it is likely that the intron remains in the form of a branched lariat after processing and therefore migrates more slowly during electrophoresis. Similar observations have been made for the LAT of HSV-1 [<xref ref-type="bibr" rid="B35">35</xref>]. We also detected a doublet of bands corresponding to the spliced mRNA in northern blot analysis. Again, sequencing of 5′-RACE products did not reveal any splicing variations that could account for the size differences. The polyadenylation chain lengths could differ for the individual spliced mRNA molecules representing the doublet observed for the mRNA. Identification of the transcriptional start sites rules out the possibility that a cluster of miRNAs mapped near to the 7.2 kb intron splice donor site originate from the same primary transcript [<xref ref-type="bibr" rid="B36">36</xref>]. It remains unknown, however, what functional relationship the miRNAs may have with the MCMV 7.2 kb intron locus, if any, during virus replication and pathogenesis.</p><p>Identification of the transcriptional start sites of the primary transcript expressed from the intron locus led us to identify putative transcriptional control elements that contribute to regulating transcription of the intron precursor. We identified two TATA box elements located within 135 bp of the TSS. However, using a luciferase reporter system, this region did not confer significant transcriptional activity above background. Transcriptional control elements driving expression of the adjacent M112/113 locus have been characterized and functioned in our assay as expected [<xref ref-type="bibr" rid="B30">30</xref>]. We were able to rule out the possibility that the M112/113 promoter controls transcription of the intron locus since function of this promoter is orientation dependent. Intron RNA and mRNA expression levels in cells infected with recombinant virus bearing a deletion spanning the distal and proximal TATA boxes and transcriptional start site were reduced compared to wild-type MCMV. However, the reduction in transcript levels was not nearly as robust as that observed in recombinants bearing mutations that affect intron stability or processing. We hypothesize that most of the variations we observed in the differing effects of promoter mutations on relative intron and mRNA levels can be accounted for by their different half-lives: since the intron is unusually stable, it appears to be less affected by the promoter deletions whereas the mRNA has a shorter half-life and the relative levels of RNA are more sensitive to modest reductions in transcriptional output associated with the promoter deletions.</p><p>Our studies did not identify sequence elements that robustly contribute to transcriptional control of this late transcriptional unit. As a late gene, amplification of genome copy number by DNA replication is thought to be critical for robust L transcription [<xref ref-type="bibr" rid="B2">2</xref>]. More recently, it has been demonstrated that viral replication and L gene expression also relies on a distinct set of five genes conserved across beta and gamma herpesviruses: UL79, UL87, UL91, UL92, and UL95 [<xref ref-type="bibr" rid="B37">37</xref>-<xref ref-type="bibr" rid="B39">39</xref>]. It is hypothesized that an RNA polymerase II transcriptional complex including one or more of these gene products is assembled to drive transcription of L genes. MCMV homologs of HCMV UL87, UL91, UL92, and UL95 have been annotated, but not tested for transcriptional activating functions [<xref ref-type="bibr" rid="B38">38</xref>-<xref ref-type="bibr" rid="B40">40</xref>]. M79, the MCMV homolog of HCMV UL79, has been shown to regulate L gene expression, although it does not appear to promote transcription of the intron locus [<xref ref-type="bibr" rid="B41">41</xref>]. It is also possible that we did not include the sequences responsible for binding of L gene transactivators within our reporter assay. Clearly, transcriptional regulation of L genes remains largely unexplored and it is likely that this process is significantly different from the activation of immediate early and early transcriptional units.</p><p>Post-transcriptional regulation, metabolism and function of lncRNAs is poorly understood in general. We demonstrated that the 7.2 kb intron accumulates as a consequence of a slow decay rate. Different properties contribute to lncRNA stability include GC content, the presence of specific decay elements, and if they are intronic. Given its intronic and AT-rich nature, the MCMV 7.2 kb lncRNA would be predicted to be highly unstable since introns are rapidly degraded on formation and AT-rich sequences do not form strong secondary structures that might protect the RNA from degradation. A stem-loop structure located near the 3′ end of the intron between the polypyrimidine track and putative branch point was identified using the structural prediction software mFold. Deletion of the stem loop does not impact processing of the precursor transcript, but does prevent accumulation of the intron during infection [<xref ref-type="bibr" rid="B19">19</xref>]. We hypothesize that the intron remains in the form of a lariat, similar to the Latency Associated Transcript (LAT) of HSV-1, thereby protecting it from degradation [<xref ref-type="bibr" rid="B42">42</xref>,<xref ref-type="bibr" rid="B43">43</xref>]. While the sequence and structural determinants of stability remain largely unknown for the 7.2 kb intron, its long half-life accounts for its accumulation and could be a key component of its functionality.</p><p>The spliced mRNA produced by processing of the 7.2 kb RNA spans the m106 ORF. Using an epitope-tagging strategy, we showed that this ORF could be translated during MCMV infection. The GFP-tagged protein co-localized to the nucleus of infected cells with the 7.2 kb RNA. This may reflect a related function of the intron and the m106 protein. Recombinant viruses that specifically disrupt m106 expression without impacting intron production will be useful reagents to probe the function for this viral protein. The m106 protein and its orthologues encoded by other CMVs, including UL106 of HCMV, have some unusual properties. Despite not sharing significant sequence homology, all UL106 orthologues are small (<150 amino acids), highly basic, arginine-rich peptides. It is unknown if other UL106 orthologues are expressed during infection, but given the conservation of the genomic organization of the intron locus among CMVs, it is a distinct possibility to be explored.</p><p>Production of the 7.2 kb RNA is required for the establishment of persistence in the salivary glands of mice. By analysis of multiple time points between the acute and persistent phases of infection in mice, we showed that recombinant virus lacking the intron appears to disseminate to the salivary gland as efficiently as wild-type MCMV. However, it is unable to maintain a highly productive replication program in the salivary glands as observed at 14 days post infection. In addition, we did not detect infectious MCMV<italic>del</italic>SD in any organs at 14 dpi and genome copy number of the mutant virus was substantially reduced in liver and kidney. It is possible that the mechanisms that prevent establishment of intron-mutant virus persistence in the salivary gland may also promote accelerated clearance of that virus from liver and kidney. At this time, the adaptive immune response acts to limit viral replication and it is possible that the 7.2 kb intron is involved in modulating immune surveillance in some way. Some cellular lncRNAs are involved in transcriptional regulatory processes, therefore, a possible mechanism for evading the immune response could be to regulate cellular or viral genes that are involved in this host pathogen relationship [<xref ref-type="bibr" rid="B44">44</xref>,<xref ref-type="bibr" rid="B45">45</xref>].</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>This current study has provided a detailed transcriptional and functional analysis of the MCMV 7.2 kb RNA locus. Mapping the topography of the 7.2 kb RNA locus allows us to understand the genomic elements that not only comprise the locus but also control its expression. The unusual stability of the 7.2 kb RNA compensates for the weak transcriptional output by the putative promoter region observed and may also provide insight towards molecular function within the nucleus of infected cells. Uncovering the stability determinants of the 7.2 kb RNA will therefore allow us to understand the mechanisms that promote its retention within infected cells. Although a function has yet to be determined to the 7.2 kb RNA, this current analysis has provided the framework for investigating its function during viral persistence.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Cell culture</title><p>10.1-mouse embryonic fibroblasts were propagated in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% newborn calf serum, 100 U/ml penicillin, and 100 ug/ml streptomycin. Cells were maintained at 37°C with 5% CO<sub>2</sub>.</p></sec><sec><title>Viruses</title><p>The BAC clone of the wild type Smith MCMV strain, pSM3fr, was used as the parent strain in this study [<xref ref-type="bibr" rid="B46">46</xref>,<xref ref-type="bibr" rid="B47">47</xref>]. Recombinant viruses were generated by linear recombination using either the seamless, red-mediated recombination in the DH10B <italic>Escherichia coli</italic> strain GS1783 method or by using the FLP-recombinase method as previously described [<xref ref-type="bibr" rid="B19">19</xref>]. Primer sequences used to generate the recombinant viruses are indicated in Table  <xref ref-type="table" rid="T2">2</xref>. Recombinant BAC DNA was isolated and electroporated into 10.1 fibroblasts to produce viral stocks. Multi-step growth analysis was performed by infecting 10.1 fibroblasts at a multiplicity of infection (MOI) of 0.05, collecting supernatant every 24 hours for five days starting at time zero, and titrating the supernatant by plaque assay or TCID<sub>50</sub> assay on 10.1 fibroblasts.</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Primers used for recombinant virus production</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Target</bold></th><th align="left"><bold>Sequence 5′ to 3′</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">MCMV<italic>del</italic>20<hr/></td><td align="left" valign="bottom">GGAGTGTAGGTATTCACCGTCAGACGCAACCTGACGCATCCCGGCTAGAATCGATTTATTCAACAAAGCCACG<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">CACCTGAGCCTGCTCGGCCGTTCGCTCAGGTGTGATAATGCACCTTTCAGCGCGTATATCTGGCCCGTACATCG<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">TATTCACCGTCAGACGCAACCTGACGCATCCCGGCTAGAACTGAAAGGTGCATTATCACACCTGAGCGAACGGCCGAGCA<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">TGCTCGGCCGTTCGCTCAGGTGTGATAATGCACCTTTCAGTTCTAGCCGGGATGCGTCAGGTTGCGTCTGACGGTGAATA<hr/></td></tr><tr><td align="left" valign="bottom">MCMV<italic>del</italic> 100<hr/></td><td align="left" valign="bottom">GATCACGCTACCACCGTGTGTCTCCGTACTCCGCTATTATACTTTGCGGCTCGATTTATTCAACAAAGCCACG<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">CGCTACCACCGTGTGTCTCCGTACTCCGCTATTATACTTTGCGGCCTGAAAGGTGCATTATCACACCTGAGCGAACGGCCGAGCAGGCTC<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">GAGCCTGCTCGGCGTTCGCTCAGGTGTGATAATGCACCTTTCAGGCCGCAAAGTATAATAGCGGAGTACGGAGACACACGGTGGTAGCG<hr/></td></tr><tr><td align="left" valign="bottom">MCMV<italic>del</italic> 135<hr/></td><td align="left" valign="bottom">CGGCACGGGGAAATAAAATGATCACGCTACCACCGTGTGTCTCCGTACTCTCGATTTATTCAACAAAGCCACG<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">GATGCGTCCGCCGCCTCACCTGAGCCTGCTCGGCCGTTCGCTCAGGTGTGCGCGTATATCTGGCCCGTACATCG<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">AAATAAAATGATCACGCTACCACCGTGTGTCTCCGTACTCCACACCTGAGCGAACGGCCGAGCAGGCTCAGGTGAGGCGG<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">CCGCCTCACCTGAGCCTGCTCGGCCGTTCGCTCAGGTGTGGAGTACGGAGACACACGGTGGTAGCGTGATCATTTTATTT<hr/></td></tr><tr><td align="left" valign="bottom">M106-GFP<hr/></td><td align="left" valign="bottom">TCCACCAACACGATCCCCGAGATACCCAGAATCGTGGTCGAGGTGGTAGACGCCGGAAGAAGATGGAAAAAG<hr/></td></tr><tr><td align="left"> </td><td align="left">GTTTTCTGACATGAGTCTGTGTGTTTATTTATTAATTATCTGTCAGTTTACGTCGTGGAATGCCTTCG</td></tr></tbody></table></table-wrap></sec><sec><title>Plasmids</title><p>RACE products were cloned into pGEM-T-Easy (Promega) and sequenced. Reporter constructs were generated by PCR amplifying sequence upstream of the 7.2 kb RNA splice donor site. Primers used to generate the reporter constructs are indicated in Table  <xref ref-type="table" rid="T1">1</xref>. Amplicons were resolved by gel electrophoresis and gel purified using the Qiaquick Gel Extraction Kit (Qiagen). Amplicons were cloned into pGEM-T-Easy. After insertion into the pGEM-T-Easy plasmid, the inserts were digested from the plasmid using the flanking EcoRI sequences then subcloned into the pGL3 Basic vector at a newly generated EcoRI site using the site directed mutatgenesis kit (Stratagene). Orientation of the cloned insert was determined by sequencing the pGL3 plasmid. The pGL3-SV40 control plasmid and the renilla phRL-TK normalization control plasmid were used in the luciferase assays (Promega).</p></sec><sec><title>Promoter analysis</title><p>Mouse fibroblasts were seeded in 24-well dishes and co-transfected with a pGL3 construct (see above) and phRL-TK using polyethylenimine (PEI) at a 6:1 ratio of PEI to plasmid DNA. The plasmids were cotransfected at a 1:1 ratio with 3 × 10<sup>10</sup> copies per well of each plasmid. Protein lysates were harvested 48 hours post transfection and assayed following the Duo-Glow Luciferase assay kit (Promega, Madison, WI). Luciferase activity was normalized to renilla activity in each well and the data is expressed as the fold change of luciferase induction relative to the luciferase induction from the pGL3 promoter-less vector.</p></sec><sec><title>RNA analyses</title><p>To determine expression kinetics, 10.1 fibroblasts were pretreated with 100 μg/mL cyclohexamide or 200 μg/mL PAA 1 hour before MCMV infection. Total RNA was harvested from 10.1 fibroblasts at either 24 hours post infection (h p.i.) or 48 h p.i. with TRIzol LS (Life Technologies) according to the manufacturer’s protocol. RNA was resolved on either a 1.4% or 0.7% glyoxal gel for detection of the spliced mRNA or 7.2 kb RNA respectively. Northern blot analysis for intron locus RNAs was carried out as previously described using specific radio-labeled riboprobes [<xref ref-type="bibr" rid="B14">14</xref>].</p><p>RNA half-life analysis was performed by infecting fibroblasts with wild-type MCMV at an MOI of 1.0. At 30 hours post infection, 4 ug/mL Actinomycin D was added to infected cells and RNA was harvested over a time course starting at time 0 and ending at 32 hours post treatment. Transcript levels were quantified by qRT-PCR at the different times points relative to RNA at time 0.</p></sec><sec><title>5′ and 3′ RACE</title><p>Total RNA harvested from mock or WT MCMV infected 10.1 mouse fibroblasts at 48 hours post infection was analyzed by the First Choice RNA ligase-mediated rapid amplification of cDNA ends kit as recommended by the manufacturer (RLM-RACE, Ambion). Amplification products were purified and TA cloned into pGEM-T-Easy (Promega, Madison, WI) and sequenced (Table  <xref ref-type="table" rid="T1">1</xref>). For 3′ RACE, RNA was reverse transcribed using a poly(A)-adapter. Amplification products were cloned and sequenced. MacVector software was used to align RACE sequences to MCMV reference sequence.</p></sec><sec><title>Primer extension</title><p>Oligonucleotides 497 and 50 were end radiolabeled and used in primer extension reactions on total RNA from mock- or MCMV-infected cells as previously described (Table  <xref ref-type="table" rid="T1">1</xref>) [<xref ref-type="bibr" rid="B19">19</xref>]. Primer extension products were analyzed by denaturing 10% urea-polyacrylamide gel electrophoresis followed by phosphorimager analysis.</p></sec><sec><title>qRT-PCR and qPCR</title><p>Total RNA was DNase treated and reverse transcribed using the Quantitect Reverse Transcription kit (Qiagen). Quantitative PCR was performed using the LightCycler 480 Probes Master Mix (Roche) along with IDT hydrolysis probes specific for the intron locus RNAs and selected housekeeping genes (Table  <xref ref-type="table" rid="T1">1</xref>). Ct values were determined using the Basic Relative Quantification analysis module of the LightCycler 480 (Roche) software. Primer-probe efficiencies were determined by three biological replicates of 10-fold dilutions. The 18S rRNA was used as a reference gene and the relative target levels were quantified by a delta-delta CT method, the Pfaffl method, that incorporates the calculated primer-probe efficiencies (Table  <xref ref-type="table" rid="T1">1</xref>) [<xref ref-type="bibr" rid="B48">48</xref>].</p></sec><sec><title>Immunoblotting</title><p>Cells were trypsinized, centrifuged, and collected in PBS. Cells were lysed in RIPA buffer (150 mM NaCl, 1% v/v Nonidet P-40, 0.5% w/v deoxycholate, 0.1% w/v SDS, 5 mM EDTA, 50 mM Tris; pH 8.0) containing protease inhibitor cocktail (Roche). The cell lysate was briefly sonicated to facilitate nuclear protein release and insoluble debris was centrifuged. GFP tagged m106 protein was immunoblotted using a rabbit polyclonal antibody and detected with a fluorescently conjugated secondary antibody using the SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific). HP1, heterochromatin associated protein 1, was detected similarly as a loading control (Santa Cruz).</p></sec><sec><title>FISH and Immunofluorescence</title><p>Fluorescently labeled RNA probes antisense to the 7.2 kb intron were generated using the FISH Tag kit (Table  <xref ref-type="table" rid="T1">1</xref>) (Invitrogen). Briefly, probes were in vitro transcribed from linearized pGEM-T-Easy constructs using an amino allyl modified base in which an alexa flour can be chemically attached to. Following in vitro transcription of the probes, the DNA templates are digested using DNase I and the amino modified RNA is purified over a column then ethanol precipitated. The purified probes are fluorescently labeled according to the manufacturer’s instructions then column purified and subsequent ethanol precipitation. Cells were fixed for 20 minutes in 4% paraformaldehyde, 10% acetic acid in 1x PBS. The fixation was quenched for 20 minutes in PBS with 0.1 M glycine. Cells were washed twice with PBS then permeabilized with 70% ethanol overnight at 4°C. Cells were rehydrated by washing twice with 50% formamide/2x SSC. The probe was denatured by heating at 65°C for 10 minutes in probe buffer then cells were incubated overnight with the denatured probe at 37°C. The following day, cells were washed twice with 0.1X SSC/50% formamide at 50°C then washed once with PBST. Immunoflourescence of m106-GFP was carried out as previously described [<xref ref-type="bibr" rid="B49">49</xref>].</p></sec><sec><title>Mice</title><p>All animal procedures were approved by the Institutional Animal Care and Use Committee of the University of Colorado Denver. BALB/c mice were inoculated by intraperitoneal injection with 5×10<sup>6</sup> pfu of tissue culture derived wild type or recombinant MCMV in 300 ul DMEM. At designated times mice were sacrificed and liver, spleen, lungs, kidneys, and salivary glands were removed and weighed. Part of the tissue was homogenized and titrated on mouse fibroblasts. The remaining tissue was processed for DNA isolation in order to quantify viral genomes using the DNeasy Blood and Tissue kit according to the manufacturer’s protocol (Qiagen). 250 ng of DNA was analyzed by qPCR for each sample.</p></sec></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Author contributions</title><p>TMS, L-AMV, CGA, and CAK participated in the design and execution of the study. TMS and CAK drafted the manuscript. All authors read and approved the final manuscript.</p></sec>
Inhibition of tetrodotoxin-resistant sodium current in dorsal root ganglia neurons mediated by D1/D5 dopamine receptors	<sec><title>Background</title><p>Dopaminergic fibers originating from area A11 of the hypothalamus project to different levels of the spinal cord and represent the major source of dopamine. In addition, tyrosine hydroxylase, the rate-limiting enzyme for the synthesis of catecholamines, is expressed in 8-10% of dorsal root ganglia (DRG) neurons, suggesting that dopamine may be released in the dorsal root ganglia. Dopamine has been shown to modulate calcium current in DRG neurons, but the effects of dopamine on sodium current and on the firing properties of small DRG neurons are poorly understood.</p></sec><sec><title>Results</title><p>The effects of dopamine and dopamine receptor agonists were tested on the tetrodotoxin-resistant (TTX-R) sodium current recorded from acutely dissociated small (diameter ≤ 25 μm) DRG neurons. Dopamine (20 μM) and SKF 81297 (10 μM) caused inhibition of TTX-R sodium current in small DRG neurons by 23% and 37%, respectively. In contrast, quinpirole (20 μM) had no effects on the TTX-R sodium current. Inhibition by SKF 81297 of the TTX-R sodium current was not affected when the protein kinase A (PKA) activity was blocked with the PKA inhibitory peptide (6–22), but was greatly reduced when the protein kinase C (PKC) activity was blocked with the PKC inhibitory peptide (19–36), suggesting that activation of D1/D5 dopamine receptors is linked to PKC activity. Expression of D1and D5 dopamine receptors in small DRG neurons, but not D2 dopamine receptors, was confirmed by Western blotting and immunofluorescence analysis. In current clamp experiments, the number of action potentials elicited in small DRG neurons by current injection was reduced by ~ 30% by SKF 81297.</p></sec><sec><title>Conclusions</title><p>We conclude that activation of D1/D5 dopamine receptors inhibits TTX-R sodium current in unmyelinated nociceptive neurons and dampens their intrinsic excitability by reducing the number of action potentials in response to stimulus. Increasing or decreasing levels of dopamine in the dorsal root ganglia may serve to adjust the sensitivity of nociceptors to noxious stimuli.</p></sec>	<contrib contrib-type="author" id="A1"><name><surname>Galbavy</surname><given-names>William</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>william.galbavy@stonybrookmedicine.edu</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Safaie</surname><given-names>Elham</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>elham.safaie@stonybrook.edu</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Rebecchi</surname><given-names>Mario J</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>mario.rebecchi@stonybrook.edu</email></contrib><contrib contrib-type="author" corresp="yes" id="A4"><name><surname>Puopolo</surname><given-names>Michelino</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I3">3</xref><email>michelino.puopolo@stonybrook.edu</email></contrib>	Molecular Pain	<sec><title>Background</title><p>Small DRG neurons with unmyelinated axons signal nociceptive sensory information from the body surface and viscera to the dorsal spinal horn. Small DRG neurons are unusual in expressing both tetrodotoxin-sensitive (TTX-S) (Na<sub>V</sub>1.1, Na<sub>V</sub>1.2, Na<sub>V</sub>1.6, and Na<sub>V</sub>1.7) and TTX-R (Na<sub>V</sub>1.8 and Na<sub>V</sub>1.9) voltage-dependent sodium channels [<xref ref-type="bibr" rid="B1">1</xref>-<xref ref-type="bibr" rid="B9">9</xref>]. Na<sub>V</sub>1.8 and Na<sub>V</sub>1.9 channels are preferentially expressed in small DRG neurons and fulfill critical aspects of their physiology [<xref ref-type="bibr" rid="B10">10</xref>-<xref ref-type="bibr" rid="B14">14</xref>]. In addition to their contribution to different aspects of pain [<xref ref-type="bibr" rid="B15">15</xref>-<xref ref-type="bibr" rid="B21">21</xref>], Na<sub>V</sub>1.8 channels play a major role during the upstroke of the action potential in small DRG neurons [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B22">22</xref>,<xref ref-type="bibr" rid="B23">23</xref>], and thus represent an ideal target for neuromodulators to affect their intrinsic excitability.</p><p>Dopaminergic fibers from area A11 of the hypothalamus project to different levels of the spinal cord and represent the major source of dopamine [<xref ref-type="bibr" rid="B24">24</xref>-<xref ref-type="bibr" rid="B28">28</xref>]. A functional role of dopamine in the spinal cord is suggested by: <italic>i</italic>) expression of D1-D5 dopamine receptors [<xref ref-type="bibr" rid="B29">29</xref>-<xref ref-type="bibr" rid="B31">31</xref>]; <italic>ii</italic>) peripheral nocifensive action of dopamine [<xref ref-type="bibr" rid="B32">32</xref>-<xref ref-type="bibr" rid="B37">37</xref>]; <italic>iii</italic>) post-synaptic effects of dopamine in the dorsal spinal horn that can inhibit nociceptive transmission [<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B39">39</xref>].</p><p>In addition to descending dopaminergic fibers from the hypothalamus, several subpopulations of DRG neurons, including those specialized in detecting low-threshold mechanosensory stimuli and those innervating pelvic organs, express tyrosine hydroxylase (TH), the rate-limiting enzyme for the synthesis of catecholamines, and thus most likely release dopamine as a modulatory transmitter [<xref ref-type="bibr" rid="B40">40</xref>-<xref ref-type="bibr" rid="B45">45</xref>]. Cultured DRG neurons release various peptides and ATP from their soma [<xref ref-type="bibr" rid="B46">46</xref>-<xref ref-type="bibr" rid="B48">48</xref>], suggesting that they may also release dopamine as well [<xref ref-type="bibr" rid="B49">49</xref>].</p><p>Dopamine has been shown to modulate calcium current in DRG neurons [<xref ref-type="bibr" rid="B50">50</xref>-<xref ref-type="bibr" rid="B52">52</xref>], but the effects of dopamine on sodium current and on the intrinsic excitability of small DRG neurons are poorly understood. Despite some reports showing inhibition of sodium current in chick sensory neurons [<xref ref-type="bibr" rid="B50">50</xref>], intracellular recordings from DRG neurons in isolated spinal ganglia have reported contrasting results showing depolarization, hyperpolarization or biphasic responses of the cell membrane upon application of dopamine [<xref ref-type="bibr" rid="B53">53</xref>-<xref ref-type="bibr" rid="B55">55</xref>].</p><p>Here, using an <italic>in vitro</italic> preparation of acutely dissociated DRG neurons, we show that dopamine inhibits the TTX-R sodium current in small DRG neurons and dampens their intrinsic excitability. Pharmacological studies indicate that the effect of dopamine is mediated by D1/D5 dopamine receptors.</p></sec><sec sec-type="results"><title>Results</title><sec><title>Dopamine effect on TTX-resistant sodium current</title><p>Small DRG neurons express both TTX-S and TTX-R sodium channels. In order to isolate the TTX-R sodium current, TTX-S sodium current was blocked by 300 nM TTX in the presence of 30 μM Cd<sup>2+</sup> to block calcium current. The inward current remaining in 300 nM TTX was completely blocked when 151 mM NaCl was replaced with equimolar tetraethylammonium-Cl (Figure <xref ref-type="fig" rid="F1">1</xref>A). Tetrodotoxin (300 nM) was only a weak inhibitor of the total sodium current, reducing the peak by 19.6 ± 14.0% (n = 32), similar to previous reports in small DRG neurons [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B23">23</xref>]. The TTX-R sodium current isolated with this protocol (Figure <xref ref-type="fig" rid="F1">1</xref>B) is consistent with sodium current carried through Na<sub>V</sub>1.8 channels previously reported in small DRG neurons [<xref ref-type="bibr" rid="B1">1</xref>-<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B56">56</xref>]. At the holding potential of −80 mV, there was no obvious component of non-inactivating current from Na<sub>V</sub>1.9 channels. The TTX-R sodium current recorded with this protocol was quite stable over 20–25 min (Figure <xref ref-type="fig" rid="F1">1</xref>C). In collected results (Figure <xref ref-type="fig" rid="F1">1</xref>D,E), the TTX-R sodium current showed no significant decrease during 25 min, changing from 1.01 ± 0.03 (normalized peak) during the first 5 min to 0.97 ± 0.04 after 25 min (n = 11, paired t-test, p = 0.104). We next used this same protocol to test the effect of dopamine on TTX-R sodium current. After 5 min in control, the neuron was challenged with 20 μM dopamine (Figure <xref ref-type="fig" rid="F1">1</xref>F). Dopamine caused a clear inhibition of TTX-R sodium current. In collected results (Figure <xref ref-type="fig" rid="F1">1</xref>G,H), the TTX-R sodium current was reduced from 0.99 ± 0.01 to 0.77 ± 0.09 (n = 7, paired t-test, p < 0.01) after 15 min in 20 μM dopamine.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Dopamine effect on TTX-R sodium current. A)</bold> Isolation of TTX-R sodium current from a small DRG neuron (diameter ≤ 25 μm). Top: voltage clamp protocol. A single step of voltage from −80 to 0 mV, 30 msec duration, was delivered every 5 sec. Bottom: Total sodium current (Total I<sub>Na</sub>) recorded in Tyrode’s solution supplemented with 30 μM CdCl<sub>2</sub> to block calcium current. Subsequent application of 300 nM tetrodotoxin blocked the TTX-S sodium current leaving only the TTX-R sodium current which was completely blocked when 151 mM NaCl was replaced by equimolar concentration of tetraethylammonium-Cl. <bold>B)</bold> TTX-S and TTX-R sodium currents isolated by subtraction from the cell in A. <bold>C)</bold> The TTX-R sodium current recorded in a small DRG neuron was normalized to the peak TTX-R sodium current recorded during the first 5 min in whole-cell and monitored for 25 min. <bold>D)</bold> Collected results showing the relative peak of TTX-R sodium current during 25 min. <bold>E)</bold> The relative peak of TTX-R sodium current was 1.01 ± 0.03 at 5 min and 0.97 ± 0.04 at 25 min (n = 11, paired t-Test, p = 0.104). <bold>F)</bold> In a small DRG neuron, 20 μM dopamine reduced the relative peak of TTX-R sodium current from 0.99 (after 5 min in control) to 0.75 (after 15 min in 20 μM dopamine). <bold>G)</bold> Collected results showing the effect of 20 μM dopamine on the relative peak of TTX-R sodium current. <bold>H)</bold> The relative peak of TTX-R sodium current was reduced from 0.99 ± 0.01 after 5 min in control to 0.77 ± 0.09 (n = 7, paired t-test, p < 0.01) after 15 min in 20 μM dopamine.</p></caption><graphic xlink:href="1744-8069-9-60-1"/></fig></sec><sec><title>Effect of SKF 81297 on TTX-R sodium current</title><p>The next step was to identify which dopamine receptors mediate this inhibitory effect. The D1/D5 receptor agonist SKF 81297 produced inhibition very much like that of dopamine (Figure <xref ref-type="fig" rid="F2">2</xref>A). In collected results (Figure <xref ref-type="fig" rid="F2">2</xref>B,C), the TTX-R sodium current was reduced from 0.99 ± 0.02 to 0.63 ± 0.08 (n = 22, paired t-test, p < 0.01) after 15 min in 10 μM SKF 81297. In contrast, the D2 receptor agonist quinpirole had no significant effect (Figure <xref ref-type="fig" rid="F2">2</xref>D); in collected results (Figure <xref ref-type="fig" rid="F2">2</xref>E,F), the TTX-R sodium current was reduced from 0.99 ± 0.01 to 0.94 ± 0.04 (n = 5, paired t-test, p = 0.065) after 15 min in 20 μM quinpirole. These results strongly suggest that the inhibitory effect of dopamine on TTX-R sodium current in small DRG neurons is mediated by activation of D1/D5 dopamine receptors.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Effect of D1/D5 and D2 dopamine receptor agonists on TTX-R sodium current. A)</bold> In a small DRG neuron (diameter ≤ 25 μm), 10 μM SKF 81297 (D1/D5 dopamine receptors agonist) reduced the relative peak of TTX-R sodium current from 0.97 after 5 min in control to 0.51 after 15 min in SKF 81297. Inset: representative TTX-R sodium current traces recorded after 5 min in control (trace 1), after 15 min in 10 μM SKF 81297 (trace 2), and after 5 min washout (trace 3). <bold>B)</bold> Collected results showing the effect of 10 μM SKF 81297 on the relative peak of TTX-R sodium current. <bold>C)</bold> The relative peak of TTX-R sodium current was reduced from 0.99 ± 0.02 after 5 min in control to 0.63 ± 0.08 (n = 22, paired t-Test, p < 0.01) after 15 min in 10 μM SKF 81297. <bold>D)</bold> In a small DRG neuron, 20 μM quinpirole (D2 dopamine receptor agonist) caused only a little change on the relative peak of TTX-R sodium current from 0.97 after 5 min in control to 0.94 after 15 min in 20 μM quinpirole. <bold>E)</bold> Collected results showing the effect of 20 μM quinpirole on the relative peak of TTX-R sodium current. <bold>F)</bold> The relative peak of TTX-R sodium current was reduced from 0.99 ± 0.01 after 5 min in control to 0.94 ± 0.04 (n = 5, paired t-test, p = 0.065) after 15 min in 20 μM quinpirole.</p></caption><graphic xlink:href="1744-8069-9-60-2"/></fig><p>To test this further, we used SCH 23390, a selective antagonist at D1/D5 dopamine receptors [<xref ref-type="bibr" rid="B57">57</xref>,<xref ref-type="bibr" rid="B58">58</xref>]. As expected, the effect of SKF 81297 (Figure <xref ref-type="fig" rid="F3">3</xref>A, black circles) was antagonized by SCH 23390 (Figure <xref ref-type="fig" rid="F3">3</xref>A, blue squares). Similarly, the effect of SKF 81297 was also reduced by the alkylating agent ethoxycarbonyl-2-ethoxy-1,2-dihydroquinoline (EEDQ), a broad, irreversible, dopamine receptor antagonist [Figure <xref ref-type="fig" rid="F3">3</xref>A, 20 μM EEDQ (red triangles) and 100 μM EEDQ (green diamonds)]. In collected results (Figure <xref ref-type="fig" rid="F3">3</xref>B), the inhibitory effect of 10 μM SKF 81297 on TTX-R sodium current was reduced from 0.63 ± 0.08 (n = 22) when applied without the antagonist (black bars) to 0.83 ± 0.04 (n = 8, one-way ANOVA, followed by Dunnett post-hoc analysis p < 0.01) when applied in combination with 3 μM SCH 23390 (blue bars); to 0.79 ± 0.06 (n = 6, one-way ANOVA, followed by Dunnett post-hoc analysis, p < 0.01) when applied in combination with 20 μM EEDQ (red bars); to 0.92 ± 0.07 (n = 9, one-way ANOVA, followed by Dunnett post-hoc analysis, p < 0.01) when applied in combination with 100 μM EEDQ (green bars).</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>The effect of SKF 81297 (D1/D5 dopamine receptors agonist) on TTX-R sodium current was antagonized by SCH 23390 and EEDQ. A)</bold> Collected results showing the effect of 10 μM SKF 81297 on TTX-R sodium current applied either alone (black circles), or in combination with 3 μM SCH 23390 (blue squares), or 20 μM EEDQ (red triangles), or 100 μM EEDQ (green diamonds). <bold>B)</bold> The inhibitory effect of 10 μM SKF 81297 on the relative peak of TTX-R sodium current was reduced from 0.63 ± 0.08 (n = 22) when applied without the antagonist (black bars) to 0.83 ± 0.04 (n = 8, one-way ANOVA, followed by post-hoc Dunnett analysis p < 0.01) when applied on DRG neurons incubated for 15 min with 3 μM SCH 23390 (blue bars); to 0.79 ± 0.06 (n = 6, one-way ANOVA, followed by post-hoc Dunnett analysis p < 0.01) when applied on DRG neurons incubated for 30 min with 20 μM EEDQ (red bars); to 0.92 ± 0.07 (n = 9, one-way ANOVA, followed by post-hoc Dunnett analysis, p < 0.01) when applied on DRG neurons incubated for 30 min with 100 μM EEDQ (green bars).</p></caption><graphic xlink:href="1744-8069-9-60-3"/></fig></sec><sec><title>Intracellular pathways downstream to D1/D5 dopamine receptors</title><p>Previous work has described the ability of PKA and PKC to modulate both TTX-S sodium current in central neurons [<xref ref-type="bibr" rid="B57">57</xref>-<xref ref-type="bibr" rid="B59">59</xref>] and TTX-R sodium current in peripheral sensory neurons [<xref ref-type="bibr" rid="B60">60</xref>-<xref ref-type="bibr" rid="B65">65</xref>]. When PKA activity was blocked by including in the patch pipette 10 μM of PKA inhibitory fragment (6–22) [PKAi(6–22)], the inhibitory effect of 10 μM SKF 81297 on TTX-R sodium current in small DRG neurons was unaffected. In collected results (Figure <xref ref-type="fig" rid="F4">4</xref>A,B), the TTX-R sodium current was reduced from 0.99 ± 0.01 to 0.63 ± 0.08 (n = 22) after 15 min in 10 μM SKF 81297 (Figure <xref ref-type="fig" rid="F4">4</xref>A, circles) and from 0.99 ± 0.01 to 0.66 ± 0.06 (n = 11, t-test, two populations, p = 0.229) after 15 min in 10 μM SKF 81297 with 10 μM PKAi(6–22) included in the patch pipette (Figure <xref ref-type="fig" rid="F4">4</xref>A, squares).</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>Intracelluar pathway downstream to D1/D5 dopamine receptors activation. A)</bold> Collected results showing the effect of 10 μM SKF 81297 on TTX-R sodium current applied either alone (circles) or in combination with 10 μM PKAi(6–22) (squares) included in the patch pipette to block the PKA activity. <bold>B)</bold> The inhibitory effect of 10 μM SKF 81297 on the relative peak of TTX-R sodium current was slightly reduced from 0.63 ± 0.08 (n = 22) when applied alone to 0.66 ± 0.06 (n = 11, t-test, two populations, p = 0.229) when applied in combination with 10 μM PKAi(6–22) included in the patch pipette. <bold>C)</bold> Collected results showing the effect of 10 μM SKF 81297 on TTX-R sodium current applied either alone (circles) or in combination with 5 μM PKCi(19–36) (triangles) included in the patch pipette to block the PKC activity. <bold>D)</bold> The inhibitory effect of 10 μM SKF 81297 on the relative peak of TTX-R sodium current was significantly reduced from 0.63 ± 0.08 (n = 22) when applied alone to 0.78 ± 0.06 (n = 11, t-test, two populations, p < 0.01) when applied in combination with 5 μM of PKCi(19–36) included in the patch pipette.</p></caption><graphic xlink:href="1744-8069-9-60-4"/></fig><p>In contrast, when PKC activity was blocked by including in the patch pipette 5 μM of PKC inhibitory fragment (19–36) [PKCi(19–36)], the inhibitory effect of 10 μM SKF 81297 on the TTX-R sodium current was substantially affected. In collected results (Figure <xref ref-type="fig" rid="F4">4</xref>C,D), the TTX-R sodium current was reduced from 0.99 ± 0.01 to 0.63 ± 0.08 (n = 22) after 15 min in 10 μM SKF 81297 (Figure <xref ref-type="fig" rid="F4">4</xref>C, circles) and from 0.99 ± 0.01 to 0.78 ± 0.06 (n = 11, t-test, two populations, p < 0.01) after 15 min in 10 μM SKF 81297 with 5 μM of PKCi(19–36) included in the patch pipette (Figure <xref ref-type="fig" rid="F4">4</xref>C, triangles). Taken together, these results suggest that activation of D1/D5 dopamine receptors in small DRG neurons is coupled to PKC activity.</p></sec><sec><title>Effect of SKF 81297 on steady-state activation and inactivation curves of TTX-R sodium channels</title><p>We next tested whether D1/D5 dopamine receptors inhibition of TTX-R sodium channels is mediated by changes in the voltage-dependence of activation or inactivation of the channels. The steady-state activation curve had a midpoint of −15.1 ± 4.2 mV and a slope of 5.0 ± 1.8 mV (n = 17) in control (Figure <xref ref-type="fig" rid="F5">5</xref>C, circles) versus −12.7 ± 7.4 mV (t-test, two populations, p = 0.270) and 4.9 ± 2.1 mV (n = 15) in 10 μM SKF 81297 (Figure <xref ref-type="fig" rid="F5">5</xref>C, diamonds). The steady-state inactivation curve had a midpoint of −34.7 ± 3.2 mV and a slope of 4.9 ± 1.1 mV (n = 17) in control (Figure <xref ref-type="fig" rid="F5">5</xref>D, circles) versus −31.3 ± 4.2 mV (t-test, two populations, p < 0.05) and 5.2 ± 1.9 mV (n = 15) in 10 μM SKF 81297 (Figure <xref ref-type="fig" rid="F5">5</xref>D, diamonds). The changes induced by SKF 81297 on the steady-state activation and inactivation curves of TTX-R sodium channels are similar to previously reported effects of D1/D5 dopamine receptors activation on TTX-S sodium channels both in central neurons [<xref ref-type="bibr" rid="B57">57</xref>-<xref ref-type="bibr" rid="B59">59</xref>] and in retinal ganglion cells [<xref ref-type="bibr" rid="B66">66</xref>].</p><fig id="F5" position="float"><label>Figure 5</label><caption><p><bold>Effect of SKF 81297 on the steady-state activation and inactivation curves of TTX-R sodium channels. A)</bold> Voltage-dependence activation of TTX-R sodium channels. TTX-R voltage-gated sodium currents were elicited by 500 msec steps from a holding potential of – 80 mV to voltages between −120 and + 30 mV in 10 mV increments. <bold>B)</bold> Voltage-dependence of inactivation of TTX-R sodium channels. TTX-R voltage-gated sodium currents were elicited with a test pulse to −10 mV preceded by 500 msec prepulses from −120 to + 30 mV in 10 mV increments. <bold>C)</bold> Collected results showing the effect of 10 μM SKF 81297 on the steady-state activation curve of TTX-R sodium channels in small DRG neurons (diameter ≤ 25 μm). The steady-state activation curve had a midpoint for activation of −15.1 ± 4.2 mV and a slope of 5.0 ± 1.8 mV (n = 17) in control (circles) versus −12.7 ± 7.4 mV (t-test, two populations, p = 0.270) and 4.9 ± 2.1 mV (n = 15) in neurons incubated for 15 min in 10 μM SKF 81297 (diamonds). Dotted lines are best fits to the Boltzmann function. Error bars were omitted for clarity. <bold>D)</bold> Collected results showing the effect of 10 μM SKF 81297 on the steady-state inactivation curve of TTX-R sodium channels (same neurons as in <bold>C</bold>). The steady-state inactivation curve had a midpoint for inactivation of −34.7 ± 3.2 mV and a slope of 4.9 ± 1.1 mV (n = 17) in control (circles) versus −31.3 ± 4.2 mV (t-test, two populations, p < 0.05) and 5.2 ± 1.9 mV (n = 15) in 10 μM SKF 81297 (diamonds). Dotted lines are best fits to the Boltzmann function. Error bars were omitted for clarity.</p></caption><graphic xlink:href="1744-8069-9-60-5"/></fig></sec><sec><title>Effect of SKF 81297 on firing</title><p>The TTX-R sodium current plays a major role during the upstroke of the action potential in small DRG neurons [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B22">22</xref>,<xref ref-type="bibr" rid="B23">23</xref>], and thus may represent an ideal target for neuromodulators to induce changes in the intrinsic excitability of nociceptors. In order to test this hypothesis, the effect of SKF 81297 was assessed in current clamp in which trains of action potentials were elicited in small DRG neurons by current injection [<xref ref-type="bibr" rid="B56">56</xref>]. During the initial control experiments in current clamp, we observed a progressive reduction in the number of action potentials elicited with repeated current injections. The number of action potentials decreased from 15.3 ± 3.8 (n = 6) during the first ramp to 9.8 ± 2.5 (n = 6) during the fifth ramp, even with 30 sec interval between current injections, possibly reflecting accumulation of slow inactivation of sodium channels during sustained depolarization [<xref ref-type="bibr" rid="B56">56</xref>]. For this reason, we decided to limit the analysis to action potentials elicited during the first current injection by making a comparison between small DRG neurons in control to those incubated with 10 μM SKF 81297 for 15 min (Figure <xref ref-type="fig" rid="F6">6</xref>A,B). The resting potential of small DRG neurons in control (−73.6 ± 4.9 mV, n = 15) was very similar to the resting potential of small DRG neurons incubated for 15 min in 10 μM SKF 81297 (−73.7 ± 5.0 mV, n = 12, t-test, two populations, p = 0.978). The total number of action potentials during the first current injection decreased from 14.9 ± 3.4 (n = 15) in control to 10.0 ± 3.6 (n = 12, t-test, two populations p < 0.01) in 10 μM SKF 81297. Although not statistically significant, there appeared to be also a decrease in the peak of the first action potential from 38.4 ± 10.4 mV (n = 15) in control to 33.2 ± 7.3 mV (n = 12, t-test, two populations, p = 0.137) in 10 μM SKF 81297; a decrease in the maximum upstroke velocity of the first action potential from 87.0 ± 31.9 V/sec (n = 15) in control to 77.8 ± 35.2 V/sec (n = 12, t-test, two populations, p = 0.484) in 10 μM SKF 81297; a decrease in the firing frequency from 33.1 ± 7.5 Hz (n = 15) in control to 29.9 ± 13.8 Hz (n = 12, t-test, two populations, p = 0.459) in 10 μM SKF 81297 (Figure <xref ref-type="fig" rid="F6">6</xref>C). Taken together, these results suggest that activation of D1/D5 dopamine receptors induces substantial changes in the intrinsic excitability of small DRG neurons and reduces their ability to sustain volleys of action potentials in response to current injection.</p><fig id="F6" position="float"><label>Figure 6</label><caption><p><bold>Effect of SKF 81297 on the firing properties of small DRG neurons. A)</bold> Top: current clamp protocol: 0.8 sec ramp of current to 1 nA was injected to elicit action potentials. Bottom: action potentials elicited in a small DRG neuron (diameter ≤ 25 μm) in control. <bold>B)</bold> Action potentials elicited in a different small DRG neuron (diameter ≤ 25 μm) incubated for 15 min in 10 μM SKF 81297. <bold>C)</bold> In collected results, the total number of action potentials during the first current injection decreased from 14.9 ± 3.4 (n = 15) in control to 10.0 ± 3.6 (n = 12) in 10 μM SKF 81297 (t-test, two populations p < 0.01); the peak of the first action potential decreased from 38.4 ± 10.4 mV (n = 15) in control to 33.2 ± 7.3 mV (n = 12) in 10 μM SKF 81297 (t-test, two populations, p = 0.137); the maximum upstroke velocity of the first action potential decreased from 87.0 ± 31.9 V/sec (n = 15) in control to 77.8 ± 35.2 V/sec (n = 12) in 10 μM SKF 81297 (t-test, two populations, p = 0.484); the firing frequency decreased from 33.1 ± 7.5 Hz (n = 15) in control to 29.9 ± 13.8 Hz (n = 12) in 10 μM SKF 81297 (t-test, two populations, p = 0.459).</p></caption><graphic xlink:href="1744-8069-9-60-6"/></fig></sec><sec><title>Expression of dopamine receptors in small DRG neurons</title><p>Our electrophysiology data strongly suggest that the effect of dopamine on TTX-R sodium current in small DRG neurons is mediated by activation of D1/D5 dopamine receptors, without any significant contribution of D2 dopamine receptors. To further test this hypothesis, we characterized the expression of D1, D2, and D5 dopamine receptors in DRG neurons. Relative tissue expression levels of D1, D2, and D5 dopamine receptor subtypes were evaluated by Western blotting. Surprisingly, levels of D1 and D5 dopamine receptors were substantially higher in dorsal root ganglia compared to striatum, cortex, or lumbar spinal cord. In contrast, D2 dopamine receptors were undetected in dorsal root ganglia but, highly enriched in striatum and cortex (Figure <xref ref-type="fig" rid="F7">7</xref>). Immunofluorescence imaging of dissociated DRG neurons showed that nearly all small DRG neurons (diameter ≤ 30 μm), both isolectin B4 positive [IB4(+)] and IB4(−) neurons, expressed D1 and D5 dopamine receptors. In contrast, most DRG neurons with diameter ≥ 30 μm were only faintly stained (Figure <xref ref-type="fig" rid="F8">8</xref>). While some plasma membrane staining was observed in small DRG neurons, most of the immunofluorescence signal was intracellular, apparently vesicular. The immunofluorescence results from fixed dissociated primary cultures were confirmed in fixed dorsal root ganglia sections (Figure <xref ref-type="fig" rid="F9">9</xref>). D1 and D5 dopamine receptors were primarily expressed in smaller DRG neurons. The intensity varied among the smaller neurons and no clear pattern of IB4(+) or IB4(−) association emerged. Although the immunofluorescence signals for D1 and D5 dopamine receptors in the larger IB4(−) neurons were above background, the overall intensities were much lower than those in smaller diameter neurons.</p><fig id="F7" position="float"><label>Figure 7</label><caption><p><bold>Western blotting of dopamine receptors.</bold> Immunoblots for D1, D2, and D5 dopamine receptors were obtained for frontal cortex (CTX), striatum (STM), spinal cord (SC) and dorsal root ganglia (DRG) proteins. All wells were loaded with the same concentration of total protein with GAPDH serving as an internal control. <bold>A)</bold> Membranes were probed with antibodies specific for D1, D2, or D5 dopamine receptors. D1 and D5 were identified in CTX, STM, and DRG as single bands with apparent MW’s of 68 kDa. D2 dopamine receptor protein was identified in CTX and STM as a single band with an apparent MW of 62 kDa. Corresponding sections from each blot were probed with antibody against GAPDH (shown beneath the membranes in <bold>A</bold>). <bold>B)</bold> Western blotting of D1 and D5 dopamine receptors were repeated comparing equal concentrations of SC and DRG protein. Corresponding sections from each blot were probed with antibody against GAPDH (shown beneath the membranes in <bold>B</bold>).</p></caption><graphic xlink:href="1744-8069-9-60-7"/></fig><fig id="F8" position="float"><label>Figure 8</label><caption><p><bold>Expression of dopamine receptors in acutely isolated DRG neurons.</bold> Acutely dissociated DRG neurons were fixed in 4% formaldehyde, permeabilized, blocked and then probed with lectin FITC-labeled IB4 (green channel) or D1 (red channel, <bold>A</bold>) or D5 dopamine receptors antibodies (red channel, <bold>B</bold>) and secondary antibodies against mouse IgG labeled with Alexa 594. Merging the images indicated some co-localization of IB4 and either D1 or D5 dopamine receptors (yellow). Scale bars shown are 50 μm.</p></caption><graphic xlink:href="1744-8069-9-60-8"/></fig><fig id="F9" position="float"><label>Figure 9</label><caption><p><bold>Expression of dopamine receptors in intact dorsal root ganglia.</bold> Frozen sections, 20 μm thick, were prepared from lumbar DRG taken from P14 to P28 Sprague Dawley rats perfused with formaldehyde. Sections of dorsal root ganglia were stained with FITC-labeled lectin IB4 (green channel) and monoclonal antibodies to either D1 (red channel, <bold>A</bold>) or D5 dopamine receptors antibodies (red channel, <bold>B</bold>), and secondary antibodies against mouse IgG labeled with Alexa 594. Merging the images indicated some co-localization of IB4 and either D1 or D5 dopamine receptors (yellow). Scale bars are 50 μm.</p></caption><graphic xlink:href="1744-8069-9-60-9"/></fig></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>The data presented here show a strong inhibition of TTX-R sodium current in DRG neurons by activation of D1/D5 dopamine receptors, accompanied by decrease in their intrinsic excitability. Only small DRG neurons (diameter of 24.8 ± 2.6 μm and cell capacitance of 19.5 ± 4.4 pF, n = 111) sensitive to capsaicin were included in the study, and among those about 88% were also IB4(+). All these features suggest that the dopamine-sensitive neurons correspond to small nociceptive neurons [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B67">67</xref>-<xref ref-type="bibr" rid="B70">70</xref>].</p><p>Western blot and immunofluorescence analysis confirmed the expression of D1 and D5 dopamine receptors in dorsal root ganglia, but no evidence was found for the expression of D2 dopamine receptors. These results were consistent with those obtained with electrophysiology in which the D2 agonist quinpirole had little effect. Using polymerase chain reaction (PCR) and DNA sequencing, transcripts of D1-D5 dopamine receptors, including D2 dopamine receptors, have been reported in dorsal root ganglia [<xref ref-type="bibr" rid="B30">30</xref>,<xref ref-type="bibr" rid="B31">31</xref>]. A possible explanation for this discrepancy could be that, even though the transcript for D2 dopamine receptors is detectable with PCR, the protein level could be too low to be detected by Western blotting or immunofluorescence microscopy. Alternatively, it is possible that once translated into protein, D2 dopamine receptors are rapidly transported from the cell body to the terminals, and thus are missed with our analysis.</p><p>In most DRG neurons, the inhibitory effects of dopamine or SKF 81297 on TTX-R sodium current required many minutes to develop fully. In most of the neurons tested, the inhibitory effect of the drug did not reach a steady-state even after 15 min. This is slower than the inhibitory effect of dopamine on TTX-S sodium current in hippocampal or prefrontal cortex neurons [<xref ref-type="bibr" rid="B58">58</xref>,<xref ref-type="bibr" rid="B59">59</xref>], but similar to that in retinal ganglion cells [<xref ref-type="bibr" rid="B66">66</xref>]. The relative slow development of inhibition is consistent with mediation by a second messenger pathway involving PKC.</p><p>Activation of PKA or PKC has been reported to modulate Na<sub>V</sub>1.8 channels in DRG neurons with increases in current amplitude and changes in the kinetic properties [<xref ref-type="bibr" rid="B60">60</xref>-<xref ref-type="bibr" rid="B62">62</xref>]. Our results (Figure <xref ref-type="fig" rid="F4">4</xref>A) suggest that PKA activity is not required for modulation of TTX-R sodium current by D1/D5 dopamine receptor activation, in contrast with results in central neurons where activation of D1/D5 dopamine receptors inhibits the TTX-S sodium current through activation of PKA [<xref ref-type="bibr" rid="B57">57</xref>-<xref ref-type="bibr" rid="B59">59</xref>].</p><p>The reduced effect of SKF 81297 on TTX-R sodium current when blocking PKC (Figure <xref ref-type="fig" rid="F4">4</xref>C) strongly suggests that PKC activity is required for modulation of TTX-R sodium current upon activation of D1/D5 dopamine receptors. Activation [<xref ref-type="bibr" rid="B61">61</xref>,<xref ref-type="bibr" rid="B63">63</xref>,<xref ref-type="bibr" rid="B64">64</xref>] or inhibition [<xref ref-type="bibr" rid="B61">61</xref>] of PKC has been reported to up-regulate or down-regulate, respectively, the TTX-R sodium current in small DRG neurons. Even though multiple mechanisms/pathways are required to explain the full actions of PKC on TTX-R sodium current, the reduced effect of SKF 81297 on the TTX-R sodium current observed in our experiments is more consistent with inhibition of PKC activity upon activation of D1/D5 dopamine receptors. Additional experiments are needed to better characterize the intracellular pathway downstream to D1/D5 dopamine receptor activation in small DRG neurons.</p><p>The stimulation of D1/D5 dopamine receptors reduced the excitability of small DRG neurons, as manifested in reduction of action potential firing, and a decrease in the peak and the maximum upstroke velocity of action potentials. These effects are consistent with inhibition of TTX-R (Na<sub>V</sub>1.8) current, the main current responsible for the upstroke of the action potential in DRG neurons [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B22">22</xref>,<xref ref-type="bibr" rid="B23">23</xref>]. It is also likely, however, that D1/D5 dopamine receptor activation can modulate other ion channels critical to action potential generation in DRG neurons. Whether this could account for the changes observed here is unknown and will require further experiments.</p><p>Many observations support the idea that dopamine may be released locally in the dorsal root ganglia: <italic>i</italic>) TH, the rate-limiting enzyme for the synthesis of catecholamines, is expressed in about 8-10% of DRG neurons [<xref ref-type="bibr" rid="B40">40</xref>,<xref ref-type="bibr" rid="B41">41</xref>,<xref ref-type="bibr" rid="B43">43</xref>,<xref ref-type="bibr" rid="B44">44</xref>]; <italic>ii</italic>) dopamine and its metabolites dihydroxyphenylacetic acid (DOPAC) and homovanillic acid (HVA) have been detected in DRG neurons and in dorsal spinal nerve roots, but not in satellite and Schwann cells [<xref ref-type="bibr" rid="B45">45</xref>,<xref ref-type="bibr" rid="B71">71</xref>,<xref ref-type="bibr" rid="B72">72</xref>]; <italic>iii</italic>) when DRG neurons are loaded with dopamine, dopamine release could be evoked by high potassium stimulation and detected by amperometric means, suggesting that DRG neurons may carry the necessary release machinery for dopamine [<xref ref-type="bibr" rid="B49">49</xref>]. A question however remains why previous work has failed to detect other enzymes, namely the aromatic acid decarboxylase (AADC) and dopamine β-hydroxylase (DBH), involved in the synthesis of catecholamines [<xref ref-type="bibr" rid="B40">40</xref>,<xref ref-type="bibr" rid="B44">44</xref>,<xref ref-type="bibr" rid="B73">73</xref>,<xref ref-type="bibr" rid="B74">74</xref>], even though in some cases AADC and DBH have been detected in TH(−) DRG neurons [<xref ref-type="bibr" rid="B44">44</xref>]. A possible explanation for this discrepancy could be that the protein levels of AADC and DBH are too low to be detected by immunocytochemistry. Alternatively, it is possible that TH(+) DRG neurons may synthesize L-DOPA, which after release is converted to dopamine by AADC localized in other cell types.</p><p>In other sensory systems, including the retina and the olfactory bulb, spontaneously active dopaminergic cells provide a tonic release of dopamine [<xref ref-type="bibr" rid="B75">75</xref>-<xref ref-type="bibr" rid="B78">78</xref>], which plays an important function in sensory adaptation. In the retina, dopamine sets the gain of the retinal networks for vision in bright light [<xref ref-type="bibr" rid="B79">79</xref>]. Similarly, in the olfactory bulb, dopamine plays an important function in setting the gain for odor discrimination [<xref ref-type="bibr" rid="B80">80</xref>-<xref ref-type="bibr" rid="B84">84</xref>]. TH(+) DRG neurons form a selective class of unmyelinated Low-Threshold mechanoreceptors (C-LTMRs) innervating hair follicles in the skin [<xref ref-type="bibr" rid="B41">41</xref>] and associated with light touch and injury-induced mechanical hypersensitivity [<xref ref-type="bibr" rid="B85">85</xref>,<xref ref-type="bibr" rid="B86">86</xref>]. TH(+) C-LTMRs are not spontaneously active, but in response to light mechanical force they respond with a train of action potentials [<xref ref-type="bibr" rid="B41">41</xref>] that can trigger dopamine release. Light touch able to induce stretch of the skin or deflection of hair follicles able to activate C-LTMRs is a very frequent experience during the daily activity, supporting the possibility that TH(+) C-LTMRs may release dopamine in the dorsal root ganglia very frequently. Thus, increasing or decreasing levels of dopamine in the dorsal root ganglia may serve to adjust the sensitivity of nociceptors to noxious stimuli, with increasing levels of dopamine raising the threshold for pain and reduced levels increasing the sensitivity to noxious stimuli. Future behavioral experiments using animal models of pain are needed to test this hypothesis.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>In summary, our results demonstrate that DRG neurons express D1 and D5 dopamine receptors, but not D2 dopamine receptors. Agonists of D1/D5 dopamine receptors cause robust inhibition of TTX-R sodium current in small DRG neurons and decrease their ability to fire action potentials in response to stimulus. A functional implication of the inhibitory effect of dopamine could be that, by modulating the TTX-R sodium current and the intrinsic excitability of DRG neurons, increasing or decreasing levels of dopamine may adjust the sensitivity of nociceptor sensory neurons to incoming sensory stimuli.</p></sec><sec sec-type="materials\|methods"><title>Materials and methods</title><sec><title>Preparation of isolated Dorsal Root Ganglia neurons</title><p>Isolated DRG neurons were prepared from Sprague Dawley rats, postnatal day 14–28. Animal were anesthetized with isoflurane and decapitated. Both thoracic and lumbar segments of the spinal cord were removed and placed in a cold Ca<sup>2+</sup>, Mg<sup>2+</sup>-free Hank’s solution containing (in mM) 137 NaCl, 5.3 KCl, 0.33 Na<sub>2</sub>HPO<sub>4</sub>, 0.44 KH<sub>2</sub>PO<sub>4</sub>, 5 HEPES, 5.5 Glucose, pH = 7.4 with NaOH. The bone surrounding the spinal cord was removed and dorsal root ganglia were exposed and pulled out. After removing the roots, ganglia were chopped in half and incubated for 20 min at 34°C in Ca<sup>2+</sup>, Mg<sup>2+</sup>-free Hank’s solution containing 20 U/ml Papain (Worthington Biochemical, Lakewood, NJ) and 5 mM D,L-cysteine. Ganglia were washed and incubated for 20 min at 34°C in Ca<sup>2+</sup>, Mg<sup>2+</sup>-free Hank’s solution containing 3 mg/ml collagenase (Type I, Sigma-Aldrich, St. Louis, MO) and 4 mg/ml Dispase II (Boehringer Mannheim, Indianapolis, IN). Ganglia were then placed in Leibovitz’s L-15 medium (Invitrogen, San Diego, CA) supplemented with 10% fetal calf serum, 5 mM HEPES, and 50 ng/ml NGF (Invitrogen, San Diego, CA). Individual cells were dispersed by mechanical trituration using fire-polished Pasteur pipettes with decreasing bore size and plated on glass coverslip treated with 30–50 μg/ml poly-D-lysine. Cells were incubated in the supplemented L-15 solution at 34°C (in 5% CO<sub>2</sub>) for 2 hours, then stored at room temperature in Neurobasal medium (Gibco) and used over the next 4–6 hours. This protocol yields spherical cell bodies without neurites, which enhances voltage control of sodium current. The cells can be lifted from the cover slip after establishing the whole-cell configuration in order to facilitate rapid solution changes using flow pipes.</p></sec><sec><title>Western blotting</title><p>Immunoblots were performed for D1, D2 and D5 dopamine receptor proteins in dorsal root ganglia, frontal cortex, striatum and spinal cord of P14 to P28 Sprague Dawley rats. Animals were euthanized and perfused with heparinized saline. The dissected nervous tissue samples were immediately frozen on dry ice and stored at −80°C. Proteins were extracted from the frozen samples with ice cold lysis buffer prepared with complete EDTA-free protease inhibitor cocktail tablets (Roche Pharmaceuticals), 20 mM Tris, 150 mM NaCl, 2.5 mM Na<sub>4</sub>P<sub>2</sub>O<sub>7</sub>, 1% Nonidet P40, 0.1% SDS and 1 mM of EDTA, NaF, PMSF, Na<sub>3</sub>VO<sub>4</sub>, and dithiothrietol. Three 2.3 mm diameter Zircona/Silica beads (Biospec) were added to the samples which were then homogenized in a Biospec Mini Bead Beater for one min. Samples were centrifuged at 4°C at 13,000 × g for 15 min and the supernatant fluids were saved. Total protein concentrations were determined using Bio Rad Protein Assay Dye Reagent Concentrate. After the addition of concentrated Laemmli buffer and heating to 80°C, equal amounts of total protein were subjected to SDS PAGE in a 10% polyacrylamide gel and transferred to PVDF membrane in a Semi-Dry Blot apparatus (BioRad). Membranes were blocked overnight with 5% non-fat milk in TBS at 4°C and probed with D1 (Millipore MAB5290), D2 (Millipore AB5084P) or D5 (Millipore MAB5292) dopamine receptor antibodies at 1:500 with 5% non-fat milk in TBS for 3 h. Internal control GAPDH antibody (Sigma G8795) was incubated at 1:5000 with 5% non-fat milk in TBS for 1 h. After primary antibody incubation, membranes were washed 3 times with 0.05% Tween-20 in Tris-buffered saline (TBS-T) for 10 min each wash. Secondary antibodies linked to HRP were diluted 1:8000 (Santa Cruz, L0312) and 1:5000 (Santa Cruz, L1911) with 5% non-fat milk in TBS and incubated with the washed membranes for 2 h at room temperature. Membranes were then washed 3 times with TBS-T as described above and then antibody binding was detected with ECL plus reagent (GE).</p></sec><sec><title>Immunocytochemistry</title><p>Indirect immunofluorescence was used to assess expression of the receptor subtypes in dissociated DRG neurons in primary culture. DRG neurons were fixed with 4% formaldehyde in Hank’s Buffered Saline Solution (HBSS) for 20 min, washed 3 times in PBS, 10 min each, and then permeabilized with 0.2% Nonidet P40 in TBS for 20 minutes. After 3 more washes with TBS, 10 min each, cover slips were blocked for 1 h with blocking solution (10% Goat Serum, 1% BSA in TBS) and then probed with primary dopamine receptor antibody overnight in blocking solution (1:500), followed by secondary antibody (1:1000 Invitrogen Alexa Fluor 594 A11032) with or without FITC-labeled lectin IB4 for 1.5 h at room temperature. Following 3 washes with TBS, the coverslips were rinsed with deionized water and then mounted (Prolong Gold Antifade mounting fluid) on Superfrost Plus microscope slides (Fisher) and imaged with a confocal microscope (Olympus Fluoview1000).</p><p>For fixed tissue specimens, P14 to P28 Sprague Dawley rats were euthanized and perfused with heparinized saline, followed by 4% formaldehyde in PBS. Dissected specimens were then post fixed for 30 min in 4% formaldehyde in PBS before being transferred to a solution of 30% sucrose in PBS for overnight incubation at 4°C. Fixed tissues were embedded in optimum cutting temperature compound (OCT), frozen on dry ice and then sliced into 20 μm thick sections with a cryostat (Leica) and collected onto Superfrost Plus microscope slides. As soon as dry, the sections were stored up to several days at −20°C. Before the addition of antibody, the sections were permeabilized and blocked with 10% goat serum in TBS with 0.1% Nonidet P40 for 3 h at room temperature. Specimens were then probed with primary antibody overnight in 10% goat serum in TBS at 4°C. They were then washed 3 times in TBS-T and then incubated with fluorescently labeled secondary antibody (1:1000 Invitrogen Alexa Fluor 594 A11032) diluted with 10% goat serum and 1% rat serum in TBS with or without FITC-labeled IB4 for 2 h at room temperature. The wash steps were repeated. The slides were rinsed with deionized water and thoroughly drained. A drop of mounting fluid was placed on each section and coverslips were mounted. All slides were imaged on a confocal microscope (Olympus Fluoview 1000). Images were obtained with a 40× oil objective and 1.4 NA lens. All images were exported as TIFF files to Image J, which was used to process the final images.</p></sec><sec><title>Cell classification</title><p>Small DRG neurons (cell diameter ≤ 25 μm) were chosen for recording. At the end of each experiment, the recorded DRG neuron was tested for the expression of TRPV1 channels (by testing the sensitivity to 1 μM capsaicin) and classified as nociceptor or non-nociceptor [<xref ref-type="bibr" rid="B87">87</xref>]. As nociceptors are neurochemically and functionally distinct [<xref ref-type="bibr" rid="B69">69</xref>,<xref ref-type="bibr" rid="B70">70</xref>,<xref ref-type="bibr" rid="B88">88</xref>,<xref ref-type="bibr" rid="B89">89</xref>], a further classification was made by testing the ability of TRPV1-positive DRG neurons to bind the isolectin B4 (IB4) FITC conjugate and classified as peptidergic [IB4(−)] or non-peptidergic [IB4(+)]. Overall about 75-80% of small DRG neurons tested were sensitive to capsaicin and only those were included in this study. Of the neurons sensitive to capsaicin, about 88% were also IB4(+). Small DRG neurons with these properties are considered to be nociceptive neurons [<xref ref-type="bibr" rid="B67">67</xref>-<xref ref-type="bibr" rid="B70">70</xref>].</p></sec><sec><title>Electrophysiology</title><p>Whole-cell recordings were made with a Multiclamp 700B amplifier (Molecular Devices, Sunnyvale, CA). Patch pipettes were pulled from borosilicate glass (100 μl microcapillaries, VWR, South Plainfield, NJ) or A-M Systems glass 8250 (A-M Systems, Sequim, WA) using a Sutter P97 puller (Sutter Instrument, Novato, CA). The resistance of the patch pipette was 1.8-2.5 MΩ when filled with the standard internal solution. The tips of the patch pipettes were wrapped with Parafilm to reduce pipette capacitance. In whole-cell mode the capacity current was reduced by using the amplifier circuitry. Series resistance was 3.2 ± 0.8 MΩ (n = 122) when the patch pipette was filled with Cs-based internal solution for voltage clamp experiments, and 5.0 ± 1.4 MΩ (n = 27) when the patch pipette was filled with K-based internal solution for current clamp experiments. To reduce voltage errors, 70-80% of series resistance compensation was applied. In voltage clamp experiments a Cs-based internal solution was used to block outward currents through potassium channels. This solution contained (in mM): 125 CsCl, 10 NaCl, 2 MgCl<sub>2</sub>, 10 EGTA, 10 HEPES, 14 Tris-creatine PO<sub>4</sub>, 4 Mg-ATP, and 0.3 Na-GTP, pH = 7.2 with CsOH. In the experiments in Figure <xref ref-type="fig" rid="F5">5</xref>, 20 mM CsCl were replaced with 20 mM Tetraethylammonium-Cl to block outward current at depolarized potentials. In current clamp experiments a K-based internal solution was used. This solution contained (in mM): 135 K-methanesulfonate, 10 NaCl, 2 MgCl<sub>2</sub>, 0.1 EGTA, 10 HEPES, 14 Tris-creatine PO<sub>4</sub>, 4 Mg-ATP, and 0.3 Na-GTP, pH = 7.2 with KOH. Seals were formed in Tyrode’s solution containing (in mM): 151 NaCl, 2.5 KCl, 2 CaCl<sub>2</sub>, 10 HEPES, 13 glucose, pH = 7.4 with NaOH. After the whole-cell configuration had been established, cells were lifted off in front of an array of gravity-fed quartz flow pipes (I.D. 320 μm) that allowed rapid (<1 sec) exchange of external solutions. To isolate the TTX-R sodium current, the Tyrode’s solution was supplemented with 300 nM TTX and 30 μM CdCl<sub>2</sub> in order to block the TTX-S sodium current and calcium current, respectively. The concentration of TTX used (300 nM) is sufficient to fully block the TTX-S sodium current while sparing the TTX-R sodium current, which requires about 40 μM for half-block [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B90">90</xref>]. Similarly, the amount of Cd<sup>2+</sup> used (30 μM) to block the calcium current should have only minimal effect on the TTX-R sodium current [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B91">91</xref>,<xref ref-type="bibr" rid="B92">92</xref>].</p><p>Expression of TRPV1 channels in small DRG neurons was tested at the end of each experiment by challenging the neuron with 1 μM capsaicin and neurons were classified as TRPV1 positive if the inward current activated by capsaicin at – 80 mV reached at least −500 pA.</p></sec><sec><title>Data acquisition and analysis</title><p>Currents and voltages were controlled and sampled using a Digidata 1440 A interface and pClamp 10 software (Molecular Devices, Sunnyvale, CA). Current or voltage signals were filtered at 10 kHz (−3 dB, 4-pole Bessel) and digitized at 50 kHz. Analysis was performed using pClamp 10 and Igor Pro (version 6, Wavemetrics, Lake Oswego, OR), using DataAccess (Bruxton, Seattle, WA) to import pClamp files in Igor. Small DRG neurons (diameter 18 to 25 μm) were initially selected by measuring the diameter from images captured to computer by a CCD camera Oly-150 (Olympus Imaging America Inc., Center Valley, PA) using a video acquisition card (dP dPict Imaging, Inc., Indianapolis, IN). A more accurate measurement of cell diameter was obtained from measurements of whole-cell capacitance assuming a membrane capacitance of 1 μF/cm<sup>2</sup> and spherical shape. Cell capacitance was measured by integrating the average of 5–10 current responses to a – 5 mV step from −80 mV filtered at 10 KHz and acquired at 50 KHz. In Figure 5, conductance <italic>(G)</italic> was measured as <italic>G = I/(V-V</italic><sub><italic>res</italic></sub><italic>)</italic>, where <italic>I</italic> is the peak current, <italic>V</italic> is the voltage, and <italic>Vres</italic> is the reversal potential for sodium current. <italic>V</italic><sub><italic>res</italic></sub> (67.4 ± 2.7 mV, n = 13) was measured with series of steps from −100 to + 100 mV, increments of 10 mV. <italic>G</italic> is plotted normalized to <italic>G</italic><sub><italic>max</italic></sub>, the peak conductance. The lines are best fits to the Boltzmann function: <italic>1/(1 + exp[−(V-V</italic><sub><italic>1/2</italic></sub><italic>)/k]),</italic> where <italic>V</italic> is the step membrane potential in millivolts, <italic>V</italic><sub><italic>1/2</italic></sub> is the half-maximal voltage in millivolts, and <italic>k</italic> is the slope factor in millivolts. Sodium channels availability was determined by using 500 msec prepulses from −120 to + 30 mV followed by a test pulse to −10 mV. The test pulse current is normalized to its maximal value. Solid lines are best fits to the Boltzmann function: <italic>1/(1 + exp[(V-V</italic><sub><italic>1/2</italic></sub><italic>)/k])</italic>, where <italic>V</italic> is the prepulse membrane potential, <italic>V</italic><sub><italic>1/2</italic></sub> is the half-maximal voltage, and <italic>k</italic> is the slope factor in millivolts. In current clamp experiments, reported voltages were corrected for the – 8 mV junction potential between the potassium methanesulfonate-based internal solution and the Tyrode’s solution present when zeroing pipette current. The junction potential was measured using a flowing 3 M KCl bridge [<xref ref-type="bibr" rid="B93">93</xref>]. Action potentials were elicited by injecting 0.8 sec ramps of current to 1 nA [<xref ref-type="bibr" rid="B56">56</xref>]. Usually action potential firing began within 150–200 msec. The peak of the first action potential was quite positive (average + 38 mV), but gradually decreased during current injection until only oscillations were observed. Only action potentials that had a peak ≥ 0 mV and amplitude (from peak to trough) of ≥ 30 mV were included in the analysis. All experiments were carried out at room temperature (22 ± 2°C). Data are reported as Mean ± SD. Statistical differences between data sets were analyzed using the t-test or one-way ANOVA, followed by post-hoc Dunnett analysis (for the data sets in Figure <xref ref-type="fig" rid="F3">3</xref>B). Differences were considered significant at p < 0.05 or *p < 0.01.</p></sec></sec><sec><title>Abbreviations</title><p>DRG: Dorsal root ganglia; TTX: Tetrodotoxin; TTX-R: Tetrodotoxin-resistant; TTX-S: Tetrodotoxin-sensitive; TH: Tyrosine hydroxylase; EEDQ: Ethoxycarbonyl-2-ethoxy-1,2-dihydroquinoline.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>WG performed research and analyzed data. ES analyzed data. MR performed research and analyzed data. MP designed research, performed research, analyzed data, and wrote the paper. All authors read and approved the final manuscript.</p></sec>
Mutated Ca<sub>V</sub>2.1 channels dysregulate CASK/P2X3 signaling in mouse trigeminal sensory neurons of R192Q Cacna1a knock-in mice	<sec><title>Background</title><p>ATP-gated P2X3 receptors of sensory ganglion neurons are important transducers of pain as they adapt their expression and function in response to acute and chronic nociceptive signals. The present study investigated the role of calcium/calmodulin-dependent serine protein kinase (CASK) in controlling P2X3 receptor expression and function in trigeminal ganglia from <italic>Cacna1a</italic> R192Q-mutated knock-in (KI) mice, a genetic model for familial hemiplegic migraine type-1.</p></sec><sec><title>Results</title><p>KI ganglion neurons showed more abundant CASK/P2X3 receptor complex at membrane level, a result that likely originated from gain-of-function effects of R192Q-mutated Ca<sub>V</sub>2.1 channels and downstream enhanced CaMKII activity. The selective Ca<sub>V</sub>2.1 channel blocker ω-Agatoxin IVA and the CaMKII inhibitor KN-93 were sufficient to return CASK/P2X3 co-expression to WT levels. After CASK silencing, P2X3 receptor expression was decreased in both WT and KI ganglia, supporting the role of CASK in P2X3 receptor stabilization. This process was functionally observed as reduced P2X3 receptor currents.</p></sec><sec><title>Conclusions</title><p>We propose that, in trigeminal sensory neurons, the CASK/P2X3 complex has a dynamic nature depending on intracellular calcium and related signaling, that are enhanced in a transgenic mouse model of genetic hemiplegic migraine.</p></sec>	<contrib contrib-type="author" id="A1"><name><surname>Gnanasekaran</surname><given-names>Aswini</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>aswini.anibio8@gmail.com</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Bele</surname><given-names>Tanja</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>tanja.bele@ung.si</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Hullugundi</surname><given-names>Swathi</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>swathihk@sissa.it</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Simonetti</surname><given-names>Manuela</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I5">5</xref><email>manuela.simonetti@pharma.uni-heidelberg.de</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Ferrari</surname><given-names>Michael D</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>m.d.ferrari@lumc.nl</email></contrib><contrib contrib-type="author" id="A6"><name><surname>van den Maagdenberg</surname><given-names>Arn MJM</given-names></name><xref ref-type="aff" rid="I2">2</xref><xref ref-type="aff" rid="I3">3</xref><email>maagdenberg@lumc.nl</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Nistri</surname><given-names>Andrea</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>nistri@sissa.it</email></contrib><contrib contrib-type="author" corresp="yes" id="A8"><name><surname>Fabbretti</surname><given-names>Elsa</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I4">4</xref><email>Elsa.Fabbretti@ung.si</email></contrib>	Molecular Pain	<sec><title>Background</title><p>P2X3 receptors are predominantly expressed on sensory ganglion neurons where they play an important role in transducing pain signals [<xref ref-type="bibr" rid="B1">1</xref>]. A major property of these receptors is the ability to rapidly adapt their function to extracellular milieu changes by trafficking-mediated receptor redistribution, by modulation of receptor function through intracellular kinases, or by interaction with specific scaffold proteins [<xref ref-type="bibr" rid="B2">2</xref>-<xref ref-type="bibr" rid="B5">5</xref>]. We recently reported that under basal conditions P2X3 receptors are strongly associated with the multifunction scaffold protein calcium/calmodulin-dependent serine protein kinase (CASK) [<xref ref-type="bibr" rid="B6">6</xref>]. In the present study we investigated whether the CASK/P2X3 complex was altered and functionally linked to sensitization of P2X3 receptors in transgenic knock-in (KI) mice exhibiting a gain-of-function phenotype of voltage-gated Ca<sub>V</sub>2.1 (P/Q-type) calcium channels, due to a R192Q missense mutation in the channel α1 subunit that causes familial hemiplegic migraine type 1 (FHM-1) [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B8">8</xref>]. Using this KI mouse model, we previously identified multiple Ca<sub>V</sub>2.1 channel interactors (calcineurin, Cdk5 and CaMKII) that modulate P2X3 receptor function in trigeminal sensory neurons [<xref ref-type="bibr" rid="B9">9</xref>-<xref ref-type="bibr" rid="B12">12</xref>]. In particular, enhanced P2X3 receptor-mediated responses were found in KI neurons that depend on constitutive activation of CaMKII and are reversed by the selective Ca<sub>V</sub>2.1 channel blocker or by the CaMKII inhibitor [<xref ref-type="bibr" rid="B9">9</xref>]. Previous studies showed that CASK is associated with calcium channels [<xref ref-type="bibr" rid="B13">13</xref>-<xref ref-type="bibr" rid="B15">15</xref>] and, thus, provide the rational to explore if the R192Q mutation in KI mice influences CASK/P2X3 assembly and function. The present study aimed at testing, with molecular biology and electrophysiological methods, the properties of the CASK/P2X3 receptor complex in this mouse model expressing gain-of-function of Ca<sub>V</sub>2.1 channels, using primary cultures of trigeminal ganglia that fully retain the basal characteristics of the CASK/P2X3 complex <italic>in vivo</italic>[<xref ref-type="bibr" rid="B6">6</xref>].</p></sec><sec sec-type="results"><title>Results</title><sec><title>The CASK/P2X3 receptor complex is abundantly expressed in KI ganglia and is modulated by Ca<sup>2+</sup> influx</title><p>In order to study the effects of CASK on P2X3 receptors expressed in WT and KI ganglia, we first compared CASK/P2X3 complex levels in ganglion extracts. Immunoprecipitation experiments showed that the complex was significantly more abundant in KI than in WT samples (Figure  <xref ref-type="fig" rid="F1">1</xref>A; p = 0.038, n = 5). A significant increase (n = 5, p = 0.005) in CASK associated with cell membrane fractions was observed in KI tissue (Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Figure S1A), although total CASK lysate preparations did not show any difference between WT or KI samples (Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Figure S1B). Further experiments concerning the specificity of the CASK/P2X3 complex, based on immunoprecipitating CASK first and then performing western blotting with P2X3 antibodies, validated our previous findings [<xref ref-type="bibr" rid="B6">6</xref>] and are included in Additional file <xref ref-type="supplementary-material" rid="S2">2</xref>: Figure S2A, B.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>CASK/P2X3 complex in KI trigeminal neurons. A</bold>, Example of immunopurified P2X3 from trigeminal ganglia probed with anti-CASK antibodies reveals more abundant CASK/P2X3 complex in KI ganglion cultures than in WT ones. Histograms quantify this effect (n = 5, p = 0.038). Ab, indicate signal from unrelated antibody. <bold>B</bold>, CASK/P2X3 co-immunoprecipitation experiments from ganglion cultures demonstrate that CASK/P2X3 complex is sensitive to ω-Agatoxin IVA treatment (400 nM, overnight; Aga). Note the more intense CASK/P2X3 signal in KI samples as seen in <bold>A</bold>. P2X3 immunoprecipitation input and β-Tubulin expression shown as gel loading controls. <bold>C</bold>, CASK/P2X3 complex is sensitive to KN-93 pre-incubation (5 μM, 90 min). <bold>D</bold>, Examples of anti-phospho-CaMKII T286 Western immunoblots of extracts from WT and KI trigeminal cultures in control conditions and after CASK silencing. β-Tubulin is shown as loading control. Histograms quantify the effect (#, p = 0.05; p = 0.03 for WT and p = 0.025 for KI; n = 3).</p></caption><graphic xlink:href="1744-8069-9-62-1"/></fig><p>In analogy to its effect on other receptors (e.g. NMDA receptors; [<xref ref-type="bibr" rid="B15">15</xref>]), CASK might exert a role in the process of P2X3 receptor export to surface membranes. In fact, pulled-down biotinylated surface P2X3 receptors showed co-purification with intracellular CASK (Additional file <xref ref-type="supplementary-material" rid="S3">3</xref>: Figure S3), supporting the view that CASK/P2X3 complexes are membrane-bound. In these biotinylation experiments, no difference was observed in the levels of surface membrane CASK in WT and KI samples (Additional file <xref ref-type="supplementary-material" rid="S3">3</xref>: Figure S3).</p><p>We further explored whether the origin of the stronger CASK/P2X3 association in KI samples could be directly linked to the Ca<sub>V</sub>2.1 R192Q gain-of-function and concomitant higher CaMKII activation [<xref ref-type="bibr" rid="B9">9</xref>]. To this end, we used selective pharmacological tools to specifically block either Ca<sub>V</sub>2.1 channel function or apply inhibitors of downstream CaMKII activity. Trigeminal ganglion cultures were treated with the selective Ca<sub>V</sub>2.1 channel blocker ω-Agatoxin IVA (400 nM, 15 h) and then tested for CASK/P2X3 protein interaction by immunoprecipitation. Immunoprecipitation experiments with purified P2X3 receptors and CASK expression (Figure  <xref ref-type="fig" rid="F1">1</xref>B) confirmed that CASK/P2X3 receptor interaction was stronger in KI cultures and that it was significantly reversed by ω-Agatoxin IVA application. Interestingly, CASK/P2X3 complex in WT culture was not affected by the channel blocker suggesting that larger CASK/P2X3 interaction in KI neurons was likely due to the enhanced Ca<sub>V</sub>2.1 channel activity.</p><p>Likewise, after cultures treatment with the CaMKII inhibitor KN-93 (5 μM, 90 min, Figure  <xref ref-type="fig" rid="F1">1</xref>C), the CASK/P2X3 interaction was significantly inhibited in KI, but not in WT (Figure  <xref ref-type="fig" rid="F1">1</xref>C). In addition, after RNA silencing of CASK expression [<xref ref-type="bibr" rid="B6">6</xref>], there was a significant decrease in CaMKII phosphorylation (Figure  <xref ref-type="fig" rid="F1">1</xref>D).</p></sec><sec><title>P2X3 expression and function after siCASK in WT and KI ganglion cultures</title><p>Our recent findings [<xref ref-type="bibr" rid="B6">6</xref>] that showed how siCASK significantly lowered P2X3 expression in trigeminal ganglion cultures, have been further validated in the present study in which no difference between WT and KI cultures was observed as a consequence of siCASK (Figure  <xref ref-type="fig" rid="F2">2</xref>A). To further explore functional consequence of CASK/P2X3 complex in the KI model, patch clamp experiments were carried out (Figure  <xref ref-type="fig" rid="F2">2</xref>B). Sample P2X3 receptor currents elicited by pulse application of the selective agonist α,β-methylene-adenosine-5′-triphosphate (α,β-meATP; 10 μM) were clearly smaller after siCASK silencing, but proportionally similar in WT and KI neurons. As expected [<xref ref-type="bibr" rid="B9">9</xref>], KI neuronal currents were constitutively larger than WT ones (Figure  <xref ref-type="fig" rid="F2">2</xref>B).</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Effect of siCASK silencing in KI and WT neurons. A</bold>, Example of Western immunoblotting for CASK, P2X3 and β-tubulin, in control and after siCASK silencing. Note the significant decrease in P2X3 receptor expression in both WT and KI samples. Histograms show significant reduction of P2X3 protein expression in silenced samples (p = 0.026 for WT and p = 0.04 for KI; n = 5, number of experiments). <bold>B</bold>, representative examples of currents induced by α,β-meATP (α,β, black bar, 10 μM) from WT or KI trigeminal neurons in control or after siCASK. Histograms show quantification of mean current amplitude values (WT: 230 ± 19 pA or 170 ± 13 pA; KI: 296 ± 92 pA or 225 ± 16 pA for control or siCASK respectively). # p = 0.036, p = 0.02 for WT and p = 0.019 for KI; n = 15–23 cells.</p></caption><graphic xlink:href="1744-8069-9-62-2"/></fig></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>The principal finding of the present study is the demonstration that the R192Q missense mutation in the α1 subunit of Ca<sub>V</sub>2.1 calcium channels, that confers gain of function to these channels, is associated with more abundant CASK/P2X3 complex in trigeminal sensory neurons with a mechanism clearly dependent on Ca<sub>V</sub>2.1 channel function and CaMKII activation. This result outlines a molecular mechanism whereby P2X3 receptors are retained and display stronger activity in KI neurons and can, therefore, contribute to sensitization to painful stimuli.</p><p>Our former studies have shown that P2X3 receptors of trigeminal sensory neurons are under the influence of endogenous kinases such as Csk and Cdk5, which regulate the basal operational activity of these receptors [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B16">16</xref>]. We have recently observed a new member of this class of regulators, namely CASK, which strongly controls P2X3 receptor expression and function at the plasma membrane of mouse trigeminal neurons under physiological conditions [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B17">17</xref>]. The present report shows that CASK is more associated with P2X3 receptors in KI ganglia and in culture, and that this association is largely dependent on Ca<sub>V</sub>2.1 calcium channel and CAMKII activity as their specific pharmacological blockers reversed the effect. When CASK expression was inhibited by siRNA silencing, the P2X3 receptor expression and function fell to similar levels in WT and KI neurons. In biotinylation experiments, since no difference was observed in CASK fractions from WT or KI surface membrane samples, these experiments are, therefore, consistent with a differential distribution of P2X3 receptors to lipid raft compartments in KI ganglia [<xref ref-type="bibr" rid="B3">3</xref>]. When CASK was blocked, the residual P2X3 receptor currents were approximately one-third smaller, providing a first estimate about the extent of the role of CASK in regulating P2X3 receptor activity. Such data suggest that the stronger CASK/P2X3 association in KI cells was an important factor for the observed up-regulation of P2X3 receptors.</p><p>Figure  <xref ref-type="fig" rid="F3">3</xref> summarizes an idealized scheme of how we envisage the interaction between certain neuronal proteins affecting P2X3 receptor responsiveness. In WT neurons, only a fraction of P2X3 receptors is thought to be associated with CASK, which ensures their stability at the plasma membrane level [<xref ref-type="bibr" rid="B6">6</xref>]. Under these conditions, Ca<sub>V</sub>2.1 channel activity provides only a relatively minor contribution to Ca<sup>2+</sup> influx [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B19">19</xref>] and P2X3 receptors are regulated only to limited extent by CASK (Csk and Cdk5 are not depicted here). The hyper-functional Ca<sub>V</sub>2.1 channels in KI (that require a smaller membrane depolarization to reach the activation threshold [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B20">20</xref>]) induce stronger Ca<sup>2+</sup> influx and CaMKII activity [<xref ref-type="bibr" rid="B9">9</xref>], improving the stability of CASK/P2X3 receptor complexes and their activity at plasma membrane level. Furthermore, these processes probably synergize and add to the already larger release of extracellular ATP by KI ganglia [<xref ref-type="bibr" rid="B21">21</xref>], thereby further facilitating P2X3 responses and thus contributing to the process of sensitization of P2X3 receptors [<xref ref-type="bibr" rid="B9">9</xref>].</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Proposed mechanism of action of CASK/P2X3 complex in trigeminal sensory neurons.</bold> Schematic representation of CASK, P2X3, and Ca<sub>V</sub>2.1 channel protein interactions at the plasma membrane of trigeminal sensory neurons. <bold>A</bold>, In WT ganglion cells, only a fraction of P2X3 receptors is thought to be associated with CASK that ensures their stable expression. Under these conditions, Ca<sub>V</sub>2.1 channels provide only a minor contribution to Ca<sup>2+</sup> influx, and the dynamic CASK/P2X3 association ensures physiological responses and P2X3 receptors turn-over [<xref ref-type="bibr" rid="B6">6</xref>]. <bold>B</bold>, In KI ganglia, the hyperfunctional Ca<sub>V</sub>2.1 channels result in a stronger Ca<sup>2+</sup> influx, larger CaMKII activity, and more abundant CASK/P2X3 receptor complexes with improved stability of P2X3 receptors. Sub-membrane vesicle-associated CASK, potentially associated with P2X3 receptor trafficking, is also indicated. Larger release of extracellular ATP [<xref ref-type="bibr" rid="B20">20</xref>] facilitates P2X3 responses and may contribute to the process of sensitization of P2X3 receptors.</p></caption><graphic xlink:href="1744-8069-9-62-3"/></fig><p>In conclusion, the present data suggest that CASK could play the role of scaffold protein linking different membrane and sub-membrane molecules (including other channels) to elicit further downstream signaling. Since CASK is involved at presynaptic level to modulate synaptic transmitter release [<xref ref-type="bibr" rid="B22">22</xref>], it is feasible that this anchoring mechanism may also occur at central level and shows distinct alterations in chronic pain.</p></sec><sec sec-type="materials\|methods"><title>Materials and methods</title><sec><title>Tissue culture</title><p>Ca<sub>V</sub>2.1 R192Q KI and WT mice [<xref ref-type="bibr" rid="B8">8</xref>] were euthanized by slowly raising levels of CO<sub>2</sub> in accordance with the Italian Animal Welfare Act and approved by SISSA Ethical Committee. Trigeminal ganglia primary cultures were obtained as previously described [<xref ref-type="bibr" rid="B6">6</xref>]. In specific experiments ω-Agatoxin IVA (400 nM; 15 h; Sigma, Milan, Italy) or CaMKII inhibitor KN-93 (5 μM; Sigma) were added to the culture medium for the indicated time [<xref ref-type="bibr" rid="B9">9</xref>]. CASK siRNA experiments were performed as previously reported [<xref ref-type="bibr" rid="B6">6</xref>].</p></sec><sec><title>Western blotting and immunoprecipitation</title><p>Protein extracts and immunoprecipitation experiments and membrane protein biotinylation were performed as previously described [<xref ref-type="bibr" rid="B6">6</xref>]. P2X3 receptors were solubilized in 10 mM Tris–HCl (pH 7.5), 150 mM NaCl, 20 mM EDTA, and 1% Triton-X-100 plus protease inhibitors (Roche Products, Welwyn Garden City, UK) as previously reported [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B6">6</xref>]. The non-ionic detergent n-octyl-β-D-glucoside (ODG; [<xref ref-type="bibr" rid="B9">9</xref>]) was not used. For western blot quantification, signals expressed in optical density absolute units (AU) were measured with digital imaging system (UVITEC, Cambridge, UK).</p></sec><sec><title>Patch clamp recordings</title><p>Currents from siRNA-treated cultures (72 h) were recorded under whole cell voltage clamp mode at holding potential of -65 mV after correcting for the liquid junction potential [<xref ref-type="bibr" rid="B6">6</xref>]. P2X3 receptor synthetic agonist α,β-meATP (10 μM; Sigma) was applied (2 s) using a fast superfusion system (RSC-200; BioLogic Science Instruments, Claix, France).</p></sec><sec><title>Data analysis</title><p>Data are presented as mean ± standard error. Statistical significance was evaluated with paired Student’s <italic>t</italic>-test or Mann–Whitney rank sum test using OriginPro 7.5 (OriginLab, Northampton, MA, USA).</p></sec></sec><sec><title>Abbreviations</title><p>α: β-meATP α,β-methyleneATP; CaMK: Calcium/calmodulin-dependent kinase; CASK: Calcium/calmodulin-dependent serine kinase; TG: Trigeminal ganglia.</p></sec><sec><title>Competing interests</title><p>The authors state that the content of this article does not create any competing interest.</p></sec><sec><title>Authors’ contributions</title><p>AG carried out biochemical studies and statistical analysis, TB performed further data collection, SH carried out functional studies and statistical analysis, MF, AM provided animal model; EF, MS conceived of the study, EF, AN design and coordination; AG, AM, AN, EF draft the manuscript. All authors read and approved the final manuscript.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional file 1: Figure S1</title><p><bold>A</bold>, <bold>B</bold> Example of a Western blot experiment of total trigeminal ganglia extracts <bold>(A)</bold> or total membrane or ganglia samples <bold>(B)</bold> from WT and KI mice tested with anti-CASK antibodies. Histogram quantifications show no difference of total CASK expression in WT and KI (n = 3; p > 0.05), and a significant enrichment of total membrane-associated CASK expression in KI vs WT (n = 5, p = 0.005). Actin was used as gel loading control. </p></caption><media xlink:href="1744-8069-9-62-S1.pdf"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S2"><caption><title>Additional file 2: Figure S2</title><p><bold>A</bold>, Western blot experiment of trigeminal ganglia extracts after immunoprecipitation with anti-CASK and revealed with anti-CASK or anti-P2X3 antibodies. β-Actin signals quantify immunoprecipitation input. <bold>B</bold>, Membrane protein biotinylation experiments of trigeminal ganglia cultures in control and after siP2X3, analysed with western blot and probed with anti-CASK or anti-P2X3 antibodies, as indicated. Signals from total extracts (left) and streptavidin pull-down (right) of biotinylated samples are shown. No difference in membrane-bound CASK after siP2X3 is found β-Actin is used as gel loading control. Note lack of β-Actin in pull-down samples. </p></caption><media xlink:href="1744-8069-9-62-S2.pdf"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S3"><caption><title>Additional file 3: Figure S3</title><p>Membrane protein biotinylation experiments of WT and KI trigeminal ganglia cultures in control and after siP2X3, revealed with western blot and probed with anti-CASK and anti-P2X3 antibodies. Signals from total extracts (left) and streptavidin pull-down (right) of biotinylated samples are shown. β-Actin is used as gel loading control. Histograms quantify larger surface P2X3 receptors in trigeminal ganglia cultures from KI mice and no changes in surface-associated CASK between WT and KI samples (n = 3, p < 0.05). </p></caption><media xlink:href="1744-8069-9-62-S3.pdf"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
Epidemiology, management, complications and costs associated with type 2 diabetes in Brazil: a comprehensive literature review	<sec><title>Background</title><p>With an estimated 74% of all deaths attributable to non-communicable diseases (NCDs) in 2010, NCDs have become a major health priority in Brazil. The objective of the study was to conduct a comprehensive literature review on diabetes in Brazil; specifically: the epidemiology of type 2 diabetes, the availability of national and regional sources of data (particularly in terms of direct and indirect costs) and health policies for the management of diabetes and its complications.</p></sec><sec><title>Methods</title><p>A literature search was conducted using PubMed to identify articles containing information on diabetes in Brazil. Official documents from the Brazilian government and the World Health Organization, as well as other grey literature and official government websites were also reviewed.</p></sec><sec><title>Results</title><p>From 2006 to 2010, an approximate 20% increase in the prevalence of self-reported diabetes was observed. In 2010, it was estimated that 6.3% of Brazilians aged 18 years or over had diabetes. Diabetes was estimated to be responsible for 278,778 years of potential life lost for every 100,000 people. In 2013, it is estimated that about 7% of patients with diabetes has had one or more of the following complications: diabetic foot ulcers, amputation, kidney disease, and fundus changes. The estimated annual direct cost of diabetes was USD $3.952 billion in 2000; the estimated annual indirect cost was USD $18.6 billion. The two main sources of data on diabetes are the information systems of the Ministry of Health and surveys. In the last few years, the Brazilian Ministry of Health has invested considerably in improving surveillance systems for NCDs as well as implementing specific programmes to improve diagnosis and access to treatment.</p></sec><sec><title>Conclusions</title><p>Brazil has the capacity to address and respond to NCDs due to the leadership of the Ministry of Health in NCD prevention activities, including an integrated programme currently in place for diabetes. Strengthening the surveillance of NCDs is a national priority along with recognising the urgent need to invest in improving the coverage and quality of mortality data. It is also essential to conduct regular surveys of risk factors on a national scale in order to design effective preventive strategies.</p></sec>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Bertoldi</surname><given-names>Andréa D</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>andreadamaso.epi@gmail.com</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Kanavos</surname><given-names>Panos</given-names></name><xref ref-type="aff" rid="I2">2</xref><xref ref-type="aff" rid="I3">3</xref><email>P.G.Kanavos@lse.ac.uk</email></contrib><contrib contrib-type="author" id="A3"><name><surname>França</surname><given-names>Giovanny V A</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I4">4</xref><email>nutrigio@gmail.com</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Carraro</surname><given-names>André</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>andre.carraro@gmail.com</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Tejada</surname><given-names>Cesar Augusto Ovieda</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>cesaroviedotejada@gmail.com</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Hallal</surname><given-names>Pedro C</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>prchallal@gmail.com</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Ferrario</surname><given-names>Alessandra</given-names></name><xref ref-type="aff" rid="I2">2</xref><xref ref-type="aff" rid="I3">3</xref><email>A.Ferrario@lse.ac.uk</email></contrib><contrib contrib-type="author" id="A8"><name><surname>Schmidt</surname><given-names>Maria Inês</given-names></name><xref ref-type="aff" rid="I6">6</xref><email>mischmidt49@gmail.com</email></contrib>	Globalization and Health	<sec><title>Background</title><p>Brazil is an upper middle-income country with a population of 190,755,799 inhabitants [<xref ref-type="bibr" rid="B1">1</xref>] and a per capita gross domestic product of USD $ 10,993 (current exchange rate) in 2011. With a land area covering 47% of Latin America [<xref ref-type="bibr" rid="B2">2</xref>], Brazil has marked regional inequalities in terms of climate, social development, income and other indicators.</p><p>Following democratisation of the country from 1994 onwards, Brazil has experienced a period of economic growth, which allowed the implementation of social development policies [<xref ref-type="bibr" rid="B3">3</xref>]. This has led to slow but stable improvements in social indicators, particularly reductions in poverty and in regional inequalities. In the 70s and early 80s, Brazil underwent a period of social mobilisation in which people campaigned for basic rights, including universal health care access. The demand for greater decentralisation of public resources led to an increase in the budget of cities and states. These factors contributed to the implementation of the Brazilian Unified Health System (SUS - <italic>Sistema Único de Saúde</italic>) in 1990 [<xref ref-type="bibr" rid="B4">4</xref>].</p><p>SUS is intended to provide healthcare free of charge to the whole Brazilian population, financed through direct and indirect sources such as tax revenues, social contributions, out-of-pocket spending, and employers’ health-care spending [<xref ref-type="bibr" rid="B5">5</xref>]. It includes primary health care units, hospitals, emergency departments, laboratories and blood centres. In 2006, SUS budget reached around USD $15 billion, which represents 54% of the country’s total health expenses [<xref ref-type="bibr" rid="B6">6</xref>]. Although access has expanded over the years, the increasing demands on SUS have had negative repercussions on the quality of the services delivered and on waiting times in hospitals and emergency departments [<xref ref-type="bibr" rid="B5">5</xref>].</p><p>In 2011, 22% of total health expenditure was spent on the payment of private health insurance [<xref ref-type="bibr" rid="B7">7</xref>]. The proportion of out-of-pocket expenses has continued to rise in spite of the implementation of SUS, from 9% in 1981 to 15% in 2003 and 19% in 2008 [<xref ref-type="bibr" rid="B5">5</xref>]. Out-of-pocket expenses are particularly concerning due to the difficulty in accurately predicting these costs [<xref ref-type="bibr" rid="B8">8</xref>] which can lead to catastrophic health spending. This is a problem affecting up to 16% of all Brazilian families [<xref ref-type="bibr" rid="B8">8</xref>-<xref ref-type="bibr" rid="B11">11</xref>].</p><p>Brazil and various other Latin American countries have undergone rapid demographic, epidemiological and nutritional transitions [<xref ref-type="bibr" rid="B12">12</xref>]. Dietary shifts towards low consumption of fiber and heavy consumption of saturated fatty acids and sugar and sedentary lifestyles are key contributors to the incidence of obesity, type 2 diabetes, and other chronic diseases [<xref ref-type="bibr" rid="B13">13</xref>]. Non-communicable diseases (NCDs) have become a major health priority in Brazil with an estimated 74% of all deaths attributable to NCDs in 2010 [<xref ref-type="bibr" rid="B14">14</xref>]. National estimates indicate that people with diabetes experience a 57% greater risk of death than the general population [<xref ref-type="bibr" rid="B15">15</xref>]. Beyond the health burden, diabetes is also responsible for increased use of health services and increased costs. Between 1999–2001, it was estimated that about 7.4% of all non-pregnancy related admissions to hospitals and 9.3% of all hospital costs in Brazil were attributable to diabetes [<xref ref-type="bibr" rid="B16">16</xref>].</p><p>In the present study we aimed to: (i) identify existing data sources on the prevalence of diabetes and its complications, as well as the direct and indirect costs of diabetes in Brazil; (ii) describe the prevalence of diabetes and its complications - retinopathy, nephropathy, neuropathy, diabetic foot ulcers, amputation, kidney disease, fundus changes, vascular complications; (iii) report evidence on direct and indirect costs; and (iv) review health policies for the management of diabetes and its complications.</p><sec><title>Methodology</title><p>A comprehensive literature search was conducted to identify articles containing information on type 2 diabetes in Brazil. The following PubMed search strategy was used: ("diabetes mellitus" [MeSH Terms] OR ("diabetes" [All Fields] AND "mellitus" [All Fields]) OR "diabetes mellitus" [All Fields] OR "diabetes" [All Fields] OR "diabetes insipidus" [MeSH Terms] OR ("diabetes" [All Fields] AND "insipidus" [All Fields]) OR "diabetes insipidus" [All Fields]) AND ("brazil" [MeSH Terms] OR "brazil" [All Fields]). The search was limited to articles published in Portuguese, English or Spanish between 2000 and October 2011, without any restrictions on the study design or the level (national or regional) at which the data were collected.</p><p>We included all publications providing information on one or more of the following end-points related to diabetes type 2 in Brazil: prevalence and incidence, management (treatment, access, and inequalities), complications (retinopathy, nephropathy, neuropathy, diabetic foot ulcers, amputation, kidney disease, fundus changes, vascular complications) and direct and indirect costs.</p><p>Articles were first screened by title and then by abstract. Full-text of selected publications were retrieved and examined regarding eligibility. Reference lists of the selected articles were scrutinized in order to identify relevant references. Official documents from the Brazilian government and the World Health Organization (WHO) were also examined. In addition, we identified unpublished work in the grey literature through Google, the researchers’ own knowledge and consultations with diabetes experts in Brazil.</p></sec></sec><sec><title>Results and discussion</title><p>We identified 2,699 articles published between 2000 and October 2011. The screening phase enabled us to identify 87 publications, which were retrieved for detailed evaluation. Forty-two publications met the eligibility criteria (Table <xref ref-type="table" rid="T1">1</xref>).</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Literature review</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="center"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Area of diabetes management</bold></th><th align="center"><bold>Number of publications retrieved</bold></th><th align="left"><bold>References</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Prevalence, incidence and mortality<hr/></td><td align="center" valign="bottom">15<hr/></td><td align="left" valign="bottom">[<xref ref-type="bibr" rid="B17">17</xref>-<xref ref-type="bibr" rid="B30">30</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Prevalence and costs of complications<hr/></td><td align="center" valign="bottom">13<hr/></td><td align="left" valign="bottom">[<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B31">31</xref>-<xref ref-type="bibr" rid="B40">40</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Management: treatment, access and inequality<hr/></td><td align="center" valign="bottom">6<hr/></td><td align="left" valign="bottom">[<xref ref-type="bibr" rid="B41">41</xref>-<xref ref-type="bibr" rid="B46">46</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Outcomes<hr/></td><td align="center" valign="bottom">6<hr/></td><td align="left" valign="bottom">[<xref ref-type="bibr" rid="B47">47</xref>-<xref ref-type="bibr" rid="B52">52</xref>]<hr/></td></tr><tr><td align="left">Direct and indirect costs</td><td align="center">7</td><td align="left">[<xref ref-type="bibr" rid="B23">23</xref>,<xref ref-type="bibr" rid="B53">53</xref>-<xref ref-type="bibr" rid="B58">58</xref>]</td></tr></tbody></table></table-wrap><sec><title>Data sources on diabetes in Brazil</title><p>The Ministry of Health has developed a comprehensive surveillance system for NCDs and their risk factors [<xref ref-type="bibr" rid="B17">17</xref>]. For diabetes, data is available on morbidity (Hospital Information Systems, Ambulatory Information System, and Hypertension and Diabetes Registration and Follow-up system), mortality (single cause or multiple causes) and risk factors (routine data collection through surveillance systems and surveys) [<xref ref-type="bibr" rid="B17">17</xref>].</p><p>The Hospital Information System (SIH-SUS) [<xref ref-type="bibr" rid="B17">17</xref>] is a national system that aggregates patient level data on hospital admissions, primary cause of admission, diagnosis, procedures, length of stay and reimbursement by SUS. The system is set up to allow download and tabulation of data at the municipal level. The scope of the system is limited to SUS admissions and does not include any information on admissions covered by private health insurance or paid out-of-pocket. It is estimated that SIH-SUS covers 60% to 70% of all hospital admissions in the country, although with large variations across regions.</p><p>As part of the Ambulatory Information System (SIA-SUS) [<xref ref-type="bibr" rid="B17">17</xref>] information is collected on so-called ‘highly complex procedures’. This includes data on treatment and exams in the areas of nephrology, cardiology, oncology, orthopaedics, ophthalmology among others. From this dataset it is possible to extract relevant information on screening and management of diabetes and its complications. For example, Georg et al. [<xref ref-type="bibr" rid="B59">59</xref>] performed an economic analysis using secondary data from the SIA-SUS (fasting plasma glucose measurement in order to confirm diagnosis of diabetes), aiming to estimate the cost-effectiveness of the screening programme for diabetes mellitus in Brazil.</p><p>The registration and follow-up system for hypertension and diabetes (HiperDia) [<xref ref-type="bibr" rid="B15">15</xref>] is a computerised system restricted to health system units that register prospective information on patients with hypertension and diabetes who are registered with a health unit or primary health care team. Aggregate data and reports are accessible online. This database includes information on the number of patients with hypertension, types 1 and 2 diabetes, the number of patients who are obese, smokers, physically inactive, as well as those diagnosed with other chronic complications (e.g. dyslipidaemia) [<xref ref-type="bibr" rid="B17">17</xref>]. It is estimated that 31.1% of patients with known diabetes in Brazil are registered in the HiperDia System [<xref ref-type="bibr" rid="B15">15</xref>]. However, there are concerns about the quality of the data. A recent study identified inaccuracies and contradictions in the information reported, indicating the need for additional training and more specific clinical and laboratory criteria to enhance identification of diabetes and hypertension-related complications [<xref ref-type="bibr" rid="B60">60</xref>].</p><p>The mortality information system (SIM) collects information on deaths nationwide [<xref ref-type="bibr" rid="B61">61</xref>]. The system includes reliable information on age, gender, place of residence and cause of death classified according to the International Classification of Diseases version 10 (ICD-10). Problems of misclassification regarding cause of death and coverage gaps are known in the north and northeast of the country [<xref ref-type="bibr" rid="B62">62</xref>]. However, even in these regions, major improvements have been documented in the recent years [<xref ref-type="bibr" rid="B17">17</xref>].</p><p>VIGITEL is a surveillance system of risk and protective factors for chronic NCDs through telephone interviews [<xref ref-type="bibr" rid="B63">63</xref>]. It was launched in 2006 in all capitals of the 26 Brazilian states including the Federal District and has been conducted since then on an annual basis. Each annual survey includes around 2,000 participants from each of the 27 capital cities with results weighted for the availability of land lines in each region.</p><p>The National Household Sample Survey (PNAD) provides periodic surveillance data on NCDs nationwide. Reports summarising data by region, by state, and by rural/urban area are accessible online [<xref ref-type="bibr" rid="B64">64</xref>]. The three surveys conducted to date provided information on access to and utilisation of health services in 1998, 2003 and 2008. In addition, the 2008 survey also included information on morbidities caused by chronic diseases, including diabetes [<xref ref-type="bibr" rid="B65">65</xref>].</p><p>The family budget survey (POF) [<xref ref-type="bibr" rid="B66">66</xref>] is a household survey measuring consumption, expenses and income of Brazilian families. Previous survey rounds were conducted in 1974/1975, 1987/1988, 1995/1996, 2002/2003 and 2008/2009. The survey provides information on the cost of treating diabetes, which allows for assessment of the disease’s impact on households budgets, for example [<xref ref-type="bibr" rid="B46">46</xref>].</p><p>The national demographic and health survey (PNDS) is part of the MEASURE DHS project [<xref ref-type="bibr" rid="B67">67</xref>], focusing on women of child-bearing age and children under five in Brazil. The PNDS was first conducted in 1986 and subsequently in 1996 and 2006; however, data on the prevalence of diabetes among women and access to medicines were only collected in 2006 [<xref ref-type="bibr" rid="B68">68</xref>].</p><p>The Brazilian longitudinal study of adult health (ELSA-Brasil) [<xref ref-type="bibr" rid="B69">69</xref>,<xref ref-type="bibr" rid="B70">70</xref>] is a multicentre cohort study funded by the Ministry of Health to investigate diabetes and cardiovascular disease (CVD) incidence, risk factors and complications. The baseline evaluation was completed in December 2010 and included 15,105 civil servants aged 35–74 years from six public universities in the northeast, south and southeast regions of Brazil. Annual telephone interviews are conducted to monitor the health status of each participant enrolled in the baseline [<xref ref-type="bibr" rid="B69">69</xref>].</p></sec><sec><title>Prevalence of diabetes and diabetes-related mortality</title><p>In 2012, the International Diabetes Federation (IDF) estimated the prevalence of diabetes in Brazil to be 10.3% [<xref ref-type="bibr" rid="B71">71</xref>]. In the next paragraphs we summarise evidence on the prevalence of diabetes and diabetes-related mortality since 1986 across different regions in Brazil.</p><p>From 1986 to 1988, a multicentre study on diabetes was conducted in nine Brazilian state capitals, including a sample of 21,847 subjects first screened by fasting capillary glucose (FCG) [<xref ref-type="bibr" rid="B21">21</xref>]. The prevalence of diabetes was estimated at 7.6% among subjects aged 30–69 years. A concerning finding was that 46.5% of the cases were undiagnosed. In addition, out of those who were aware of their diabetes condition, 22.3% were not receiving any type of diabetes treatment. The prevalence of diabetes did not vary according to sex, ethnicity and level of education, but increased markedly with age, from 2.7% among those aged 30–39 years to 17.4% among those aged 60–69 years [<xref ref-type="bibr" rid="B21">21</xref>].</p><p>Since then, several other studies have been conducted with different scopes and methodologies, as summarised in Table <xref ref-type="table" rid="T2">2</xref>. Most of the studies presented are based on self-reported diabetes.</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p><bold>Studies of the prevalence of diabetes in Brazil</bold><sup> <bold>1</bold> </sup></p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="center" valign="middle"><bold>First author (year)</bold></th><th align="center" valign="middle"><bold>Site</bold></th><th align="center" valign="middle"><bold>Year of study</bold></th><th align="center" valign="middle"><bold>Sample size</bold></th><th align="center" valign="middle"><bold>Age group</bold></th><th align="center" valign="middle"><bold>Diabetes prevalence</bold></th><th align="center" valign="middle"><bold>Criteria</bold></th></tr></thead><tbody valign="top"><tr><td align="center" valign="bottom">Goldenberg (1996) [<xref ref-type="bibr" rid="B25">25</xref>]<hr/></td><td align="center" valign="bottom">São Paulo, SP<hr/></td><td align="center" valign="bottom">1986-1988<hr/></td><td align="center" valign="bottom">2,007<hr/></td><td align="center" valign="bottom">30-69 years<hr/></td><td align="center" valign="bottom">4.7%<hr/></td><td align="center" valign="bottom">Self-report<hr/></td></tr><tr><td rowspan="2" align="center" valign="bottom">Malarbi (1992) [<xref ref-type="bibr" rid="B49">49</xref>]<hr/></td><td align="center" valign="bottom">Nine Brazilian<hr/></td><td rowspan="2" align="center" valign="bottom">1988<hr/></td><td rowspan="2" align="center" valign="bottom">21,847<hr/></td><td rowspan="2" align="center" valign="bottom">30-69 years<hr/></td><td rowspan="2" align="center" valign="bottom">7.6%<hr/></td><td rowspan="2" align="center" valign="bottom">OGTT<sup>2</sup> and self-report<hr/></td></tr><tr><td align="center" valign="bottom">state capitals<hr/></td></tr><tr><td align="center" valign="bottom">Torquato (2003) [<xref ref-type="bibr" rid="B30">30</xref>]<hr/></td><td align="center" valign="bottom">Ribeirão Preto, SP<hr/></td><td align="center" valign="bottom">1996-1997<hr/></td><td align="center" valign="bottom">1,473<hr/></td><td align="center" valign="bottom">30-69 years<hr/></td><td align="center" valign="bottom">12.1%<hr/></td><td align="center" valign="bottom">OGTT and self-report<hr/></td></tr><tr><td rowspan="2" align="center" valign="bottom">Passos (2005) [<xref ref-type="bibr" rid="B27">27</xref>]<hr/></td><td rowspan="2" align="center" valign="bottom">Bambuí, MG<hr/></td><td rowspan="2" align="center" valign="bottom">1997<hr/></td><td rowspan="2" align="center" valign="bottom">816 adults and 1,494 elderly<hr/></td><td rowspan="2" align="center" valign="bottom">Adults (18-59 years); elderly (60+ years).<hr/></td><td align="center" valign="bottom">Elderly 14.6%<hr/></td><td rowspan="2" align="center" valign="bottom">FPG<sup>3</sup> and self-report<hr/></td></tr><tr><td align="center" valign="bottom">Adults 2.3%<hr/></td></tr><tr><td align="center" valign="bottom">Dias da Costa (2006) [<xref ref-type="bibr" rid="B24">24</xref>]<hr/></td><td align="center" valign="bottom">Pelotas, RS<hr/></td><td align="center" valign="bottom">2000<hr/></td><td align="center" valign="bottom">1,968<hr/></td><td align="center" valign="bottom">20-69 years<hr/></td><td align="center" valign="bottom">5.6%.<hr/></td><td align="center" valign="bottom">Self-report<hr/></td></tr><tr><td rowspan="2" align="center" valign="bottom">Souza (2003) [<xref ref-type="bibr" rid="B29">29</xref>]<hr/></td><td rowspan="2" align="center" valign="bottom">Campos dos Goytacazes, RJ<hr/></td><td rowspan="2" align="center" valign="bottom">2001<hr/></td><td rowspan="2" align="center" valign="bottom">1,039<hr/></td><td rowspan="2" align="center" valign="bottom">>18 years<hr/></td><td align="center" valign="bottom">6.0%<hr/></td><td rowspan="2" align="center" valign="bottom">FPG<hr/></td></tr><tr><td align="center" valign="bottom">(age-adjusted prevalence)<hr/></td></tr><tr><td align="center" valign="bottom">Mendes (2011) [<xref ref-type="bibr" rid="B26">26</xref>]<hr/></td><td align="center" valign="bottom">São Paulo, SP<hr/></td><td align="center" valign="bottom">2003<hr/></td><td align="center" valign="bottom">872<hr/></td><td align="center" valign="bottom">60+ years<hr/></td><td align="center" valign="bottom">17.9%<hr/></td><td align="center" valign="bottom">Self-report<hr/></td></tr><tr><td rowspan="2" align="center" valign="bottom">Schmidt (2009) [<xref ref-type="bibr" rid="B28">28</xref>]<hr/></td><td align="center" valign="bottom">27 Brazilian<hr/></td><td rowspan="2" align="center" valign="bottom">2006<hr/></td><td rowspan="2" align="center" valign="bottom">54,369<hr/></td><td rowspan="2" align="center" valign="bottom">aged ≥18 years<hr/></td><td rowspan="2" align="center" valign="bottom">5.3%<hr/></td><td rowspan="2" align="center" valign="bottom">Self-report<hr/></td></tr><tr><td align="center" valign="bottom">state capitals<hr/></td></tr><tr><td rowspan="2" align="center">Bosi (2009) [<xref ref-type="bibr" rid="B23">23</xref>]</td><td rowspan="2" align="center">São Carlos, SP</td><td rowspan="2" align="center">2007-2008</td><td rowspan="2" align="center">1,116</td><td rowspan="2" align="center">30-79 years</td><td align="center" valign="bottom">5% and 13.5%<hr/></td><td rowspan="2" align="center">OGTT and fasting capillary glycaemia</td></tr><tr><td align="center">(age-adjusted prevalence)</td></tr></tbody></table><table-wrap-foot><p><sup>1</sup>Data from publications of the Ministry of Health and other institutions were not included in the table.</p><p><sup>2</sup>OGTT: oral glucose tolerance test (old diagnostic criteria for fasting glucose).</p><p><sup>3</sup>FPG: fasting plasma glucose.</p></table-wrap-foot></table-wrap><p>Self-reported prevalence of diabetes has been studied on an annual basis in all state capitals since 2006. As shown in Figure <xref ref-type="fig" rid="F1">1</xref>, within only four years, self-reported prevalence increased from 5.3% in 2006 to 6.3% in 2010. It is not clear whether this increase is due to increased prevalence, increased diagnosis or both.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p>Prevalence of diabetes in Brazil between 2006 and 2010, according to the VIGITEL.</p></caption><graphic xlink:href="1744-8603-9-62-1"/></fig><p>As shown in Figure <xref ref-type="fig" rid="F2">2</xref>, women were more likely than men to report having diabetes, which may reflect their higher utilisation of medical care and therefore increased likelihood of being diagnosed [<xref ref-type="bibr" rid="B63">63</xref>], supporting the argument of increased detection. However, it seems likely that higher incidence of diabetes must also have played a role in increasing the reported prevalence of diabetes, particularly given the parallel increase in the prevalence of obesity epidemics in Brazil [<xref ref-type="bibr" rid="B72">72</xref>].</p><fig id="F2" position="float"><label>Figure 2</label><caption><p>Prevalence of diabetes in Brazil by sex and age groups. VIGITEL, 2010.</p></caption><graphic xlink:href="1744-8603-9-62-2"/></fig><p>Franco et al. [<xref ref-type="bibr" rid="B73">73</xref>] analysed diabetes-relates deaths in São Paulo, including data from 1975 to 1992. Diabetes was mentioned on the certificate of 13,786 deaths (6.8%), and referred as the underlying cause of 2.6% of all deaths. Diabetes was also reported as an associated cause of deaths whose underlying cause was cardiovascular and respiratory diseases, as well as neoplasia.</p><p>Cesse et al. [<xref ref-type="bibr" rid="B18">18</xref>] analysed time trends in diabetes-related mortality and found that mortality increased in most state capitals between 1950 to 2000, while the largest proportional variations were observed in Teresina-PI (55.1%), Recife-PE (27%) and Natal (21.7%). This is consistent with the rapid demographic transition seen in Brazil during this period [<xref ref-type="bibr" rid="B2">2</xref>] as well as with the increased prevalence of diabetes. Mortality figures underestimate the burden of diabetes, since the underlying cause of death (including diabetes) is not accounted in the final cause of death estimates. Coeli et al. [<xref ref-type="bibr" rid="B19">19</xref>] examined 2,974 death certificates of older adults and found that 291 subjects had diabetes as one of the reported causes of death. However, only 150 subjects had diabetes as the underlying cause of death.</p><p>Diabetes was estimated to be responsible for 278,778 years of potential life lost for every 100,000 people [<xref ref-type="bibr" rid="B17">17</xref>]. Disability adjusted life years (DALYs) due to diabetes and its complications were estimated in the five regions of Brazil [<xref ref-type="bibr" rid="B20">20</xref>,<xref ref-type="bibr" rid="B22">22</xref>] – results showed that diabetes was responsible for 5.1% (6.0% among women and 4.4% among men) of the total DALYs in the country.</p></sec><sec><title>Diabetes complications</title><p>It is estimated that about 7% of patients with diabetes had one or more of the following complications: diabetic foot ulcers, amputation, kidney disease, fundus changes [<xref ref-type="bibr" rid="B15">15</xref>]. Detailed information regarding studies on diabetes complication in Brazil is shown in Table <xref ref-type="table" rid="T3">3</xref>.</p><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>Studies on diabetes complications in Brazil</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th rowspan="2" align="center" valign="middle"><bold>First author (year)</bold></th><th rowspan="2" align="center" valign="middle"><bold>City, State</bold></th><th rowspan="2" align="center" valign="middle"><bold>N of patients with type 2 diabetes</bold></th><th colspan="2" align="center" valign="middle"><bold>Data collection</bold><hr/></th><th rowspan="2" align="center" valign="middle"><bold>Complication</bold></th><th rowspan="2" align="center" valign="middle"><bold>Prevalence</bold></th></tr><tr><th align="center" valign="middle"><bold>Place</bold></th><th align="center" valign="middle"><bold>Source of data</bold></th></tr></thead><tbody valign="top"><tr><td rowspan="3" align="center" valign="bottom">Bruno (2000) [<xref ref-type="bibr" rid="B32">32</xref>]<hr/></td><td rowspan="3" align="center" valign="bottom">Porto Alegre, Rio Grande do Sul<hr/></td><td rowspan="3" align="center" valign="bottom">93<hr/></td><td rowspan="3" align="center" valign="bottom">Dialysis centres<hr/></td><td rowspan="3" align="center" valign="bottom">Standardized questionnaire, clinical interview, and review of medical records<hr/></td><td align="center" valign="bottom">Diabetic nephropathy<hr/></td><td align="center" valign="bottom">58.0%<hr/></td></tr><tr><td align="center" valign="bottom">Diabetic retinopathy<hr/></td><td align="center" valign="bottom">85.0%<hr/></td></tr><tr><td align="center" valign="bottom">Peripheral vascular disease<hr/></td><td align="center" valign="bottom">73.0%<hr/></td></tr><tr><td rowspan="4" align="center" valign="bottom">Scheffel (2004) [<xref ref-type="bibr" rid="B37">37</xref>]<hr/></td><td rowspan="4" align="center" valign="bottom">Rio Grande do Sul<hr/></td><td rowspan="4" align="center" valign="bottom">927<hr/></td><td rowspan="4" align="center" valign="bottom">Health care units<hr/></td><td rowspan="4" align="center" valign="bottom">Clinical examination and laboratory tests<hr/></td><td align="center" valign="bottom">Coronary artery disease<hr/></td><td align="center" valign="bottom">36.0%<hr/></td></tr><tr><td align="center" valign="bottom">Diabetic retinopathy<hr/></td><td align="center" valign="bottom">48.0%<hr/></td></tr><tr><td align="center" valign="bottom">Ischemic heart disease<hr/></td><td align="center" valign="bottom">36.0%<hr/></td></tr><tr><td align="center" valign="bottom">Peripheral vascular disease<hr/></td><td align="center" valign="bottom">33.0%<hr/></td></tr><tr><td rowspan="3" align="center" valign="bottom">Tres (2007) [<xref ref-type="bibr" rid="B39">39</xref>]<hr/></td><td rowspan="3" align="center" valign="bottom">Passo Fundo, Rio Grande do Sul<hr/></td><td rowspan="3" align="center" valign="bottom">340<hr/></td><td rowspan="3" align="center" valign="bottom">Outpatient Diabetes Clinic of Hospital São Vicente de Paulo<hr/></td><td rowspan="3" align="center" valign="bottom">Questionnaire and neurological tests<hr/></td><td align="center" valign="bottom">Diabetic nephropathy<hr/></td><td align="center" valign="bottom">29.5%<hr/></td></tr><tr><td align="center" valign="bottom">Diabetic neuropathy.<hr/></td><td align="center" valign="bottom">22.0%<hr/></td></tr><tr><td align="center" valign="bottom">Diabetic retinopathy<hr/></td><td align="center" valign="bottom">28.8%<hr/></td></tr><tr><td align="center">Vieira-Santos (2008) [<xref ref-type="bibr" rid="B40">40</xref>]</td><td align="center">Recife, Pernambuco</td><td align="center">1,374</td><td align="center">Primary health care units</td><td align="center">Medical records</td><td align="center">Diabetic foot</td><td align="center">9.0%</td></tr></tbody></table></table-wrap><p>A study of 1,374 patients with diabetes seen in family health units in Recife, Pernambuco state found a 9% prevalence of diabetic foot [<xref ref-type="bibr" rid="B40">40</xref>]. Routine screening for diabetic foot is limited by the lack of trained podiatrists and appropriate supplies. With the exception of a few treatment centres, most health services, particularly primary health care, do not perform foot screening for patients at high risk of developing complications [<xref ref-type="bibr" rid="B36">36</xref>].</p><p>According to the Brazilian Ministry of Health, diabetic retinopathy (DR) is the leading cause of irreversible blindness in Brazil. Asymptomatic in its early stages, retinopathy evolves over time, affecting the majority of patients who have lived with diabetes for more than 20 years [<xref ref-type="bibr" rid="B21">21</xref>]. It is estimated that 20 to 40% of patients with type 2 diabetes are affected by DR, based on studies among specific groups and restricted areas [<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B74">74</xref>-<xref ref-type="bibr" rid="B76">76</xref>]. The narrow focus of these studies and limited geographical coverage make it difficult to estimate the national prevalence.</p><p>Diabetic nephropathy (DN) is another common and devastating complication in patients with diabetes, with a slightly lower frequency than retinopathy [<xref ref-type="bibr" rid="B31">31</xref>]. Similar to other countries, chronic kidney disease has been an important public health problem in Brazil. It is estimated that at least one third of Brazilians with type 2 diabetes are affected by DN [<xref ref-type="bibr" rid="B34">34</xref>,<xref ref-type="bibr" rid="B35">35</xref>].</p><p>Data available from the High Complexity Procedures Authorisation Subsystem on Renal Replacement Therapy (APAC/TRS) [<xref ref-type="bibr" rid="B17">17</xref>] shows that between 2000 and 2006, 148,284 patients began dialysis treatments (predominantly haemodialysis) in Brazil. The incidence of terminal disease patients starting replacement therapy was estimated to be 119.8/1,000,000 inhabitants/year, varying from 143.6/1,000,000/year in the south of Brazil to 66.3/1,000,000/year in the north of the country [<xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B33">33</xref>]. Hypertension was reported as the leading cause of renal disease (22%), followed by diabetes mellitus (13.8%) and glomerulonephritis (7.2%) [<xref ref-type="bibr" rid="B17">17</xref>]. Undetermined causes were cited frequently (44.8%), indicating the need to improve the quality of the information recorded. Incidence of terminal disease patients starting dialysis increased in patients over 65 years, most likely related to population aging and greater use of renal replacement therapy among the elderly [<xref ref-type="bibr" rid="B33">33</xref>].</p><p>A population based study conducted in all 18 dialysis centres located in the metropolitan area of Porto Alegre between July 1995 and October 1996 followed 111 patients with type 2 diabetes for an average period of 3.6 years. The prevalence of DN was 58% and it was the leading cause of renal disease in 61% of all patients in the follow-up period (63%) [<xref ref-type="bibr" rid="B32">32</xref>].</p><p>Ischaemic heart disease and hypertension are the most frequent cardiovascular diseases in patients with diabetes. In women with diabetes, the protective effects observed for cardiovascular disease in general disappears [<xref ref-type="bibr" rid="B31">31</xref>]. In 2004, a cross-sectional study using a sample of 927 patients with type 2 diabetes treated at three medical centres in Rio Grande do Sul observed a prevalence of coronary artery disease, peripheral vascular disease and hypertension of 36%, 33%, and 73% respectively [<xref ref-type="bibr" rid="B37">37</xref>].</p><p>Regarding neuropathy, it is estimated that the most common form of the disease is distal symmetrical sensory polyneuropathy [<xref ref-type="bibr" rid="B15">15</xref>]. In 2007, a cross-sectional study with 340 patients with type 2 diabetes in Passo Fundo (southern Brazil) found a prevalence of 22% of diabetic peripheral neuropathy [<xref ref-type="bibr" rid="B39">39</xref>].</p><p>Despite the existence of multiple data sources, evidence on the prevalence and incidence of diabetes and its complications at national and regional level is very scarce and originates mainly from surveys. Prevalence data mainly originates from a number of studies that rely on self-reported data, and no study on the incidence of diabetes was found. It seems that there is a missed opportunity to leverage the data available through some of the national databases such as SUS and HiperDia, among others.</p></sec><sec><title>Management of diabetes in Brazil: treatment, access, inequality</title><p>Evaluation of the health care delivered by SUS is still done infrequently, particularly with regard to chronic diseases. A study carried out by Assunção et al. [<xref ref-type="bibr" rid="B41">41</xref>] in 1998/1999 evaluated the structure, process, and outcomes of diabetes treatment in primary health care in Pelotas, in Southern Brazil. Approximately 85% of the physicians in the study reported prescribing a diet plan during their first consultation and 72% prescribed physical activity. In terms of laboratory monitoring of the patients, all physicians requested fasting blood glucose, while only 60% requested glycosylated haemoglobin.</p><p>In 2006, the Ministry of Health published primary health care guidelines [<xref ref-type="bibr" rid="B31">31</xref>] for the management of diabetes at primary care level. The guidelines provide recommendations on diabetes screening and prevention, diagnosis, initial evaluation and basic treatment. Screening is recommended for asymptomatic individuals at higher risk of diabetes according to the following indicators: age >45 years, BMI > 25 Kg/m<sup>2</sup>, waist circumference >102 cm for men and >88 cm for women, family history of diabetes, hypertension (>140/90 mmHg), HDL cholesterol <35 mg/dl and/or triglycerides >150 mg/dl. Recommendations on lifestyle changes, pharmacological treatment, prevention and management of acute and chronic outcomes of diabetes are also provided. Similar guidelines are available for hypertension and prevention of CVD at the primary care level.</p><p>Some studies using regional samples investigated availability, affordability and access to medicines used for the treatment of diabetes. Pinto et al. [<xref ref-type="bibr" rid="B45">45</xref>] analysed medicine prices and availability using WHO/HAI methodology. The study was performed in 2007 in 30 cities in Brazil and found that metformin 500 mg and glibenclamide 5 mg were available in 23% and 93% of public sector facilities respectively.</p><p>In contrast, another study [<xref ref-type="bibr" rid="B42">42</xref>] carried out in six cities in the south of Brazil found total availability of metformin 500 mg to be 100% in the public sector. In terms of affordability, the study found that both metformin and glibenclamide could cost up to two working days of salary for non-skilled workers to purchase a monthly course treatment. A cross-sectional study [<xref ref-type="bibr" rid="B43">43</xref>] evaluating 41 municipalities in South and Northeast Brazil reported that 78.6% of patients with diabetes had access to diabetes medicines. Another study using the same population [<xref ref-type="bibr" rid="B44">44</xref>] looked at access to diabetes medicine among the elderly and found that 95.8% had access to medicines, with the majority of medicines provided by SUS (76.7%).</p><p>A National Survey on Medicine Access and Utilization (PNAUM) started in 2013 and data collection is ongoing [<xref ref-type="bibr" rid="B77">77</xref>]. The aim of this survey is to evaluate the national pharmaceutical policy and whether the policy is achieving its main objective of ensuring high levels of access to medicine for the entire population. It is the first national study exclusively designed to evaluate the result of the current pharmaceutical policy.</p><p>The impact of diabetes on family expenses was investigated in a study using data from POF 2002-2003 [<xref ref-type="bibr" rid="B46">46</xref>]. This study showed that 1.7% of the population purchased at least one medicine for diabetes. The annual average spending for those who acquired one or more medicines for diabetes care amounted to USD $102.81.</p><p>Data from 2004 showed that glycaeted haemoglobin (HbA1c) control (<7.0%) was attained only by 40% of patients with diabetes [<xref ref-type="bibr" rid="B51">51</xref>].</p></sec><sec><title>Diabetes outcomes indicators</title><p>Table <xref ref-type="table" rid="T4">4</xref> identifies indicators on diabetes outcomes available in Brazil. The main sources of data are information systems from the Ministry of Health. Minimal baseline information on each indicator is available.</p><table-wrap position="float" id="T4"><label>Table 4</label><caption><p>Indicators of diabetes outcomes available in Brazil</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Indicator</bold></th><th align="left"><bold>Source</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Percent of persons with diabetes mellitus with a HbA1c tested in last 12 months<hr/></td><td align="left" valign="bottom">Surveys<hr/></td></tr><tr><td align="left" valign="bottom">Percent of persons tested, who have HbA1c value >7.5%<hr/></td><td align="left" valign="bottom">Surveys<hr/></td></tr><tr><td align="left" valign="bottom">Percent of the persons with diabetes mellitus with microalbuminuria tested in last 12 months<hr/></td><td align="left" valign="bottom">Surveys<hr/></td></tr><tr><td align="left" valign="bottom">Percent of those tested with microalbuminuria<hr/></td><td align="left" valign="bottom">Surveys<hr/></td></tr><tr><td align="left" valign="bottom">Percent of the persons with diabetes mellitus with blood pressure measurements in last 12 months<hr/></td><td align="left" valign="bottom">SisHiperDia<hr/></td></tr><tr><td align="left" valign="bottom">Percent of the persons with diabetes mellitus who are smoking<hr/></td><td align="left" valign="bottom">SisHiperDia Vigitel<hr/></td></tr><tr><td align="left" valign="bottom">Percent of persons with diabetes mellitus with BMI ≥ 25 kg/m<sup>2</sup>, BMI ≥ 30 kg/m<sup>2</sup><hr/></td><td align="left" valign="bottom">SisHiperDia Vigitel<hr/></td></tr><tr><td align="left" valign="bottom">Percent of persons with diabetes mellitus with fundus inspection in the last 12 months<hr/></td><td align="left" valign="bottom">SisHiperDia<hr/></td></tr><tr><td align="left" valign="bottom">Percent of those tests, with proliferate retinopathy in the last 12 months<hr/></td><td align="left" valign="bottom">SisHiperDia<hr/></td></tr><tr><td align="left" valign="bottom">Annual incidence of amputations above the ankle in patients diabetes mellitus/100,000 general population<hr/></td><td align="left" valign="bottom">Surveys<hr/></td></tr><tr><td align="left">Annual incidence of myocardial infarction in patients with diabetes mellitus/100,000 general population</td><td align="left">SisHiperDia</td></tr></tbody></table></table-wrap><p>Data from 2004 showed that glycaeted haemoglobin (HbA1c) control (<7.0%) was attained only by 40% of patients with diabetes [<xref ref-type="bibr" rid="B51">51</xref>]. Further, it is estimated that about 7% of individuals with diabetes had one or more of the following complications: diabetic foot ulcers, amputation, kidney disease, fundus changes [<xref ref-type="bibr" rid="B15">15</xref>].</p><p>A multicentre study conducted in five countries, including Brazil, identified that no country has reached the standard for HbA1c or blood pressure set by the American Diabetes Association Diabetes Physician Recognition Programme [<xref ref-type="bibr" rid="B52">52</xref>]. In 2007, a cross-sectional multicentre study conducted in nine Latin American countries (Argentina, Brazil, Chile, Costa Rica, Ecuador, Guatemala, Mexico, Peru, and Venezuela), including a sample of 878 Brazilians aged 18 to 75 years with type 2 diabetes, showed that about 40% of participants had controlled glycosylated haemoglobin (HbA1c <7.0%) [<xref ref-type="bibr" rid="B51">51</xref>].</p><p>Very few studies have been conducted to evaluate the quality of treatment and to measure differences between SUS and privately insured patients. A retrospective cohort study was carried out in Southern Brazil involving 80 patients treated in a SUS outpatient clinic and 277 patients treated at a private clinic. Patients receiving treatment from SUS generally showed worse metabolic control, although only the values of HbA1c and total cholesterol were statistically different between the two groups [<xref ref-type="bibr" rid="B50">50</xref>]. However, due to the small sample size and the regional coverage of this study, these findings are not representative of the whole Brazilian population.</p></sec><sec><title>Costs related to diabetes and its complications</title><p>In 2008, the World Bank estimated that countries such as Brazil, China, India and Russia lose more than 20 million productive life-years due to NCDs annually [<xref ref-type="bibr" rid="B78">78</xref>].</p><p>A study across several Latin and Central American and Caribbean countries [<xref ref-type="bibr" rid="B79">79</xref>] estimated that in 2000, the total annual costs (direct and indirect) of diabetes in Brazil were USD $22.6 billion. Direct costs included medications, hospitalisations, consultations, and treatment for complications and totalled to US $3.952 billion. This represented a direct cost per capita of US $872. Indirect costs included loss of income by permanent and temporary incapacity as well as premature death, and amounted to USD $18.6 billion. Across all twenty-five Latin American and Caribbean countries included in the analysis, Brazil had the highest estimated indirect and direct costs for diabetes among the countries studied.</p><p>Bahia et al. [<xref ref-type="bibr" rid="B54">54</xref>] estimated direct and indirect costs of type 2 diabetes using data collected during 1,000 interviews carried out in 2007 in eight Brazilian cities. The total annual cost per patient was USD $2,108, of which 63.3% were direct costs (USD $1,335) and 36.7% indirect costs (USD $773).</p><p>McLellan et al. [<xref ref-type="bibr" rid="B55">55</xref>] estimated the cost of clinical treatment and hospital expenses to be around USD $710 per patient/year in 2001. This estimate was based on 93 people with diabetes in the city of Piracicaba - São Paulo - hospitalised between March and June 2001, and therefore unlikely to be nationally representative.</p><p>Rosa et al. [<xref ref-type="bibr" rid="B57">57</xref>] calculated expenses for hospitalisation due to diabetes using national data for the period of 1999–2001. It was estimated that the average cost per hospitalisation resulting in patient death is USD $275.27; in comparison to USD $143.45 when hospitalisation did not result in death [<xref ref-type="bibr" rid="B57">57</xref>]. Hospitalisation rates for patients with diabetes have been stable in the past few years, ranging from 65 to 75 per 100,000 inhabitants per year.</p><p>Abegunde et al. [<xref ref-type="bibr" rid="B53">53</xref>] predicted that losses due to reduced productivity at work and the decreased family income as a result of diabetes, heart disease and stroke would lead to an economic loss amounting to USD $4.18 billion from 2006 to 2015 in low and middle-income countries.</p><p>A study using DATASUS data estimated the direct cost of hospitalisation due to diabetes to be USD $362,945,412 in 2000 [<xref ref-type="bibr" rid="B58">58</xref>]. Another study [<xref ref-type="bibr" rid="B56">56</xref>] simulated a hypothetical cohort including 6.48 million participants with type 2 diabetes, based on estimates from the Brazilian Ministry of Health, hospital budgets and expense records in 2008. The estimated annual total cost of hospitalisation was USD $264 million (converted using 2008 rate of exchange US$1 = R$1.64), while the costs related to amputation totalled USD $128 million [<xref ref-type="bibr" rid="B56">56</xref>].</p></sec><sec><title>Health policy related to diabetes</title><p>In 1987, a multicentre study on the prevalence of diabetes and impaired glucose tolerance was conducted in nine Brazilian capitals among adults aged 30–69. This study indicated that half of the individuals with diabetes were not aware of their health condition [<xref ref-type="bibr" rid="B21">21</xref>].</p><p>In an attempt to address the high level of unawareness about diabetes, the first national diabetes screening campaign was launched in 2001 and implemented by public health services in Brazil. The target population was SUS users aged 40 years or older. The estimated national coverage of the campaign among the SUS target population was 73% [<xref ref-type="bibr" rid="B80">80</xref>]. Twenty million people were screened using capillary glycaemia tests and approximately 3.3 million (16.5%) suspected cases of diabetes were identified [<xref ref-type="bibr" rid="B81">81</xref>].</p><p>The Primary Health Care Department within the Health Care Secretariat develops measures to control and assess services from the primary health care and provides technical support to states, cities, and the Federal District. The Department organises basic health services including the Family Health Programme (PSF), oral health, hypertension and diabetes (HiperDia), food and nutrition, management and strategies, evaluation and follow-up activities [<xref ref-type="bibr" rid="B81">81</xref>].</p><p>According to the guidelines from HiperDia, risk prevention and care of patients with diabetes should take place at primary health care level [<xref ref-type="bibr" rid="B17">17</xref>]. The Family Health Strategy [<xref ref-type="bibr" rid="B5">5</xref>] was introduced in 1994, aiming to reorganise primary health care through the implementation of multi-disciplinary professional teams. These teams are responsible for the follow-up of a defined number of families located in a limited geographical area. The teams work on health promotion actions, prevention, recuperation, rehabilitation, and the maintenance of community health. The strategy aims to rationalise the use of all levels of assistance (primary, secondary and tertiary) and it has produced positive results for the main health indicators in the populations benefitting from the family health teams.</p><p>SUS provides essential medicines for diabetes control without additional costs for the system’s users. The free distribution of medicines in Brazil began in 1971, focusing on the poor population [<xref ref-type="bibr" rid="B82">82</xref>]. The Brazilian programme Popular Pharmacy was created in 2004 as a partnership between the federal government and states/municipalities aiming at increasing access to low-cost essential medications for the Brazilian population [<xref ref-type="bibr" rid="B82">82</xref>]. In 2006, this strategy was expanded to include private pharmacies and drug stores, named “<italic>Aqui Tem Farmácia Popular</italic>” (Popular Pharmacy is Available Here) [<xref ref-type="bibr" rid="B82">82</xref>]. As part of this programme, the Ministry of Health began subsidising 90% of the reference price of 24 medicines for the treatment of hypertension, diabetes, asthma, rhinitis, Parkinson disease, osteoporosis and glaucoma. This programme covers more than 2,500 municipalities and is available to 1.3 million Brazilians in need of medication (patients for whom drugs were prescribed), including 300,000 patients with diabetes [<xref ref-type="bibr" rid="B17">17</xref>].</p><p>In September 2006, a law was enacted to ensure the free distribution of diabetic medicines and the necessary equipment to monitor capillary glycaemia for all SUS insurees. In 2007, it was established that free medicines would be available to patients with diabetes, although the free distribution was restricted to patients whose treatment was provided by the SUS in primary health care units. In March 2011, the Brazilian government launched a programme called <italic>“Saúde Não Tem Preço”</italic> (Health has no price), to expand access to medicines for diabetes and hypertension. In this programme, the pharmacies and drugstores linked to the popular pharmacy network started to offer free medicines for the treatment of hypertension and diabetes (glibenclamide, metformin and insulin) in more than 17,500 registered private pharmacies [<xref ref-type="bibr" rid="B17">17</xref>]. A month after its launch, more than 3.7 million treatments were distributed, representing an increase of 70% in the distribution of medicines for hypertension and diabetes [<xref ref-type="bibr" rid="B17">17</xref>].</p><p>Brazil has participated in health promotion campaigns related to diabetes such as World Diabetes Day. The main strategy adopted by the Government to prevent chronic diseases is to control risk factors. Preventive efforts include anti-tobacco programmes, food and health nutrition policies (industry self-regulation code on advertising of food and beverages directed at children, regulation requiring the inclusion of warnings in all forms of advertising for products containing high levels of fat, sugars or salt), school health promotion, and actions to ensure essential medicines are provided in the public sector for hypertension and diabetes [<xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B83">83</xref>].</p><p>The Health Gym Programme was created in order to promote physical activity and provide free-of-charge spaces and support for living a healthy lifestyle [<xref ref-type="bibr" rid="B17">17</xref>]. According to the strategic action plan for coping with chronic diseases in Brazil from 2011 to 2022, the programme’s goal is to reach 4,000 municipalities by 2015 [<xref ref-type="bibr" rid="B84">84</xref>].</p></sec><sec><title>Actions for the future</title><p>Recently the Brazilian Ministry of Health launched the National Strategy for the Prevention and Control of NCDs for the period 2011–2022 [<xref ref-type="bibr" rid="B17">17</xref>]. The plan aims to prepare Brazil to confront and prevent the major chronic NCDs in the next ten years.</p><p>The Brazilian National Policy on Health Promotion [<xref ref-type="bibr" rid="B85">85</xref>] has prioritised drafting regulatory measures aimed at promoting healthy eating to reduce the prevalence of NCDs, with special emphasis on the regulation of food marketing and advertising, encouraging physical activity through gym classes at community levels, and implementing health promotion strategies in schools.</p><p>The expansion of pharmaceutical care and the free distribution of more than 15 medications for hypertension and diabetes play an important role in the Brazilian Government’s effort to tackle diabetes. In September 2011, Brazilian President Dilma Rousseff attended a general assembly summit at the UN headquarters in New York, contributing to global efforts in confronting the problem of NCDs [<xref ref-type="bibr" rid="B6">6</xref>]. The President reported that one of the first measures of her government was to increase access to medicines for poor patients with hypertension and diabetes. According to the President, 5.4 million Brazilians have taken advantage of the programme.</p></sec></sec><sec sec-type="conclusions"><title>Conclusions</title><p>According to the latest IDF estimates, the prevalence of diabetes in Brazil was 10.3% in 2012. However, this national level estimate hides important intra-country variation.</p><p>In the last few years, the Brazilian Ministry of Health has invested considerably in surveillance systems on NCDs. As a result, our review identified a number of data sources relevant to the study of diabetes covering morbidity (SIH-SUS HiperDia), mortality (SIM), risk factors (VIGITEL, ELSA), access and utilisation of health care services (PNAD, POF). However, it seems that the country is still not capitalising on available national data to produce the necessary evidence to identify gaps and formulate appropriate policy responses.</p><p>Data on diabetes costs are patchy and out-of-date. A multicountry study estimated that the total annual costs (direct and indirect) of diabetes in the country were USD $22.6 billion in 2000, representing a direct cost per capita of US $872. A more recent study estimated the direct and indirect costs of diabetes to be USD $ 2,108 per capita in 2007. There is some evidence on hospitalisation costs but no evidence on the cost of various types of complications.</p><p>A number of policies and programmes have been introduced by the Brazilian government in an attempt to improve access to diabetes care and reduce the prevalence of the disease. These include a national diabetes screening campaign in 2001, the Brazilian Popular Pharmacy programme introduced in 2004 and preventive efforts addressing risk factors (regulation of the food industry, promotion of physical activity through the health gym programme and anti-tobacco programmes).</p><p>Considering the magnitude of diabetes in Brazil, the Ministry of Health has adopted several strategies to reduce the costs of the disease in the Brazilian population, highlighting the interventions to be taken at the primary health care level. Specific programmes were implemented aimed at managing diabetes. However, some of the gaps include weak evaluation of the SUS in providing good quality care for patients with diabetes and lack of data on inequalities in access to medicines and health care services including annual testing for complications.</p><p>In conclusion, Brazil has the capacity to address and respond to NCDs due to the availability of federal, state and local integrated health programmes currently in operation. There is funding available for NCDs treatment, control and prevention, as well as health promotion, surveillance, monitoring and evaluation activities. However, these resources need to be used in the right way to be effective.</p></sec><sec><title>Abbreviations</title><p>AIH: (<italic>Autorização de Internação Hospitalar) -</italic> Hospital Admission Authorization Form; APAC: (<italic>Autorização de Procedimentos de Alta Complexidade)</italic> - Authorization for Procedures of High Complexity SUS; CNG: (<italic>Glomerulonefrite crônica)</italic> - Chronic glomerulonephritis; CKD: (<italic>Doença renal crônica</italic>) - Chronic kidney disease; DAB: (<italic>Departamento de Atenção Básica</italic>) - Primary Health Care Department; DALYs: (<italic>Anos de vida ajustados para incapacidade)</italic> - Disability adjusted life years; DATASUS: (<italic>Banco de Dados do SUS</italic>) – SUS Dataset; DHS: (<italic>Pesquisa de Demografia e Saúde)</italic> - Demographic and Health Survey; DM: (<italic>Diabetes Mellitus)</italic> - Diabetes Mellitus; DR: (<italic>Retinopatia Diabética</italic>) - Diabetic Retinopathy; ELSA: (<italic>Estudo Longitudinal de Saúde do Adulto)</italic> - Adult Health Longitudinal Study; ESF: (<italic>Estratégia Saúde da Família)</italic> - Family Health Strategy; GDP: (<italic>Produto Interno Bruto)</italic> - Gross Domestic Product; HAS: (<italic>Hipertensão Arterial Sistêmica)</italic> - Hypertension; HbA1c: Glycaeted Haemoglobin; HiperDia: (<italic>Sistema de cadastramento e acompanhamento de hipertensão e diabetes)</italic> – Hypertension and Diabetes Registration and Follow-up System; IBGE: (<italic>Instituto Brasileiro de Geografia e Estatística)</italic> - National Institute of Geography and Statistics; NCDs: (<italic>Doenças crônicas não transmissíveis)</italic> - Non-communicable diseases; PNAD: (<italic>Pesquisa Nacional de Amostra de Domicílios)</italic> - National Household Sample Survey; PNAUM: (<italic>Pesquisa Nacional sobre Acesso, Utilização e Promoção do Uso Racional de Medicamentos no Brasil)</italic> - National Research of Medicine Access and Utilization; PNDS: (<italic>Pesquisa Nacional de Demografia e Saúde)</italic> - National Demography and Health Survey; PNPS: (<italic>Política Nacional de Promoção da Saúde</italic>) - Brazilian National Policy on Health Promotion; PNS: (<italic>Pesquisa Nacional de Saúde)</italic> - National Health Research; POF: (<italic>Pesquisa de Orçamentos Familiares)</italic> - Family Budget Survey; PROESF: (<italic>Projeto de Expansão e Consolidação Saúde da Família)</italic> - Family Health Expansion and Consolidation Project; PSF: (<italic>Programa Saúde da Família)</italic> - Family Health Programme; SAMHPS: (<italic>Sistema de Assistência Médico-Hospitalar da Previdência Social)</italic> - Social Security Medical Assistance System; SIA/SUS: (<italic>Sistema de Informações Ambulatoriais</italic>) – Ambulatory Information System; SIH/SUS: (<italic>Sistema de Informações Hospitalares do SUS</italic>) - Hospital Information System; SAS: (<italic>Secretaria de Atenção à Saúde</italic>) – Health Care Secretariat; SIAB: (<italic>Sistema de Informação da Atenção Básica</italic>) – Primary Health Care Information System; SIM: (<italic>Sistema de Informação de Mortalidade</italic>) - Mortality Information System; SIS/HiperDia: (<italic>Sistema de Informação do HiperDia</italic>) - HiperDia system; SUS: (<italic>Sistema Único de Saúde</italic>) - Unified Health System; WHO: (<italic>Organização Mundial da Saúde</italic>) – World Health Organization; USAID: (<italic>Agência dos Estados Unidos para o Desenvolvimento Internacional</italic>) - http://www.usaid.gov/; VIGITEL: (<italic>Vigilância de Fatores de Risco e Proteção para Doenças Crônicas por Inquérito Telefônico</italic>) - Surveillance System of Risk and Protective Factors for Chronic Non-Communicable Diseases through Telephone Interviews.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests. The funding to conduct this study was provided by Novo Nordisk Switzerland. The sponsor had no involvement in the study design, data collection and analysis, and writing. AF received travel reimbursement and speaker fees from Novo Nordisk for delivering two presentations on diabetes in EU5 (France, Germany, Italy, Spain and UK) at national diabetes conferences in Portugal and Spain.</p></sec><sec><title>Author’s contributions</title><p>ADB was the main investigator involved in the acquisition of data and drafting the manuscript. PK coordinated the conception, design and interpretation of data. GVAF participated in the acquisition of data and in drafting the manuscript. AC and CAOT were involved in the acquisition of specific data and drafting part of the manuscript. PCH, MIS and AF revised the manuscript critically for important intellectual content. All authors read and approved the final version of the manuscript to be published.</p></sec>
Differential effects of temporal regularity on auditory-evoked response amplitude: a decrease in silence and increase in noise	<sec><title>Background</title><p>In daily life, we are continuously exposed to temporally regular and irregular sounds. Previous studies have demonstrated that the temporal regularity of sound sequences influences neural activity. However, it remains unresolved how temporal regularity affects neural activity in noisy environments, when attention of the listener is not focused on the sound input.</p></sec><sec><title>Methods</title><p>In the present study, using magnetoencephalography we investigated the effects of temporal regularity in sound signal sequencing (regular vs. irregular) in silent versus noisy environments during distracted listening.</p></sec><sec><title>Results</title><p>The results demonstrated that temporal regularity differentially affected the auditory-evoked N1m response depending on the background acoustic environment: the N1m amplitudes elicited by the temporally regular sounds were smaller in silence and larger in noise than those elicited by the temporally irregular sounds.</p></sec><sec><title>Conclusions</title><p>Our results indicate that the human auditory system is able to involuntarily utilize temporal regularity in sound signals to modulate the neural activity in the auditory cortex in accordance with the surrounding acoustic environment.</p></sec>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Okamoto</surname><given-names>Hidehiko</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>hokamoto@nips.ac.jp</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Teismann</surname><given-names>Henning</given-names></name><xref ref-type="aff" rid="I2">2</xref><xref ref-type="aff" rid="I3">3</xref><email>h.teismann@uni-muenster.de</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Keceli</surname><given-names>Sumru</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>sumru@nips.ac.jp</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Pantev</surname><given-names>Christo</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>pantev@uni-muenster.de</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Kakigi</surname><given-names>Ryusuke</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>kakigi@nips.ac.jp</email></contrib>	Behavioral and Brain Functions : BBF	<sec><title>Background</title><p>We are continuously exposed to environmental sounds in daily life. Task-relevant sound signals are often hidden in irrelevant noise. However, even when we do not voluntarily focus on surrounding sounds, we can easily detect such relevant sound signals despite the presence of ambient noises (e.g., someone calling our name in a noisy environment). Therefore, it seems plausible that humans are capable of continuously and involuntarily monitoring and segregating their acoustic environment [<xref ref-type="bibr" rid="B1">1</xref>]. However, the neural mechanisms that enable this accomplishment even when attention is not focused on the auditory input still remain elusive.</p><p>Previous studies have shown that the neural responses evoked by sounds in noisy conditions differ from those evoked during silent conditions. Auditory-evoked responses were shown to be reduced and delayed in noisy environments [<xref ref-type="bibr" rid="B2">2</xref>-<xref ref-type="bibr" rid="B4">4</xref>]. On the other hand, Alain <italic>et al.</italic>[<xref ref-type="bibr" rid="B5">5</xref>] showed that low intensity background noise could enhance the amplitude of auditory responses evoked by sound signals. These results indicated that background noise could lead to both a decrease and increase in the auditory-evoked response amplitude, depending on the situation during which the sound signals appear.</p><p>The repetition of identical sound stimuli in a silent background is known to lead to decreased N1(m) responses ([<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref>], for a review see [<xref ref-type="bibr" rid="B8">8</xref>]). However, a previous study [<xref ref-type="bibr" rid="B9">9</xref>] demonstrated that the repetition of a constant frequency sound in noisy environments resulted in significantly larger N1m responses than those elicited by randomly presented frequencies, even when the participants did not pay attention to the auditory modality. These results support the hypothesis that predictable auditory patterns aid the perception of the auditory scene [<xref ref-type="bibr" rid="B10">10</xref>] while attention is not focused on the auditory signals. The auditory neural pathway is tonotopically organized [<xref ref-type="bibr" rid="B11">11</xref>-<xref ref-type="bibr" rid="B13">13</xref>]. In noisy environments, spectral cues appear to enhance the neural activity located at the corresponding tonotopic map spot in the human auditory neural pathway even when the listener’s attention is distracted from the auditory input. In the study described above [<xref ref-type="bibr" rid="B9">9</xref>], all test sounds were presented in a temporally regular manner (inter stimulus interval (ISI) = 2400 ms). Listeners implicitly knew the timing and the frequencies of the upcoming sound stimuli in the constant frequency sound sequence; therefore, they could involuntarily assign their processing resources to the auditory neurons corresponding to these frequencies in a timely manner, resulting in larger amplitudes and shorter latencies of the auditory-evoked responses. Therefore, we assume that predictable auditory patterns may not be limited to spectral information; in noisy environments, temporal information may also play an important role for the auditory neural processing.</p><p>Based on these considerations, the goal of the present study was to investigate the effects of temporal regularity in sound signals on auditory-evoked responses in both silent and noisy environments while the participants’ attention was distracted from the auditory modality. A regular ISI may enable participants to involuntarily modulate their neural activity in the time domain to the appearance or absence of the sound signals, while that would be difficult with irregular ISIs. We hypothesized that even in an unattended situation the effects of temporal regularity on the auditory-evoked responses would differ between silence and noisy environments, in which the listeners would have to segregate sound signals from ambient noise.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Participants</title><p>Thirteen healthy participants (5 females, age range 23 – 44 years) participated in the present study. All participants had normal hearing and no neurological disorders. All participants were fully informed about the study and gave written informed consent for their participation in accordance with procedures approved by the Ethics Commission of the National Institute for Physiological Sciences. The study thus conformed to The Code of Ethics of the World Medical Association (Declaration of Helsinki).</p></sec><sec><title>Stimuli and experimental design</title><p>The experimental design is schematically represented in Figure <xref ref-type="fig" rid="F1">1</xref>. The test stimulus (TS) was a 1000 Hz pure tone. The TS was presented either together with broad-band background noise or in silence. Sound onset asynchrony was fixed to 2000 ms in the regular-sequencing condition, or pseudo-randomly selected from 1000, 1500, 2000, 2500, and 3000 ms in the irregular-sequencing condition (Figure <xref ref-type="fig" rid="F1">1</xref>). The TS had a fixed duration of 500 ms, including 10 ms onset- and offset-ramps. Therefore, the inter-stimulus interval (ISI) between the offset of a TS and the onset of the following TS was either regular (1500 ms) or irregular (500, 1000, 1500, 2000, or 2500 ms). All sounds were diotically presented through plastic tubes of 1.5 m length and earpieces fitted to the participant’s ears. The noise recorded at the earpiece using an ear simulator (Type 4157, Brüel & Kjaer Sound & Vibration Measurement, Naerum, Denmark) showed a low-pass filtered frequency characteristic, reflecting the frequency response of our sound delivery system (Figure <xref ref-type="fig" rid="F2">2</xref>). Before starting magnetoencephalography (MEG) data acquisition, each participant’s hearing threshold for the TS was individually determined for each ear. During the MEG recording session, the 1000 Hz TS was presented at an intensity of 40 dB above the individual sensation level. The broadband masking noise had 10 dB more power (not loudness) than the TS (cf. Additional file <xref ref-type="supplementary-material" rid="S1">1</xref> (audio file: regular sequencing in noise) and Additional file <xref ref-type="supplementary-material" rid="S2">2</xref> (audio file: irregular sequencing in noise)). In order to keep the test participants alert and distracted from the auditory signals, a self-chosen silent movie was presented during the MEG recordings. At the end of the measurement, questions regarding the content of the movie were asked to ensure that the participants had paid attention to the movie.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Experimental design.</bold> Schematic display of auditory stimulation in the regular sequencing (left column) and irregular sequencing conditions (right column), and in the silent (upper panels) and noisy (lower panels) conditions. The test stimulus (TS) and background noise are represented by short black solid lines and gray areas, respectively. Sound onset asynchrony between two successive TS was 2 s in the regular sequencing condition (left column), and either 1.0, 1.5, 2.0, 2.5, or 3.0 s in the irregular sequencing condition (right column).</p></caption><graphic xlink:href="1744-9081-9-44-1"/></fig><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>The amplitude spectrum of white noise recorded at the ear piece.</bold> Because of our sound delivery system (plastic tube and ear piece), the sound spectrum exhibited a 2 kHz low-pass filtered characteristic. Sample audio files of the test stimulus embedded in the background noise are available as Additional files <xref ref-type="supplementary-material" rid="S1">1</xref> (Regular Sequencing) and <xref ref-type="supplementary-material" rid="S2">2</xref> (Irregular Sequencing).</p></caption><graphic xlink:href="1744-9081-9-44-2"/></fig><p>In order to investigate temporal regularity (Regular vs. Irregular) and noise level (Silent vs. Noisy) effects, we used four different conditions (Figure <xref ref-type="fig" rid="F1">1</xref>): regular sequencing in silence (Regular_Silent), irregular sequencing in silence (Irregular_Silent), regular sequencing in noise (Regular_Noisy), and irregular sequencing in noise (Irregular_Noisy). Each MEG session consisted of eight blocks (two blocks per condition) of 150 trials, resulting in 300 trials per condition. The block order was pseudo-randomized among participants. The mean ISI of the irregular-sequencing condition was kept equal to the mean ISI of the regular-sequencing condition (mean ISI = 1500 ms).</p></sec><sec><title>Data acquisition and analysis</title><p>Auditory-evoked fields were recorded with a helmet-shaped, 306 channel MEG system (Vector-view, ELEKTA, Neuromag, Helsinki, Finland) with 102 identical triple sensor elements located in a silent, magnetically shielded room. We analyzed the MEG signals recorded by 204 planar-type gradiometers, and detected the largest signals over the corresponding cerebral sources. Signals were passed through a 0.03 – 200 Hz band-pass filter and digitized at 600 Hz. The magnetic fields evoked by TS were averaged selectively for each condition, starting 300 ms prior to TS onset, and ending 200 ms after TS offset. Participants were instructed not to move their heads during the recordings; compliance was monitored through a video camera by the experimenter. Epochs containing amplitude changes greater than 3 pT were discarded as artifact-contaminated epochs.</p><p>The locations and orientations of the equivalent current dipoles were estimated using the BESA software (BESA Research 5.3.7, BESA GmbH, Germany). To analyze the N1m component, which is the major deflection of the auditory-evoked field (for reviews see [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B14">14</xref>]), the averaged fields were 30 Hz low-pass filtered (zero-phase shift Butterworth filter, 24 dB/oct), and the baseline was corrected relative to the 250 ms pre-stimulus interval. Previous studies [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B15">15</xref>-<xref ref-type="bibr" rid="B17">17</xref>] showed that the calculated source locations and orientations of the N1m responses were not influenced by the presence of acoustic noise. Moreover, the estimated single dipole source strength was shown to be dependent on the depth of the estimated location [<xref ref-type="bibr" rid="B18">18</xref>]. Thus, in order to improve the signal-to-noise ratio, we grand-averaged the magnetic fields of all conditions and used these grand-averaged magnetic waveforms to estimate the single equivalent current dipoles reflecting the N1m response. The peak N1m response was initially identified as the maximal root-mean square value of the global field power around 100 ms after TS onset. The 10 ms time window around the peak was then used for dipole source estimation. Source locations and orientations were then estimated at the N1m amplitude peak by means of single equivalent current dipole modeling (one dipole per hemisphere) for each participant individually. Estimated sources, which were fixed in location and orientation for each hemisphere of each participant, served as a spatial filter [<xref ref-type="bibr" rid="B19">19</xref>] during the calculation of the source strength waveforms for each condition. The mean source strength within the 10 ms time window around the peak N1m latency in each hemisphere and each condition in the time range between 80 and 300 ms was used for statistical analysis.</p><p>In order to evaluate the effects of noise and temporal regularity, the source strengths and latencies of the N1m responses averaged across hemispheres in each condition were analyzed separately via a repeated-measures analysis of variance (ANOVA) using the two factors NOISE_LEVEL (Silent vs. Noisy) and SEQUENCING (Regular vs. Irregular).</p></sec></sec><sec sec-type="results"><title>Results</title><p>After artifact rejection, more than 90% of trials could be averaged for each condition in all participants (mean ± standard deviation: Regular_Silent = 298.7 ± 2.6 trials, Irregular_Silent = 297.5 ± 2.3 trials, Regular_Noisy = 296.4 ± 3.6 trials, Irregular_Noisy = 297.8 ± 3.1 trials). Dipolar magnetic field patterns over the left and right hemispheres were observed in all conditions (Figure <xref ref-type="fig" rid="F3">3</xref>A and B). The amplitudes of auditory-evoked fields were much larger in silent conditions than in noisy conditions. However, magnetic field distributions were very similar between silent and noisy conditions, indicating that the foci of the neural sources were similar. The grand-averaged waveforms used for the equivalent current dipole estimation and estimated source locations and orientations of the N1m response overlaid on the axial slice of the structural magnetic resonance image of one representative participant are displayed in Figure <xref ref-type="fig" rid="F3">3</xref>C. The goodness-of-fit of the underlying dipolar source models for the grand-averaged MEG waveforms was above 90% in all cases. The estimated dipolar sources were located at the superior temporal plane, which corresponded to the N1m generator [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B20">20</xref>].</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Representative subject result. (A)</bold> Individual auditory-evoked magnetic fields under each condition. N1m responses are indicated by the red vertical lines. <bold>(B)</bold> Isocontour maps of the magnetic fields at the N1m latency. The magnetic contour maps show clear dipolar patterns above the left auditory cortex. Red and blue contour lines represent the outbound and inbound flows of magnetic fields from and into the brain. Different scales were used between the silent and noisy conditions. <bold>(C)</bold> Source estimation. Estimated equivalent current dipoles at the latency of the maximal N1m response are illustrated together with a three-dimensional head and brain model reconstructed from the individual MRI. The spheres and barrels indicate the locations and orientations of the single dipoles in the left (red) and right (blue) hemispheres.</p></caption><graphic xlink:href="1744-9081-9-44-3"/></fig><p>The time courses (time range from -100 to +650 ms) of the source strengths averaged across all participants and hemispheres are displayed in Figure <xref ref-type="fig" rid="F4">4</xref>. The N1m responses had larger amplitudes and shorter latencies overall in the silent condition than in the noisy condition. Figure <xref ref-type="fig" rid="F5">5</xref> shows the mean N1m source strengths and latencies in each condition together with the corresponding 95% confidence intervals. Repeated-measures ANOVAs evaluating N1m source strength and N1m latency resulted in significant main effects for NOISE_LEVEL (Source strength: F (1, 12) = 39.01, <italic>p</italic> < 0.001; Latency: F (1, 12) = 284.86, <italic>p</italic> < 0.001) and SEQUENCING (Source strength: F (1, 12) = 13.81, <italic>p</italic> < 0.01; Latency: F (1, 12) = 47.73, <italic>p</italic> < 0.001). Additionally, there were significant interactions between NOISE_LEVEL and SEQUENCING (Source strength: F (1, 12) = 24.69, <italic>p</italic> < 0.001; Latency: F (1, 12) = 34.26, <italic>p</italic> < 0.001). Planned contrasts (Regular_Silent vs. Irregular_Silent; Regular_Noisy vs. Irregular_Noisy) were calculated in order to further explore the interactions between NOISE_LEVEL and SEQUENCING. The Bonferroni multiple comparison correction was used to control the family-wise error rate. The N1m source strength in the silent condition was significantly larger for irregular than for regular (two-tailed paired t-test: t(12) = 4.876, <italic>p</italic> < 0.001 (Bonferroni-corrected)), whereas the N1m source strength in the noisy condition was significantly larger for regular than for irregular (t(12) = 3.27, <italic>p</italic> < 0.02 (Bonferroni-corrected)). The N1m latency was significantly longer for irregular than for regular both in the silent (t(12) = 4.351, <italic>p</italic> < 0.002 (Bonferroni-corrected)) and noisy (t(12) = 6.410, <italic>p</italic> < 0.001 (Bonferroni-corrected)) conditions. The significant interaction between NOISE_LEVEL and SEQUENCING for the N1m source strength demonstrated that regular (compared to irregular) SEQUENCING increased neural activity under the noisy condition, while neural activity was decreased under the silent condition. Moreover, the regular sequencing shortened the N1m latency in both the silent and noisy conditions.</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>Time courses of the mean source strengths across all participants (N = 13) and hemispheres.</bold> Each colored line represents an experimental condition (see legends in the right upper corner).</p></caption><graphic xlink:href="1744-9081-9-44-4"/></fig><fig id="F5" position="float"><label>Figure 5</label><caption><p><bold>N1m source strengths and latencies.</bold> Group means (N = 13) of the N1m source strengths (left graph) and latencies (right graph) in the “Silent” and “Noisy” conditions including error bars denoting the 95% confidence intervals. Open and filled bars denote the “Regular” and “Irregular” conditions (* <italic>p</italic> < 0.05, ** <italic>p</italic> < 0.01 (Bonferroni-corrected)).</p></caption><graphic xlink:href="1744-9081-9-44-5"/></fig></sec><sec sec-type="discussion"><title>Discussion</title><p>The results of the present study demonstrated that the magnitude and latency of neural responses elicited in silent and noisy environments depended on temporal regularity in the sound sequences used to evoke the responses, even when the participants did not pay attention to the auditory signals. N1m latencies were shorter with regular ISI than with irregular ISI in both the silent and noisy conditions. However, N1m response amplitudes were smaller with regular ISI than with irregular ISI in the silent condition, whereas they were larger in the noisy condition. Notably, the test stimulus sequence was identical between the silent and noisy conditions. To the best of our knowledge, these results are the first to demonstrate the differential effects of temporal regularity on auditory-evoked response amplitudes in noisy versus silent backgrounds during distracted listening.</p><p>In a silent environment, repeated exposure to sounds with identical features may induce neural adaptation and a consequent decline in the auditory-evoked response amplitude [<xref ref-type="bibr" rid="B21">21</xref>-<xref ref-type="bibr" rid="B23">23</xref>]. In the present study, participants could implicitly foresee the timing of the onset of the upcoming TS during the regular ISI condition, whereas this was difficult in the irregular sequencing condition. Previous electroencephalography and MEG studies [<xref ref-type="bibr" rid="B24">24</xref>-<xref ref-type="bibr" rid="B26">26</xref>] demonstrated that knowledge of the stimulus onset timing could reduce N1(m) amplitude and shorten N1(m) latency in silent background conditions. The present study also confirmed that N1m source strengths were smaller and N1m latencies were shorter with the regular than with the irregular condition in silence (Figures <xref ref-type="fig" rid="F4">4</xref> and <xref ref-type="fig" rid="F5">5</xref>). In the silent environment, neural adaptation in response to stimulus timing may have led to lower N1m amplitudes and shortened N1m latencies elicited by the TS with regular ISI than with irregular ISI. Moreover, a recent functional MRI (fMRI) study [<xref ref-type="bibr" rid="B27">27</xref>] investigated neural activity while participants were listening to temporally regular or irregular sequences of tones and were performing an intensity discrimination task (not periodicity detection task). The authors found that in silence, the regular sequences caused larger neural activity in the putamen and smaller neural activity in the primary and secondary auditory cortices than the irregular sequences. In the present study, after regular sequences in silence, we also observed smaller N1m responses, which originate in non-primary auditory cortex; however, we were not able to study neural activity in the putamen. The reason is that the MEG-sensors are very sensitive to superficial activity originating in cortical sulci, but almost insensitive to the neural activity in the putamen. A cortical-striatal system (for reviews see [<xref ref-type="bibr" rid="B28">28</xref>,<xref ref-type="bibr" rid="B29">29</xref>]) appears to be involved in the neural processing of temporal regularity in sound sequences. Enhanced neural activity in the striatum appears to take charge of the neural processing of temporally regular sounds, resulting in reduced neural activity in the auditory cortex.</p><p>The present results observed in the noisy condition, showing larger N1m responses with the regular than with the irregular sequencing condition, seem at first sight contradictory to the results obtained both in the silent condition and in previous studies [<xref ref-type="bibr" rid="B24">24</xref>-<xref ref-type="bibr" rid="B26">26</xref>,<xref ref-type="bibr" rid="B30">30</xref>,<xref ref-type="bibr" rid="B31">31</xref>]. In contrast to the current knowledge of the neural adaptation of auditory-evoked responses in silence, very little is known for noisy environments. While participants were distracted from the auditory modality, Lagemann <italic>et al.</italic>[<xref ref-type="bibr" rid="B32">32</xref>] presented a train of four consecutive tones of the same frequency separated by a regular silent interval of 500 ms under a broadband masking environment. The authors demonstrated that the auditory-evoked field amplitudes were similar between the first, second, third, and fourth tones. Therefore, the neural mechanisms of adaptation to the sound signals might differ in silent and noisy environments. Temporal regularity in sound inputs in a noisy environment may have driven the involuntary constitution of a specific auditory stream as a figure, which may then have been easily segregated from the background noise by the listeners [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B33">33</xref>-<xref ref-type="bibr" rid="B35">35</xref>]. However, the formation of an auditory stream may have been unstable in case of the irregular ISI condition. The significant results observed in this study suggest that the bottom-up driven formation of auditory figure-ground segregation may facilitate neural activity tracking of the regular ISI test sound signals. Teki <italic>et al.</italic>[<xref ref-type="bibr" rid="B35">35</xref>] investigated the neural bases of auditory stimulus-driven figure-ground segregation by using a unique stimulus that incorporated stochastic variation of the signal components in frequency-time space. Figure and ground auditory signals overlapped in spectrotemporal space, but differed in their statistics of fluctuations. By means of fMRI the authors measured the brain activity related to figure-ground decomposition while the participants performed an irrelevant task. The authors observed significantly increased activations in the intraparietal sulcus, the superior temporal sulcus, and the right planum temporale as a function of increasing duration of the figures, and increased activations in the intraparietal sulcus and the superior temporal sulcus as a function of increasing the number of components of the figures. In the present study, in the Regular_Noisy condition we also observed an enlarged N1m response, which appears to originate mainly in the planum temporale [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B20">20</xref>,<xref ref-type="bibr" rid="B36">36</xref>]. Neural activity in the supra temporal sulcus might partially contribute to the increased N1m amplitude in the Regular_Noisy condition, since the dipole location merely represents the center of gravity of the neural responses, and not the extent of activated areas. On the other hand, we did not observe intraparietal sulcus activity in the Regular_Noisy condition, as shown in Figure <xref ref-type="fig" rid="F3">3</xref>. This inconsistency could be due to the different functional neuroimaging procedures (fMRI vs. MEG). In the present MEG study, we analyzed the auditory-evoked N1m response that was precisely time-locked to the bottom-up sound inputs and had a specific latency in a millisecond temporal resolution. Therefore, the neural activity in the intraparietal sulcus, which is an area outside the classical auditory cortex, might not be time-locked to the bottom-up sound inputs, or might have a different latency from the N1m response, leading to almost no contribution to the auditory-evoked N1m response measured by MEG.</p><p>In the present study, we used five different ISIs (500, 1000, 1500, 2000, or 2500 ms) in the irregular sequencing condition and only one ISI (1500 ms) in the regular condition. Longer ISIs are known to elicit larger N1m responses than shorter ISIs, and this effect was shown to be non-linear [<xref ref-type="bibr" rid="B37">37</xref>-<xref ref-type="bibr" rid="B41">41</xref>]. Therefore, even though the mean values of the irregular and regular ISIs were identical, the non-linearity of the ISI effect may have led to differential N1m response amplitudes for the irregular and regular ISI conditions. The amplitudes of the auditory-evoked responses elicited in a silent environment were shown to have a negative exponential dependence on the ISI [<xref ref-type="bibr" rid="B37">37</xref>,<xref ref-type="bibr" rid="B42">42</xref>]. This negative exponential ISI dependence may have led to smaller N1m amplitudes in the irregular sequencing condition than in the regular sequencing condition. However, the present study showed that the N1m amplitudes obtained in the silent condition were significantly larger in the irregular than in the regular condition. Therefore, the negative exponential dependence of the N1m amplitude on ISI alone cannot explain the obtained results.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>Our present findings demonstrated that the human auditory system is able to implicitly utilize temporal regularities in sound signals to modulate neural activity. Under circumstances permitting trouble-free sound detection, for instance in silence, the auditory system seems to be able to reduce the amount of neural activity allocated for processing temporally regular sounds. In contrast, when the circumstances are less optimal and auditory figure-ground segregation is required (e.g., in the presence of disturbing noise) the auditory system may use temporal regularity to properly allocate neural resources along the time axis and to effectively segregate a sound signal from the background.</p></sec><sec><title>Abbreviations</title><p>ANOVA: Analysis of variance; fMRI: functional magnetic resonance imaging; ISI: Inter-stimulus interval; MEG: Magnetoencephalography; TS: Test stimulus.</p></sec><sec><title>Competing interests</title><p>All authors declare that they have no conflicts of interests.</p></sec><sec><title>Authors’ contributions</title><p>Conceived and designed the experiments: HO. Performed the experiments: HO, SK. Analyzed the data: HO, HT. Drafted the manuscript: HO. Revised the manuscript: HO, HT, SK, CP, RK. All authors read and approved the final manuscript.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional file 1</title><p>An exemplary sound representing regular sequencing in the noisy condition.</p></caption><media xlink:href="1744-9081-9-44-S1.mp3"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S2"><caption><title>Additional file 2</title><p>An exemplary sound representing irregular sequencing in the noisy condition.</p></caption><media xlink:href="1744-9081-9-44-S2.mp3"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
A randomized controlled trial investigating the neurocognitive effects of Lacprodan® PL-20, a phospholipid-rich milk protein concentrate, in elderly participants with age-associated memory impairment: the Phospholipid Intervention for Cognitive Ageing Reversal (PLICAR): study protocol for a randomized controlled trial	<sec><title>Background</title><p>Age-related cognitive decline (ARCD) is of major societal concern in an ageing population, with the development of dietary supplements providing a promising avenue for amelioration of associated deficits. Despite initial interest in the use of phospholipids (PLs) for ARCD, in recent years there has been a hiatus in such research. Because of safety concerns regarding PLs derived from bovine cortex, and the equivocal efficacy of soybean-derived PLs, there is an important need for the development of new PL alternatives. Phospholipids derived from milk proteins represent one potential candidate treatment.</p></sec><sec><title>Methods</title><p>In order to reduce the effects of age-associated memory impairment (AAMI) the Phospholipid Intervention for Cognitive Ageing Reversal (PLICAR) was developed to test the efficacy of a milk protein concentrate rich in natural, non-synthetic milk phospholipids (Lacprodan® PL-20). PLICAR is a randomized, double-blind, placebo-controlled parallel-groups study where 150 (N = 50/group) AAMI participants aged > 55 years will be randomized to receive a daily supplement of Lacprodan® PL-20 or one of two placebos (phospholipid-free milk protein concentrate or inert rice starch) over a 6-month (180-day) period. Participants will undergo testing at baseline, 90 days and 180 days. The primary outcome is a composite memory score from the Rey Auditory Verbal Learning Test. Secondary outcomes include cognitive (verbal learning, working memory, prospective and retrospective memory, processing speed and attention), mood (depression, anxiety, stress and visual analogue scales), cardiovascular (blood pressure, blood velocity and pulse wave pressure), gastrointestinal microbiota and biochemical measures (oxidative stress, inflammation, B vitamins and Homocysteine, glucoregulation and serum choline). Allelic differences in the Apolipoprotein E and (APOE) and <italic>Methylenetetrahydrofolate reductase</italic> (MTHFR) gene will be included for subgroup analysis. A subset (N = 60; 20/group)) will undergo neuroimaging using functional magnetic resonance imaging (fMRI) and magnetoencephalography (MEG) in order to further explore <italic>in vivo</italic> central mechanisms of action of Lacprodan® PL-20. This study will enable evaluation of the efficacy of milk-derived phospholipids for AAMI, and their mechanisms of action.</p></sec><sec><title>Trial Registration</title><p>The trial is jointly funded by Arla Foods and Swinburne University of Technology, currently recruiting and is registered on the Australian New Zealand Clinical Trials Registry as <ext-link ext-link-type="uri" xlink:href="https://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=363906">ACTRN12613000347763</ext-link>.</p></sec>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Scholey</surname><given-names>Andrew B</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>andrew@scholeylab.com</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Camfield</surname><given-names>David A</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>david.camfield@gmail.com</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Hughes</surname><given-names>Matthew E</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>matthewhughes@swin.edu.au</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Woods</surname><given-names>Will</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>woods@swin.edu.au</email></contrib><contrib contrib-type="author" id="A5"><name><surname>K Stough</surname><given-names>Con K</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>cstough@gmail.com</email></contrib><contrib contrib-type="author" id="A6"><name><surname>White</surname><given-names>David J</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>dawhite@swin.edu.au</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Gondalia</surname><given-names>Shakuntla V</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>sgondalia@swin.edu.au</email></contrib><contrib contrib-type="author" id="A8"><name><surname>Frederiksen</surname><given-names>Pernille D</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>pernille.dorthea.frederiksen@arlafoods.com</email></contrib>	Trials	<sec><title>Background</title><p>Cognitive deficits including learning and memory impairment are one of the most prominent and debilitating consequences of normal and pathological ageing in humans. A consistent finding in research on ageing and cognition is that performance across various tests of memory is lower with increased age. Meta-analyses of cross-sectional and longitudinal datasets have demonstrated approximately 40 to 60% decline in cognitive speed at age 80 compared to age 20 years in non-demented adults [<xref ref-type="bibr" rid="B1">1</xref>]. These cognitive deficits are of considerable concern to elderly individuals, with up to 50% of adults aged 64 years or older reporting difficulties with their memory [<xref ref-type="bibr" rid="B2">2</xref>]. There is an increasing awareness of the possibility that dietary modification can alter the course of age-related cognitive decline.</p><p>Frequent dairy food intake is associated with better cognitive function, although the exact underlying mechanisms are yet to be determined. The positive correlation between increased dairy intake and cognitive function seems partly to be a result of counteracting metabolic syndrome, by reducing cardiovascular (CV) risk factors such as type 2 diabetes, hypertension, obesity and hyperlipedimia [<xref ref-type="bibr" rid="B3">3</xref>-<xref ref-type="bibr" rid="B5">5</xref>]. In a recent paper by Crichton <italic>et al</italic>. [<xref ref-type="bibr" rid="B6">6</xref>], the authors support this link and in addition argue that apart from counteracting the CV risk factors, dairy consumption provides additional benefits in relation to cognitive function (for example, after adjusting for CV disease, lifestyle and dietary factors). Recent advances in dairy manufacturing, which enable the enrichment of phospholipid components in milk also offer a promising new avenue for the treatment of age-related cognitive decline.</p><sec><title>Phospholipids and the amelioration of cognitive decline</title><p>Bovine milk contains a vast range of phospholipids and complex lipids, with important biological functions. Phospholipids are substances with both a hydrophilic (water-liking) head and hydrophobic (water-repellent) tail segment which are major building blocks for cellular membranes, including neuronal cells, where they arrange themselves into lipid bilayers [<xref ref-type="bibr" rid="B7">7</xref>]. Relevant phospholipids in relation to cognitive performance include phosphatidylserine (PS) and phosphatidylcholine (PC) as well as the related substances sphingomyelin and the sialic-acid-containing gangliosides. Pharmacokinetic studies of PS have revealed good bioavailability when consumed orally. Following ingestion, the headgroup is absorbed intact through the intestinal wall into the bloodstream, while the fatty-acid tail segments at position 2 are often removed and later re-added. After crossing the blood–brain barrier the fatty acid at position 2 is typically occupied by either oleic (C18:1) or docosahexaenoic acid (DHA, C22:6) in the brain [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B9">9</xref>].</p><sec><title>Phosphatidylserine (PS)</title><p>PS is a naturally occurring membrane phospholipid that is found in high concentrations in brain tissue, where it comprises 10 to 20% of the total phospholipid pool [<xref ref-type="bibr" rid="B10">10</xref>]. PS plays an important role in a host of cellular functions including mitochondrial membrane integrity, presynaptic neurotransmitter release, postsynaptic receptor activity and activation of protein kinase C in memory formation [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B11">11</xref>]. PS enhances the activities of membrane-bound enzymes involved in signal transduction [<xref ref-type="bibr" rid="B12">12</xref>] and plays a key role in the biosynthesis and release of several neurotransmitters, including acetylcholine [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B14">14</xref>], norepinephrine [<xref ref-type="bibr" rid="B15">15</xref>], serotonin and dopamine [<xref ref-type="bibr" rid="B16">16</xref>]. PS has also been found to elevate glucose metabolism [<xref ref-type="bibr" rid="B17">17</xref>].</p><p>In relation to the cholinergic system, PS has been found to restore age-related decreases in choline-acetyltransferase-positive neurons [<xref ref-type="bibr" rid="B18">18</xref>] as well as densities of muscarinic and N-methyl-D-aspartate (NMDA) receptors [<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>]. Increases in nerve growth factor (NGF)-receptor density have been reported in aged animals following PS supplementation, as well as increases in neuronal numbers and size [<xref ref-type="bibr" rid="B21">21</xref>]. Other beneficial effects of PS supplementation include protection from cell death [<xref ref-type="bibr" rid="B22">22</xref>] due to increased membrane fluidity [<xref ref-type="bibr" rid="B23">23</xref>] as well as anti-inflammatory and antioxidant effects [<xref ref-type="bibr" rid="B24">24</xref>]. Behavioural animal experiments using PS have provided evidence of improvements in spatial memory [<xref ref-type="bibr" rid="B25">25</xref>], retention of passive avoidance [<xref ref-type="bibr" rid="B15">15</xref>], exploration and memory retrieval [<xref ref-type="bibr" rid="B26">26</xref>] as well as improvement of avoidance performance [<xref ref-type="bibr" rid="B27">27</xref>]. Prevention of scopolamine-induced learning deficits have also been reported [<xref ref-type="bibr" rid="B28">28</xref>-<xref ref-type="bibr" rid="B31">31</xref>], together with an attenuation of memory deficits associated with reserpine-induced catecholamine depletion [<xref ref-type="bibr" rid="B32">32</xref>].</p><p>PS is the most studied of the phospholipids in regards to human clinical trials of dementia and cognitive decline. Five double-blind randomized trials have been conducted using PS in Alzheimer’s disease (AD) [<xref ref-type="bibr" rid="B33">33</xref>-<xref ref-type="bibr" rid="B37">37</xref>]. Clinical global impressions of change and activities of daily living were found to be improved with daily doses of 200 to 300 mg up to six months. In milder cases of AD, significant improvements to concentration, learning and memory for names, locations and recent events were also observed [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B11">11</xref>]. In the largest of these studies involving 494 elderly patients with moderate to severe cognitive decline, Cenacchi <italic>et al</italic>. [<xref ref-type="bibr" rid="B37">37</xref>] reported improvements to memory and learning following 300 mg PS/day for 6 months.</p><p>In elderly populations with mild cognitive impairment and age-associated memory impairment (AAMI), PS has also been found to be effective in ameliorating cognitive declines. Crook <italic>et al</italic>. [<xref ref-type="bibr" rid="B38">38</xref>] administered 300 mg/day PS to 149 elderly patients (aged 50 to 75 years) with age-associated memory impairment for 12 weeks, and observed significant improvements in performance tests of learning and recall abilities such as name-face matching. In a multi-centre trial by Villardita <italic>et al</italic>. [<xref ref-type="bibr" rid="B39">39</xref>], 300 mg/day PS versus placebo was administered to 170 elderly patients for 90 days. Significant improvements in attention, concentration and short-term memory were found in those receiving PS. These earlier trials were conducted using PS derived from bovine cortex, but more recently trials have also been conducted using soybean (SB)-derived PS due to concerns over bovine spongiform encephalopathy. The results in relation to SB-PS have been more mixed, with Jorissen <italic>et al</italic>. [<xref ref-type="bibr" rid="B40">40</xref>] reporting no significant effects on learning and memory, whereas more recently Vakhaporva <italic>et al</italic>. [<xref ref-type="bibr" rid="B41">41</xref>] and Kato-Kataoka <italic>et al</italic>. [<xref ref-type="bibr" rid="B42">42</xref>] reported significant improvements to verbal learning following ≥3 months supplementation. An intriguing study by Hellhammer <italic>et al</italic>. [<xref ref-type="bibr" rid="B43">43</xref>] using the same milk-derived phospholipids as the current study (PL-20) also reported improvements to working memory function following 3 weeks supplementation in healthy adults aged 30 to 55 years.</p></sec><sec><title>Phosphatidylcholine (PC)</title><p>Choline is an essential nutrient critically needed for synthesis of the neurotransmitter acetylcholine; important in brain functions, such as memory and mood, but also important in skeletal-muscle control, heart rate and breathing. Numerous animal studies demonstrate that choline is necessary for normal development of the memory function and sub-optimal dietary intake of choline by the pregnant mother and later by the infant and child directly affects brain development and results in permanent changes in brain function (Zeisel <italic>et al</italic>. 1991). Choline or its metabolites, are also needed for the structural integrity and signaling functions of cell membranes; it is the major source of methyl-groups in the diet (one of choline’s metabolites, betaine, participates in the methylation of homocysteine (HCy) to form methionine), and it directly affects cholinergic neurotransmission, transmembrane signaling and lipid transport/metabolism [<xref ref-type="bibr" rid="B44">44</xref>,<xref ref-type="bibr" rid="B45">45</xref>]. Nagata <italic>et al</italic>. [<xref ref-type="bibr" rid="B46">46</xref>] documented recently that dietary supplements of PhosChol compounds (1,2-dilinoleoyl-sn-glycero-3-phosphocholine; DL-PC and 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine; PO-PC) ) enhanced the memory and learning ability in the elderly in the Mini Mental State Examination (MMSE). In another placebo-controlled clinical trial, Ladd <italic>et al</italic>. [<xref ref-type="bibr" rid="B47">47</xref>] found that the supplementation of SB-PC in normal college students lead to an improvement in explicit memory function due to increased choline supply and improved cholinergic function.</p><p>A poor folate status is associated with cognitive decline and dementia in older adults [<xref ref-type="bibr" rid="B48">48</xref>]. Although impaired brain methylation activity and HCy toxicity are widely thought to account for this association, how folate deficiency impairs cognition is still uncertain. Troen <italic>et al</italic>. [<xref ref-type="bibr" rid="B49">49</xref>] found a correspondence of cognitive outcomes to changes in brain membrane PC content (in rats), which suggests that altered PC and possibly choline metabolism might contribute to the manifestation of folate deficiency-related cognitive dysfunction. Up to 50% of older people have been reported to have folate deficiency, with higher levels in those institutionalized [<xref ref-type="bibr" rid="B50">50</xref>]. For this reason, there is a sound theoretical reasoning to speculate on a positive beneficial outcome in relation to cognition and boosting memory with choline and choline-containing compounds [<xref ref-type="bibr" rid="B44">44</xref>].</p><p>Elevated total homocysteine (tHcy), a risk factor for many chronic diseases including cognitive decline, can be remethylated to methionine by folate [<xref ref-type="bibr" rid="B51">51</xref>]. Alternatively, tHcy can be metabolized by other 1-carbon nutrients, that is, betaine and its precursor, choline. Elias <italic>et al</italic>. [<xref ref-type="bibr" rid="B48">48</xref>] reported that tHcy levels are inversely associated with visual-spatial organization, working memory, scanning-tracking, and abstract reasoning. Chiuve <italic>et al</italic>. [<xref ref-type="bibr" rid="B52">52</xref>] assessed the association between the dietary intakes of betaine and choline and the concentration of tHcy. They found the total choline + betaine intake to be inversely associated with tHcy. In a double-blind, placebo-controlled clinical trial Olthof <italic>et al</italic>. [<xref ref-type="bibr" rid="B53">53</xref>] investigated the supplementation with soybean PC (PhosChol) on homocysteine plasma concentrations in men with mildly elevated levels. They found that PC was able to reduce homocysteine levels, thus supporting the link between dietary intake of PC and homocysteine plasma levels.</p></sec><sec><title>Gangliosides</title><p>Brain content of specific gangliosides (for example, GM1) has been documented to decrease with age, and a low GM1 content has been observed in the brains of patients with AD [<xref ref-type="bibr" rid="B54">54</xref>]. Exogenously administered gangliosides have been shown to exhibit neurotrophic action, and to increase the release of brain-derived neurotrophic factor (BDNF) <italic>in vitro</italic>[<xref ref-type="bibr" rid="B55">55</xref>]. Experimental data have shown that gangliosides, and in particular, GM1, exhibit properties similar to the neurotrophins. The neurotrophins promote neurogenesis, which is essential for specific cognitive functions that decline in some neurological disorders and in ageing [<xref ref-type="bibr" rid="B56">56</xref>]. A systemic administration of GM1 in rats ameliorated the age-related decreased activity of choline acetyl transferase and choline uptake in the brain of aged rats as well as improved spatial learning and memory tasks in the aged rats [<xref ref-type="bibr" rid="B57">57</xref>]. Dietary gangliosides increase total brain-ganglioside content in rats [<xref ref-type="bibr" rid="B58">58</xref>].</p></sec><sec><title>Sphingomyelin (SM)</title><p>The myelin content of the brain decreases with age and the age-related slowing in cognitive processing speed is associated with myelin integrity in a very healthy elderly population [<xref ref-type="bibr" rid="B59">59</xref>]. Dietary bovine SM contributes to central nervous system (CNS) myelination [<xref ref-type="bibr" rid="B60">60</xref>]. Sphingomyelin is also a source of choline. Clinical trials are yet to be conducted in order to assess the <italic>in vivo</italic> neurocognitive effects of SM in humans.</p></sec></sec><sec><title>Methods/Design</title><p>Phospholipid intervention for cognitive ageing reversal (PLICAR) is a randomized, double-blind, placebo-controlled, three-arm, stratified parallel-groups clinical trial with participants randomized to receive a minimum daily dosage of 2.7 g phospholipids (and minimum daily dose of 300 mg/day PS) from Lacprodan® PL-20 or one of two placebos (milk protein concentration or rice starch) over a 180-day (6-month) period. The maximum daily dose of Lacprodan® PL-20 will be 16 g/day. Participants will be stratified according to age, IQ using Raven’s Progressive Matrices (RPM) [<xref ref-type="bibr" rid="B61">61</xref>] and baseline score on the Wechlser Memory Scale-Revised (WMS-R) [<xref ref-type="bibr" rid="B62">62</xref>].</p></sec><sec><title>Aims and study hypotheses</title><p>The primary objective of the current study is to evaluate the chronic effects of daily Lacprodan® PL-20 supplementation on cognitive performance in a healthy elderly population with age-associated memory impairment (AAMI). The secondary objectives of the study are to investigate mood and CV effects of Lacprodan® PL-20 as well as the <italic>in vivo</italic> mechanisms of action by which Lacprodan® PL-20 may improve cognitive function. To this end, a range of measurement modalities will be employed including assessments of CV function, blood biomarkers, gastrointestinal (GI) microbiota, pharmacogenomics and brain activity assessed by functional magnetic resonance imaging (fMRI), magnetoencephalography (MEG) and diffusion tensor imaging (DTI).</p><p>Performance on the various outcome measures for participants receiving Lacprodan® PL-20 will be compared to the performance of participants receiving (i) an inert placebo powder made of rice starch, and (ii) a milk-protein concentrate without phospholipids that has been matched for protein content and total calories. The reason/rationale for including two comparator treatments (rather than one) is the opportunity to address separate research questions. The first comparison (to inert placebo) addresses the research question of whether Lacprodan® PL-20 as a whole has benefits for cognitive function and other secondary outcomes. The second comparison to a milk-protein concentrate without phospholipids addresses a more specific research question regarding whether benefits to cognition and other outcomes associated with Lacprodan® PL-20 can be attributed specifically to the phospholipid content in the formula. In consideration of previous research demonstrating cognitive benefits associated with dairy components other than phospholipids [<xref ref-type="bibr" rid="B3">3</xref>], the inclusion of a third treatment arm was deemed necessary in order to properly delineate the effects of phospholipids versus other dairy components present in Lacprodan® PL-20.</p><p>On the basis of previous human clinical studies with bovine and plant-derived phospholipids, it is hypothesised that in comparison to the inert placebo, Lacprodan® PL-20 supplementation over 180 days will result in significant improvements on our primary variable, namely a composite memory score computed from the Rey’s Verbal Learning Test (RVLT). We are also exploring the possibly that the treatment may benefit other elements of cognitive performance (including processing speed and global functioning). A second hypothesis is that cognitive benefits of a lesser magnitude will be observed when comparing supplementation with Lacprodan® PL-20 with milk protein concentrate (MPC) without phospholipids placebo over 180 days. Additionally we are exploring the effects of Lacprodan® PL-20 on a number of central, CV and GI biomarkers.</p></sec><sec><title>Study site</title><p>PLICAR will be conducted at the Centre for Human Psychopharmacology, Swinburne University of Technology, Melbourne, Australia.</p></sec><sec><title>Participants</title><p>A total of 150 healthy elderly participants (≥55 years) with AAMI will take part in the study. AAMI is defined on the basis of criteria first outlined by Crook <italic>et al</italic>. [<xref ref-type="bibr" rid="B63">63</xref>,<xref ref-type="bibr" rid="B64">64</xref>]: (i) a score >25 on the Memory Complaint Questionnaire (MAC-Q [<xref ref-type="bibr" rid="B64">64</xref>]) and (ii) a score ≤1 standard deviation below the mean for healthy young adults on the paired associates test from the WMS-R [<xref ref-type="bibr" rid="B62">62</xref>]. Participants will be excluded from participation if they are currently diagnosed with dementia and/or score <24 on the MMSE [<xref ref-type="bibr" rid="B65">65</xref>]; have a neurological, cardiac, endocrine, GI or bleeding disorder; have a psychiatric illness, including moderate-to-severe depression, as defined as a score ≥20 on the Beck Depression Inventory II (BDI-II) [<xref ref-type="bibr" rid="B66">66</xref>]; have a current or previous history of alcoholism and/or substance abuse; have a known or suspected allergy to cow’s milk and/or lactose intolerance; are a current smoker, or are not fluent in the English language. To be eligible, participants also cannot be currently taking any medications or herbal/dietary supplements with known cognitive effects. The study is jointly funded by Arla Foods (Denmark) and Swinburne University of Technology. It was ethically approved by the Swinburne University Human Research Ethics Committee (project number 2012/294) and all participants will provide written informed consent. The trial has been registered with the Australian and New Zealand Clinical Trials Registry (ACTRN12613000347763).</p></sec><sec><title>Procedure</title><p>Eligible participants are required to attend four testing sessions. An overview of the testing sessions is provided in the clinical trial flow chart (Figure <xref ref-type="fig" rid="F1">1</xref>).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Phospholipid intervention for cognitive ageing reversal (PLICAR) protocol flow diagram.</bold> AAMI, age-associated memory impairment; MAC-Q, Memory Complaint Questionnaire; MMSE, Mini Mental State Examination; BDI-II, Beck Depression Inventory II; BMI, body mass index; RPM, Raven’s Progressive Matrices; APOE, Apoliprotein E; MTHFR, Methylenetetrahydrofolate reductase; MRI, magnetic resonance imaging; MEG, magnetoencephalography; MPC, milk protein concentrate; TICS-M telephone interview for cognitive status modified.</p></caption><graphic xlink:href="1745-6215-14-404-1"/></fig><sec><title>Visit 1 (screening/practice)</title><p>During the first visit, voluntary written informed consent is obtained from all participants. Then the participants are further screened for eligibility and administered the MMSE, MAC-Q, WMS-R and BDI-II. A detailed medical history is also taken, a dietary questionnaire is administered [<xref ref-type="bibr" rid="B67">67</xref>,<xref ref-type="bibr" rid="B68">68</xref>] and demographic information is collected, which includes body mass index (BMI), age, educational background and general intelligence (as measured by RPM). All eligible participants are then required to complete practice versions of all the cognitive outcome measures to be used in the study.</p></sec><sec><title>Visit 2 (Baseline)</title><p>The second visit will be scheduled for one week following the screening/practice visit. In preparation for the baseline visit, all participants will be required to collect and store their faecal sample (as per the procedure provided in the faecal sample collection kit) a day before the actual visit. The faecal sample for GI microbiota analysis will be deposited at the Centre for Human Psychopharmacology, Swinburne University, when the participant comes for the baseline visit. The faecal samples will be stored at -80°C until further analysis. Also in preparation, participants will be required to fast from 10 pm the night before. A fasting blood sample will then be taken in order to assess the baseline biochemical measures, together with a separate blood sample, which will be used for Apolipoprotein E (APOE) and <italic>Methylenetetrahydrofolate reductas</italic>e (<italic>MTHFR</italic>) genotyping. Following the blood samples, participants will eat a light breakfast. Half an hour after breakfast, participants will be required to complete all pre-dose cognitive, mood and CV measures. Participants will then be randomized to their treatment group (Lacprodan PL-20® or MPC or rice starch) and consume their first sachet of study treatment dissolved in water. A lunch break of 90 minutes will then follow, before post dose testing on all cognitive, mood and cardiovascular measures. At the conclusion of the baseline study visit, participants will be provided with enough study treatment for the next 90 days.</p></sec><sec><title>Visits 3 and 4 (90 and 180 days)</title><p>The schedule of events for the 90-day and 180-day visits is identical to that for the baseline visit. On the day of each study visit participants will be required to wait until the completion of pre-dose assessments before consuming their daily study treatment.</p></sec><sec><title>Neuroimaging sub-study</title><p>Participants involved in the neuroimaging sub-study will be required to attend two additional sessions, the first in the week preceding their baseline visit and the second in the week preceding their 180-day visit (see Figure <xref ref-type="fig" rid="F1">1</xref>). In the second visit they will be required to consume their study treatment as usual in the morning before the scans. This is due to the fact that the neuroimaging sub-study is only concerned with the chronic effects of PL-20 on brain function.</p></sec></sec><sec><title>Sample size</title><p>Based on the reviewed literature, we predict a small-medium effect size (<italic>f</italic> = 0.14) on the primary variable. The sample size for this study is 150 participants, with 50 participants in each treatment group (Lacprodan® PL-20, MPC or rice starch). Allowing for a 20% drop-out rate over the course of the 180-day testing period, this will give 80% power to detect significant treatment × time interactions from baseline to 180 days for the primary outcome (calculated using G*Power 3.1, with α = 0.05 and <italic>r</italic> = 0.5 for the correlation between repeated measures).</p></sec><sec><title>Treatments</title><p>Lacprodan® PL-20, a powdered MPC rich in phospholipids, is manufactured by Arla Foods Ingredients Group P/S, Viby J, Denmark. The content of individual phospholipids in Lacprodan® PL-20 is displayed in Table <xref ref-type="table" rid="T1">1</xref>.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Phospholipid composition of Lacprodan® PL-20 by percentage and daily dose</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"> </th><th align="left"><bold>Percentage (%)</bold></th><th align="left"><bold>Minimum dose per day</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Sphingomyelin<hr/></td><td align="left" valign="bottom">4.3<hr/></td><td align="left" valign="bottom">688 mg<hr/></td></tr><tr><td align="left" valign="bottom">Phosphatidyl choline (PC)<hr/></td><td align="left" valign="bottom">4.3<hr/></td><td align="left" valign="bottom">688 mg<hr/></td></tr><tr><td align="left" valign="bottom">Phosphatidyl serine (PS)<hr/></td><td align="left" valign="bottom">1.9<hr/></td><td align="left" valign="bottom">304 mg<hr/></td></tr><tr><td align="left" valign="bottom">Phosphatidyl ethanolamine (PE)<hr/></td><td align="left" valign="bottom">3.5<hr/></td><td align="left" valign="bottom">560 mg<hr/></td></tr><tr><td align="left" valign="bottom">Phosphatidyl inositol (PI)<hr/></td><td align="left" valign="bottom">1.3<hr/></td><td align="left" valign="bottom">208 mg<hr/></td></tr><tr><td align="left">Ganglioside and others</td><td align="left">0.7</td><td align="left">112 mg</td></tr></tbody></table></table-wrap><p>Lacprodan® PL-20 will be administered orally at a maximum dose of 16 g/day, providing minimum daily dosages of 2.7 g PL and 300 mg PS. The powder is dissolved in 150 to 200 mL water and drunk once per day with breakfast. Two placebo treatments will also be administered: (i) an inert placebo consisting of rice starch (20 g/day), and (ii) an MPC without phospholipids (Arla Foods); (12 g/day). Both placebo treatments are also administered orally as powders dissolved in ≤250 mL water and matched to Lacprodan® PL-20 for colour and taste in order to ensure treatment blinding.</p></sec><sec><title>Randomization and safety</title><p>Randomization of participants to treatment groups will be determined by random allocation. For the neuroimaging sub-study 60 randomization numbers will be set aside, which correspond to 10 female participants receiving Lacprodan® PL-20, 10 receiving MPC and 10 receiving rice starch placebo, and 10 male participants receiving Lacprodan® PL-20, 10 receiving MPC and 10 receiving rice starch placebo. Blinding for both the main study as well as the neuroimaging sub-study will be conducted by an independent staff member at the Centre for Human Psychopharmacology, who is outside of the project, to code the treatments, and maintain the key to this code until data collection is completed. All potential adverse events will be monitored throughout the trial, with oversight from the Swinburne University of Technology Human Research Ethics committee.</p></sec><sec><title>Primary outcomes</title><p>The primary study outcome is the effect of Lacprodan® PL-20 supplementation on memory as measured using a compound score from the RVLT [<xref ref-type="bibr" rid="B69">69</xref>,<xref ref-type="bibr" rid="B70">70</xref>], similar to that used previously [<xref ref-type="bibr" rid="B71">71</xref>]. The memory score will be derived using the formula:</p><p><disp-formula><mml:math id="M1" name="1745-6215-14-404-i1" overflow="scroll"><mml:mtable columnalign="left"><mml:mtr><mml:mtd><mml:mfenced open="("><mml:mrow><mml:msub><mml:mi mathvariant="normal">Z</mml:mi><mml:mrow><mml:mn>15</mml:mn><mml:mo>‒</mml:mo><mml:mi>Word</mml:mi><mml:mspace width="0.25em"/><mml:mi>Learning</mml:mi><mml:mspace width="0.25em"/><mml:mi>test</mml:mi><mml:mo>‘</mml:mo><mml:mi>total</mml:mi><mml:mspace width="0.25em"/><mml:mi>immediate</mml:mi><mml:mspace width="0.25em"/><mml:mi>recall</mml:mi></mml:mrow></mml:msub><mml:mo>’</mml:mo></mml:mrow></mml:mfenced></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mrow><mml:mspace width="3em"/><mml:mo>+</mml:mo><mml:msub><mml:mi mathvariant="normal">Z</mml:mi><mml:mrow><mml:mn>15</mml:mn><mml:mo>‒</mml:mo><mml:mi>Word</mml:mi><mml:mspace width="0.25em"/><mml:mi>Learning</mml:mi><mml:mspace width="0.25em"/><mml:mi>Test</mml:mi><mml:mo>‘</mml:mo><mml:mi>maximum</mml:mi><mml:mspace width="0.25em"/><mml:mi>immediate</mml:mi><mml:mspace width="0.25em"/><mml:mi>recall</mml:mi><mml:mo>’</mml:mo></mml:mrow></mml:msub></mml:mrow></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mfenced close=")"><mml:mrow><mml:mspace width="3em"/><mml:mo>+</mml:mo><mml:msub><mml:mi mathvariant="normal">Z</mml:mi><mml:mrow><mml:mn>15</mml:mn><mml:mo>‒</mml:mo><mml:mi>Word</mml:mi><mml:mspace width="0.25em"/><mml:mi>Learning</mml:mi><mml:mspace width="0.25em"/><mml:mi>Test</mml:mi><mml:mo>‘</mml:mo><mml:mi>delayed</mml:mi><mml:mspace width="0.25em"/><mml:mi>recall</mml:mi><mml:mo>’</mml:mo></mml:mrow></mml:msub></mml:mrow></mml:mfenced><mml:mo>/</mml:mo><mml:mn>3</mml:mn><mml:mo>.</mml:mo></mml:mtd></mml:mtr></mml:mtable></mml:math></disp-formula></p><p>The RVLT is a test of verbal learning and memory that has a long history of use both in the assessment of clinical memory disturbances as well as cognitive decline associated with normal ageing [<xref ref-type="bibr" rid="B70">70</xref>]. Verbal learning, as measured by the RVLT and similar tests, has previously been found to be sensitive to the effects of phospholipid interventions in AAMI populations [<xref ref-type="bibr" rid="B39">39</xref>,<xref ref-type="bibr" rid="B41">41</xref>]. Similarly, verbal memory has also been found to be sensitive to other nutraceutical interventions such as <italic>Bacopa monnieri</italic>[<xref ref-type="bibr" rid="B72">72</xref>] and folic acid [<xref ref-type="bibr" rid="B71">71</xref>] in elderly populations. For these reasons, the inclusion of the RVLT will enable direct comparison of the efficacy of the milk-derived phospholipids present in Lacprodan® PL-20 to previous cognitive intervention studies in the elderly.</p></sec><sec><title>Secondary outcomes</title><p>A range of secondary outcomes will be used, encompassing cognitive performance, mood, CV, GI microbiota, biochemical, genetic and brain imaging modalities. Secondary outcomes will include other elements of cognitive performance as measured by a battery of well-validated and highly sensitive cognitive tests. Traditional paper-and-pencil neuropsychological tests as well as computerized tasks have been included. These tests will be implemented at baseline, 90 days and 180 days (pre-dose) in order to capture chronic effects. This battery will consist of the MMSE [<xref ref-type="bibr" rid="B65">65</xref>], the Prospective and Retrospective Memory Questionnaire (PRMQ) [<xref ref-type="bibr" rid="B73">73</xref>], RVLT [<xref ref-type="bibr" rid="B69">69</xref>,<xref ref-type="bibr" rid="B70">70</xref>], Spatial Working memory and Contextual Memory tasks from the Swinburne University Computerized Cognitive Ageing Battery (SUCCAB [<xref ref-type="bibr" rid="B74">74</xref>]), rapid visual information processing (RVIP), serial 3 s and 7 s subtraction and the Hick’s reaction time paradigm. In order to capture potential acute cognitive effects associated with Lacprodan® PL-20 supplementation, the SUCCAB tests, RVIP, serial 3 s and 7 s subtraction and Hick’s reaction time will also be administered 90 minutes post dose.</p><p>The MMSE [<xref ref-type="bibr" rid="B65">65</xref>] is a global measure of cognitive function that has been used extensively both as a diagnostic tool for dementia screening as well as a cognitive outcome measure for gauging the efficacy of chronic nutraceutical interventions in elderly participants, for example [<xref ref-type="bibr" rid="B71">71</xref>,<xref ref-type="bibr" rid="B75">75</xref>,<xref ref-type="bibr" rid="B76">76</xref>]. The MMSE was included in the current study due to its widespread use in previous research; however problems with ceiling effects and insensitivity to change amongst high-functioning individuals have been previously well-documented [<xref ref-type="bibr" rid="B77">77</xref>].</p><p>The PRMQ [<xref ref-type="bibr" rid="B73">73</xref>] is a self-report instrument which provides a measure of retrospective as well as prospective memory slips in everyday life. The PRMQ was included in the current study in order to provide an ecologically valid measure of typical memory complaints that may be of concern to elderly individuals. Although most memory complaint questionnaires focus exclusively on failures to remember previous information (retrospective memory), the PRMQ is unique in that it additionally provides a measure of prospective memory failures, which are failures relating to tasks that need to be completed at a certain time (for example, remembering to turn up to an appointment on time) [<xref ref-type="bibr" rid="B73">73</xref>,<xref ref-type="bibr" rid="B78">78</xref>].</p><p>In addition to the use of traditional psychometric tests, the importance of including computerized tests that may accurately gauge the speed of well-differentiated cognitive functions has emerged in recent years [<xref ref-type="bibr" rid="B79">79</xref>]. In the current study the spatial working memory and contextual memory tasks from the SUCCAB have been included due to high degrees of sensitivity to the effects of ageing, as measured using response times [<xref ref-type="bibr" rid="B74">74</xref>]. Similarly, significant reductions in response times on both of these tasks have previously been reported in older participants following chronic nutraceutical interventions [<xref ref-type="bibr" rid="B80">80</xref>,<xref ref-type="bibr" rid="B81">81</xref>]. For the assessment of processing speed, the highly sensitive Hick reaction time paradigm [<xref ref-type="bibr" rid="B82">82</xref>] will be used. For the assessment of cognitive function during increased demand, serial 3 s and serial 7 s subtraction will be assessed together with the RVIP computerized measure of sustained attention. Previous research from our laboratory has found the serial subtraction and RVIP tasks to be particularly sensitive to the acute effects of nutraceutical interventions [<xref ref-type="bibr" rid="B83">83</xref>-<xref ref-type="bibr" rid="B86">86</xref>].</p><sec><title>Mood</title><p>There is evidence to suggest that phospholipid supplementation may have a positive effect on chronic stress as well as mood. Mood improvements have been previously reported in a double-blind trial of PS in depressed patients [<xref ref-type="bibr" rid="B87">87</xref>]. A number of studies have also demonstrated that phospholipids may have anti-stress effects, as demonstrated by lowered levels of adrenocorticotropic hormone (ACTH), reduction in perceived stress ratings in response to acute stress [<xref ref-type="bibr" rid="B88">88</xref>] and reduced cortisol release in response to acute stress [<xref ref-type="bibr" rid="B89">89</xref>,<xref ref-type="bibr" rid="B90">90</xref>]. In relation to milk-based phospholipids, it was recently demonstrated that chronic supplementation may lead to increased morning cortisol availability in chronically stressed men [<xref ref-type="bibr" rid="B91">91</xref>] as well as a blunting of self-report stress ratings in response to an acute stressor [<xref ref-type="bibr" rid="B43">43</xref>].</p><p>Chronic changes in mood over the course of the trial will be assessed using the Depression, Anxiety and Stress Scale (DASS) [<xref ref-type="bibr" rid="B92">92</xref>], the Profile of Mood States (POMS) [<xref ref-type="bibr" rid="B93">93</xref>] and Bond-Lader visual analogue mood scales [<xref ref-type="bibr" rid="B94">94</xref>]. These measures will be administered at baseline, 90 and 180 days pre-dose. The Bond-Lader scales will additionally be administered post dose at each study visit in order to capture potential acute mood effects associated with Lacprodan® PL-20 treatment. The Bond-Lader scales have previously been used by our group in a wide range of acute and chronic intervention studies, and have been found to display excellent sensitivity to subtle affective changes.</p></sec><sec><title>Cardiovascular assessment</title><p>There is evidence to suggest that increases in arterial stiffness with ageing, which are reflected in measures of blood flow velocity, may be a contributing factor in cognitive decline [<xref ref-type="bibr" rid="B95">95</xref>,<xref ref-type="bibr" rid="B96">96</xref>]. Previous research suggests that milk proteins may increase insulin secretion, as well as help to reduce blood pressure and plasma cholesterol levels [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B97">97</xref>]. Further, high levels of B12 present in milk may also help to lower HCy levels, which are a contributing factor to CV disease [<xref ref-type="bibr" rid="B98">98</xref>]. In light of the fact that Lacprodan® PL-20 is an MPC, it could be argued that chronic supplementation may have a positive influence on CV function. By this reasoning, the inclusion of CV parameters in the current study will enable exploration of whether improvements to CV function are a mechanism by which with Lacprodan® PL-20 may achieve cognitive benefits.</p><p>CV function will be assessed using brachial blood pressure, aortic blood pressure, carotid-femoral pulse wave velocity (PWV) as well as blood flow velocity in the medial carotid artery and the common carotid artery (CCA). Brachial blood pressure will be calculated with the participant seated and following a five-minute rest period using a clinically validated automated sphygmomanometer. Aortic blood pressure, pulse pressure and PWV (all aspects of arterial stiffness and CV pressures) will be measured non-invasively using the SphygmoCor device. Applanation tonometry of the radial artery will be used to estimate aortic pressures and wave reflections, and applanation of the carotid and femoral arteries will be used to measure PWV. A non-invasive transcranial Doppler system will be used to record middle cerebral artery (MCA) blood velocity by placing a sensor close to the participant’s ear and common carotid artery (CCA) blood velocity will be recorded by placing a hand-held sensor at the base of the participant’s neck. All CV measures will be assessed pre- and post dose at the baseline, 90-day and 180-day study visits in order to capture both acute and chronic effects associated with Lacprodan® PL-20 supplementation.</p></sec><sec><title>Gastrointestinal microbiota</title><p>In recent years research in the field of GI microbiota has caught major interest. Research is suggesting that modifications in the composition of the GI microbiota influence normal physiological functions and contribute to diseases ranging from inflammation to diabetes. Collectively studies now indicate that the gut microbiota also communicates with the CNS possibly through immune, neural and endocrine pathways, and by these means influences gut-brain communication, brain function and even behaviour [<xref ref-type="bibr" rid="B99">99</xref>-<xref ref-type="bibr" rid="B101">101</xref>]. Studies on germ-free animals and animals exposed to pathogenic bacterial infections, probiotic bacteria or antibiotics, suggest a role of GI microbiota in the regulation of cognition, anxiety and mood [<xref ref-type="bibr" rid="B101">101</xref>-<xref ref-type="bibr" rid="B103">103</xref>]. Moreover GI microbiota perform many important functions like protection, immune development and metabolism, which together have an enormous effect on host nutrition and health condition [<xref ref-type="bibr" rid="B104">104</xref>-<xref ref-type="bibr" rid="B106">106</xref>]. Previous studies suggest that human and bovine milk proteins prevent the adhesion and colonisation of pathogenic bacteria in the GI tract [<xref ref-type="bibr" rid="B107">107</xref>-<xref ref-type="bibr" rid="B109">109</xref>] and promote the growth of beneficial bacteria [<xref ref-type="bibr" rid="B110">110</xref>]. As knowing that GI microbiota have several physiological functions in the human health condition, it can be influenced by Lacprodan® PL-20 milk protein supplementation. Therefore, it will be valuable to study the GI microbiota at different time points across the clinical trial to identify the effect of Lacprodan® PL-20 on indigenes microbial community.</p><p>Faecal samples will be collected for the GI microbiota analysis at baseline, 90 days and 180 days to explore the possible effect of Lacprodan® PL-20 on the microbial composition. The microbiota analysis will be carried out by utilising deep next-generation shotgun sequencing [<xref ref-type="bibr" rid="B111">111</xref>] of DNA extracted from collected faecal samples. This analysis will provide insight into GI microbiota of the ageing population and also functional characterisation will provide understanding of the potential mechanism by which Lacprodan® PL-20 may influence age-related cognitive decline (ARCD).</p></sec><sec><title>Biochemical assessment</title><p>Haematological testing will be conducted at baseline, 90 days and 180 days in order to further investigate possible mechanisms by which Lacprodan® PL-20 may influence cognitive decline. These measures have been chosen on the basis of current aetiological understanding of brain ageing as well as proposed <italic>in vivo</italic> actions of Lacprodan® PL-20 constituents.</p><p>Previous research has demonstrated that administration of PC, a major phospholipid component of Lacprodan® PL-20, can increase the plasma choline as well as brain acetylcholine (Ach) supply [<xref ref-type="bibr" rid="B112">112</xref>,<xref ref-type="bibr" rid="B113">113</xref>]. Further, elevated levels of the neurotoxic substance HCy have been found to be a risk factor for cognitive decline [<xref ref-type="bibr" rid="B48">48</xref>,<xref ref-type="bibr" rid="B114">114</xref>-<xref ref-type="bibr" rid="B116">116</xref>]. Improved HCy levels have been found to result from increased intake of PC and choline [<xref ref-type="bibr" rid="B52">52</xref>,<xref ref-type="bibr" rid="B53">53</xref>]. Vitamin B12, which is present in high quantities in dairy products [<xref ref-type="bibr" rid="B117">117</xref>], has also been found to be effective in reducing HCy in elderly populations [<xref ref-type="bibr" rid="B118">118</xref>]. For these reasons plasma choline as well as B Vitamins and HCy levels will be monitored throughout the study.</p><p>Other major contributors to brain ageing are oxidative stress [<xref ref-type="bibr" rid="B119">119</xref>,<xref ref-type="bibr" rid="B120">120</xref>] and inflammation [<xref ref-type="bibr" rid="B121">121</xref>]. The phospholipids PC and PS have both been found to display anti-inflammatory and antioxidant properties, inhibiting microglial activation as well as superoxide and nitric oxide production [<xref ref-type="bibr" rid="B122">122</xref>]. Endogenous antioxidant glutathione (GSH) is the most abundant antioxidant in human cells, which plays a central role in defence from oxidative stress [<xref ref-type="bibr" rid="B123">123</xref>]. Under normal physiological conditions the ratio of reduced GSH to oxidized glutathione (glutathione disulphide, GSSG) is as high as 100:1. However, in cases of increased oxidative stress the ratio changes due to increased levels of GSSG or decreased levels of reduced GSH. For this reason peripheral blood levels of GSH as well as the ratio of GSH/GSSG in the blood are good measures of oxidative stress, and have been found to be altered in patients with mild cognitive impairment and AD [<xref ref-type="bibr" rid="B124">124</xref>]. Another widely-used biomarker of oxidative stress is F2-isoprostane, which is an excellent <italic>in vivo</italic> measure of lipid peroxidation [<xref ref-type="bibr" rid="B125">125</xref>]. Peripheral blood plasma levels of F2 isoprostane have been found to be significantly elevated in mild cognitive impairment [<xref ref-type="bibr" rid="B126">126</xref>]. A previous study from our laboratory in elderly participants found plasma F2 isoprostane levels to significantly decline following a 3-month intervention with the antioxidant Pycnogenol [<xref ref-type="bibr" rid="B127">127</xref>]. In addition to these measures of oxidative stress, serum measures of inflammation will be provided using the following inflammatory biomarkers: TNF-α, IL-1β, IL-6 and C-reactive protein (CRP). Peripheral levels of these inflammatory biomarkers have previously been found to be elevated in cases of mild cognitive impairment and AD in comparison to healthy age-matched controls [<xref ref-type="bibr" rid="B128">128</xref>,<xref ref-type="bibr" rid="B129">129</xref>].</p></sec><sec><title>Genetic</title><p>A separate blood sample will be collected pre-randomization for the analysis of single nucleotide polymorphisms (SNP) in the <italic>APOE</italic> and the <italic>MTHFR</italic> genes. The <italic>APOE</italic>-ϵ4 allele has been found to be associated with an increased risk of developing AD as well as cognitive decline in normal elderly [<xref ref-type="bibr" rid="B130">130</xref>,<xref ref-type="bibr" rid="B131">131</xref>]. Testing for allelic differences in the <italic>APOE</italic> gene was included in the current study in order to determine whether these genetic differences may affect the efficacy of Lacprodan® PL-20 as a treatment for AAMI. MTHR is an important enzyme involved in the metabolism of HCy. The <italic>MTHFR</italic> 677 T allele is associated with reduced enzymatic activity, which results in decreased serum and plasma levels of folate as well as increased plasma levels of HCy [<xref ref-type="bibr" rid="B132">132</xref>]. In consideration of the relationship between levels of phospholipids, vitamin B12 and HCy [<xref ref-type="bibr" rid="B53">53</xref>,<xref ref-type="bibr" rid="B118">118</xref>], genetic testing for allelic differences in the <italic>MTHFR</italic> gene was included in the current study in order to assess whether this may also affect the efficacy of Lacprodan® PL-20 as an AAMI treatment.</p></sec><sec><title>Neuroimaging</title><p>Neuroimaging with fMRI and MEG will be conducted in a subset of 60 participants in order to further explore the <italic>in vivo</italic> mechanisms of action of Lacprodan® PL-20 in the brain. Previous neuroimaging studies using PS supplementation in AD have been conducted using electroencephalography (EEG) as well as positron emission tomography (PET) [<xref ref-type="bibr" rid="B133">133</xref>-<xref ref-type="bibr" rid="B135">135</xref>]. PET results revealed that for the PS group there was increased glucose metabolism during a visual recognition task across a number of brain regions, most notably the temperoparietal regions [<xref ref-type="bibr" rid="B133">133</xref>]. However, to date no further neuroimaging studies have been conducted using phospholipid interventions, and to the best of our knowledge none have been conducted using MRI or MEG.</p><p>In the current study structural and functional MRI scans will be acquired using a Siemens 3 Tesla Tim Trio MRI scanner (Erlangen, Germany), located at the Centre for Human Psychopharmacology, Swinburne University of Technology. During the initial scan, a structural image will be obtained for each participant and used as a reference point for further functional scans. Scanning for DTI analysis, a measure of white matter integrity, will also be conducted. Following DTI there will be scanning in a resting state in order to assess activity in the default mode network (DMN) for approximately 6 minutes. Additional analysis of cell membrane fluidity will also be conducted by using the T2 signal timing information (relaxometry) while in a resting state. Changes in the blood oxygenation-level dependent (BOLD) signal will also be analysed while participants complete in-scanner versions of verbal episodic memory (approximately 20 minutes) and N-Back working memory tasks (approximately 20 minutes).</p><p>MEG scanning will be conducted using an Elekta Neuromag® TRIUX 306-Channel Magnetometer system (Helsinki, Finland) MEG system, also located at the Centre for Human Psychopharmacology, Swinburne University of Technology. Initial scanning while in a resting state will be conducted in order to collect information as to activity in the DMN. Following this scanning will be conducted whilst participants complete the same in-scanner tasks as used in the fMRI task: verbal episodic memory and N-Back working memory. The two tasks are kept the same across both fMRI and MEG in order for information from the two imaging modalities to be combined into a single comprehensive analysis. MEG scanning provides important complementary information, which is additional to that provided by fMRI. The temporal resolution of MEG is far superior to fMRI; MEG is capable of recording neural oscillations from delta right through to the gamma range (>40 Hz). Although the spatial resolution of MEG is less than that of fMRI, the high number of sensors (approximately 300), together with modern source reconstruction algorithms (for example, beam forming) means that the spatial resolution of MEG is far superior to conventional scalp-recorded EEG [<xref ref-type="bibr" rid="B135">135</xref>]. The combination of the two imaging modalities is state-of-the-art and will provide an unparalleled level of analysis of the effects of Lacprodan® PL-20 on memory function.</p><p>All primary and secondary outcome measures are displayed in Table <xref ref-type="table" rid="T2">2</xref>.</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Summary of PLICAR outcome measures by visit</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"> </th><th align="left"><bold>Measures</bold></th><th align="left"><bold>V1</bold></th><th align="left"><bold>V2</bold></th><th align="left"><bold>V3</bold></th><th align="left"><bold>V4</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Screening<hr/></td><td align="left" valign="bottom">Written informed consent<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Demographics<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Medical history<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Inclusion/exclusion criteria<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Concomitant medications<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Adverse events<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Dietary questionnaire<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Cognitive<hr/></td><td align="left" valign="bottom">Memory Complaint Questionnaire (MAC-Q)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Wechsler Memory Scale Revised (WMS-R)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Raven’s Progressive Matrices (RPM)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Mini Mental State Exam (MMSE)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Rey Verbal Learning Test (RVLT)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">SUCCAB Spatial Working Memory<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">SUCCAB Contextual Memory<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Prospective and Retrospective Memory (PRMQ)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Rapid visual information processing (RVIP)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Serial 3 s and 7 s subtraction<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Jensen box task<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Mood<hr/></td><td align="left" valign="bottom">Beck Depression Inventory (BDI-II)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Depression, Anxiety and Stress Scale (DASS)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Profile of Mood States (POMS)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Bond-Lader visual analogue mood scales<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Cardiovascular<hr/></td><td align="left" valign="bottom">Brachial blood pressure<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">SphygmoCor (Aortic blood pressure, pulse pressure, PWV)<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Blood velocity (MCA and CCA)<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">GI microbiota<hr/></td><td align="left" valign="bottom">Intestinal bacteria<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Biochemical<hr/></td><td align="left" valign="bottom">Oxidative stress (Glutathione and F2 isoprostanes)<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Inflammation (TNF-α, IL-1β, IL-6 and CRP)<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">B Vitamins (B6, B9 and B12)<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Homocysteine (HCy)<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Glucoregulation<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Serum choline<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Genetic<hr/></td><td align="left" valign="bottom">Apolipoprotein E and MTHFR<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Brain imaging<hr/></td><td align="left" valign="bottom">Structural magnetic resonance imaging (MRI)<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Diffusion tensor imaging (MRI)<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Default mode network activation (MEG and MRI)<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Relaxometry (MRI)<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Episodic Memory task (functional MRI and MEG)<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left"> </td><td align="left">N-Back working memory task (functional MRI and MEG)</td><td align="left"> </td><td align="left">X</td><td align="left"> </td><td align="left">X</td></tr></tbody></table><table-wrap-foot><p>V1, screening/practice session; V2, baseline visit; V3, 9-day visit; V4, 180-day visit; GI, gastrointestinal; SUCCAB, Spatial Working memory and Contextual Memory tasks from the Swinburne University Computerized Cognitive Ageing Battery; PWV, pulse wave velocity; MCA, medial carotid artery; CCA, common carotid artery; CRP, C-reactive protein; MTHFR, Methyltetrahydrofolate reductase; MEG, magnetoencephalography.</p></table-wrap-foot></table-wrap></sec></sec><sec><title>Analysis</title><p>The primary analysis will investigate the effect of treatment on all cognitive outcomes from baseline to 180 days, using the groups as randomized (intention to treat). Statistical analyses will be conducted using linear mixed modelling, whereby subject-specific random intercepts and slopes will be fitted to subject data and fixed effects will be fitted to treatment group, time and the treatment × time interaction. On the basis of <italic>APOE</italic> and <italic>MTHFR</italic> genotyping, subgroup analysis will also be conducted in order to investigate the effect of allelic differences on treatment response. Secondary outcome variables will be analysed using similar statistical techniques. Results will be considered statistically significant at an alpha level of <italic>P</italic> <0.05 corrected for multiple comparisons.</p><p>Although stratification according to age, intelligence and baseline WMS-R scores may help to explain some of the residual between-group variance unrelated to the treatment effect, further exploration of possible covariates will also be investigated. Baseline correlations between the primary cognitive outcome measures and other baseline variables, including BMI, educational background, diet, CV function, GI microbiota and biochemical parameters, will also be investigated in order to investigate other important covariates. In the event that significant correlations at the <italic>P</italic> <0.05 level are found at baseline then these additional variables will also be controlled for in the primary analysis of cognitive outcomes.</p><p>Analysis of functional neuroimaging data (both MEG and fMRI) during episodic memory and N-Back working memory tasks will be conducted using a region of interest (ROI) approach. Using this method, between-group (Lacprodan® PL-20 versus MPC and inert placebo) functional differences in predefined brain regions will be statistically analysed. The ROIs for the episodic memory task will include the medial temporal lobes, the lateral prefrontal cortices, the associative temporal and paretial regions, the cingulate gyrus and the cerebellum. The ROIs that will be analysed in the N-Back working memory task will include the dorsolateral, ventrolateral and medial prefrontal cortex, anterior cingulate, parietal cortex and sensorimotor cortex [<xref ref-type="bibr" rid="B136">136</xref>].</p></sec><sec><title>Trial status</title><p>The trial is currently recruiting.</p></sec></sec><sec><title>Abbreviations</title><p>AAMI: Age-associated memory impairment; AD: Alzheimer’s disease; APOE: Apolipoprotein E; ARCD: Age-related cognitive decline; BDI-II: Beck depression inventory II; BMI: Body mass index; CCA: Common carotid artery; CNS: Central nervous system; CRP: C-reactive protein; V: Cardiovascular; DASS: Depression anxiety and stress scale; DNM: Default mode network; DTI: Diffusion tensor imaging; EEG: Electroencephalography; fMRI: Functional magnetic resonance imaging; GI: Gastrointestinal; GSH: Glutathione; GSSG: Glutathione disulphide; HCy: Homocysteine; L: Interleukin; MAC-Q: Memory complaint questionnaire; MCA: Middle cerebral artery; MEG: Magnetoencephalography; MPC: Milk protein concentrate; MTHFR: Methylenetetrahydrofolate reductase; MSE: Mini mental state Examination; PC: Phosphatidylcholine; PET: Positron emission tomography; PL: Phospholipid; PLICAR: Phospholipid intervention for cognitive ageing reversal; POMS: Profile of mood states; PRMQ: Prospective and retrospective memory questionnaire; PS: Phosphatidylserine; PWV: Pulse wave velocity; ROI: Region of interest; RPM: Raven’s progressive matrices; RVIP: Rapid visual information processing; RVLT: Rey’s verbal learning test; SB: Soybean; SUCCAB: Spatial working memory and contextual memory tasks from the Swinburne university computerized cognitive ageing battery; tHcy: Total homocysteine; TICS-M: Telephone interview for cognitive status – modified; TNF: Tumour necrosis factor; WMS-R: Wechlser memory scale-revised.</p></sec><sec><title>Competing interests</title><p>AS and CS receive research funding from the food industry, PF is an employee of Arla Foods. The other authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>AS conceived the study, participated in its design and contributed to drafting the manuscript, DC developed the study design and drafted the manuscript, MH conceived and has responsibility for the fMRI component of the study, WW conceived and has responsibility for the MEG component of the study, CS participated in the study design and contributed to the manuscript, DW participated in the study design and contributed to the manuscript, SG developed the protocol and has responsibility for the microbiota component of the study, PF participated in the study design and contributed to the manuscript. All authors read and approved the final manuscript.</p></sec>
Surgical trials and trial registers: a cross-sectional study of randomized controlled trials published in journals requiring trial registration in the author instructions	<sec><title>Background</title><p>Trial registration and the reporting of trial results are essential to increase transparency in clinical research. Although both have been strongly promoted in recent years, it remains unclear whether they have been successfully implemented in surgery and surgery-related disciplines. In this cross-sectional study, we assessed whether randomized controlled trials (RCTs) published in surgery journals requiring trial registration in their author instructions were indeed registered, and whether the study results of registered RCTs had been submitted to the trial register and were thus publicly available.</p></sec><sec><title>Methods</title><p>The ten highest ranked surgery journals requiring trial registration by impact factor (Journal Citation Reports, JCR, 2011) were chosen. We then searched MEDLINE (in PubMed) for RCTs published in the selected journals between 1 June 2012 and 31 December 2012. Any trials recruiting participants before 2004 were excluded because the International Committee of Medical Journal Editors (ICMJE) first proposed trial registration in 2004. We then searched the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) to assess whether the identified RCTs were indeed registered and whether the results of the registered RCTs were available in the register.</p></sec><sec><title>Results</title><p>The search retrieved 588 citations. Four hundred and sixty references were excluded in the first screening. A further 25 were excluded after full-text screening. A total of 103 RCTs were finally included. Eighty-five of these RCTs (83%) could be found via the ICTRP. For 7 of 59 (12%) RCTs, which were registered on ClinicalTrials.gov, summary study data had been posted in the results database.</p></sec><sec><title>Conclusions</title><p>Although still not fully implemented, trial registration in surgery has gained momentum. In general, however, the submission of summary study data to ClinicalTrials.gov remains poor.</p></sec>	<contrib contrib-type="author" id="A1"><name><surname>Hardt</surname><given-names>Julia LS</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>julia.hardt@googlemail.com</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Metzendorf</surname><given-names>Maria-Inti</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>maria-inti.metzendorf@medma.uni-heidelberg.de</email></contrib><contrib contrib-type="author" corresp="yes" id="A3"><name><surname>Meerpohl</surname><given-names>Joerg J</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>meerpohl@cochrane.de</email></contrib>	Trials	<sec><title>Background</title><p>Selective reporting of study results distorts the body of evidence available for clinical decision making. In recent years, several guidelines and recommendations have been published in order to increase transparency of scientific research. One of the key policy initiatives of this development towards more transparency was the call for the obligatory registration of all clinical trials in public trial registers. In September 2004, the International Committee of Medical Journal Editors (ICMJE) proposed 'comprehensive trial registration as a solution to the problem of selective awareness’. In order to advance this goal, the ICMJE decided to require the registration in a public trials register as a mandatory condition for the consideration for the publication of a study report. The ICMJE trial registration requirement policy applies to all trials which started patient recruitment beginning 1 July 2005, and was adopted by all ICMJE member journals [<xref ref-type="bibr" rid="B1">1</xref>]. Since then, many other journals in addition to the ICMJE member journals have adopted the ICMJE policy.</p><p>The implementation of comprehensive mandatory trial registration would allow scientists, clinicians, and study participants to track trials and prevent biased reporting, for example the non-reporting of trials with negative or inconclusive results. Therefore, even unfavorable trial results would not be lost to the pool of medical knowledge. Based on the information provided by the register, prospective investigators would be able to formulate new research questions, plan new trials to fill the gaps in the knowledge base, and avoid unnecessary duplications [<xref ref-type="bibr" rid="B2">2</xref>]. Furthermore, public electronic access to all trials could inspire researchers to collaborate and could also support trial recruitment [<xref ref-type="bibr" rid="B3">3</xref>].</p><p>In April 2007, the ICMJE expanded the definition of clinical trials that have to be registered by adopting the World Health Organization (WHO) definition of a clinical trial, which also includes preliminary trials (phase I trials). The deadline for the implementation of these modifications was 1 July 2008 [<xref ref-type="bibr" rid="B4">4</xref>]. One month after the ICMJE’s expansion of the definition, in May 2007, the WHO launched its International Clinical Trials Register Platform (ICTRP) in order to offer an international portal for identifying, deduplicating, and searching trials from registers all over the world. The ICTRP requires a minimum trial registration data set consisting of 20 items [<xref ref-type="bibr" rid="B2">2</xref>], which is also supported by the ICMJE [<xref ref-type="bibr" rid="B5">5</xref>]. As of 1 July 2007, the member journals of the Surgery Journal Editors Group (SJEG) require registration of all prospective clinical trials prior to the enrollment of the first patient [<xref ref-type="bibr" rid="B6">6</xref>]. Trials which had started recruitment before the deadline had to register before editorial review. Manuscripts are now required to specify the registration number in the abstract [<xref ref-type="bibr" rid="B7">7</xref>].</p><p>But how did all these policy recommendations, regulations, and statements influence the practice of trial registration? By examining the development and growth of ClinicalTrials.gov, the largest public trial register, which was created as a result of the Food and Drug Administration Modernization Act (FDAMA) in 1997, it can be concluded that trial registration has gained momentum and that there has been major progress within the last decade. As of 13 November 2013, ClinicalTrials.gov included more than 154,000 studies from across all 50 American states and 185 countries worldwide (<ext-link ext-link-type="uri" xlink:href="http://www.clinicaltrials.gov/ct2/resources/trends">http://www.clinicaltrials.gov/ct2/resources/trends</ext-link>).</p><p>In September 2008, the ClinicalTrials.gov results database was launched to meet the requirement in Section 801 of the Food and Drug Administration Amendments Act (FDAAA 801) that study sponsors or principal investigators report basic results for 'applicable clinical trials’ (ACTs). A trial is considered 'applicable’ if it meets the following criteria: phase II to IV interventional study involving drugs or medical devices regulated by the FDA; at least one site in the USA; and initiated or ongoing as of 27 September 2007, or later [<xref ref-type="bibr" rid="B8">8</xref>]. For all applicable trials, the results have to be submitted no later than 12 months after the trial’s completion date (<ext-link ext-link-type="uri" xlink:href="http://www.clinicaltrials.gov/ct2/manage-recs/fdaaa">http://www.clinicaltrials.gov/ct2/manage-recs/fdaaa</ext-link> - WhenDoINeedToRegister). The summary results data in the results database are presented mainly in a tabular format and are publicly accessible. This does not only benefit researchers and journal editors but also patients and the general public. The main objectives of the ClinicalTrials.gov results database are to reduce publication bias and selective outcome reporting and to promote complete reporting by structured data entry [<xref ref-type="bibr" rid="B9">9</xref>].</p><p>Although the facts and numbers presented above indicate a remarkable success of the initiatives and efforts to promote trial registration and results reporting, it remains unclear whether these have been successfully implemented as integral parts of clinical research in surgery and surgery-related disciplines. We therefore chose to explore whether randomized controlled trials (RCTs) published in the ten highest ranked (by impact factor) surgery journals that require trial registration in their author instructions were indeed registered. We also chose to address the question of whether the study results of the registered RCTs were publicly available on the trial register website.</p></sec><sec sec-type="methods"><title>Methods</title><p>We accessed the Journal Citation Reports (JCR) Science Edition 2011 on 8 January 2013. Two authors (JH and MIM) independently identified the first ten journals among the top surgery journals by impact factor that required trial registration in the author instructions on their websites (Table <xref ref-type="table" rid="T1">1</xref>). All journals not explicitly requiring trial registration in their author instructions were excluded. We intentionally chose the ten journals with the highest impact factors assuming that their policies and publishing practices would meet the current highest standards. Moreover, we expected that trialists publishing in such top-class journals would be more likely to act in an exemplary manner with regard to trial registration and results reporting.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>The first ten journals among the top surgery journals by impact factor that required trial registration in the author instructions</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="left"><bold>Journal</bold></th><th align="center"><bold>Country</bold></th><th align="center"><bold>Impact factor (JCR 2011)</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom"><italic>Annals of Surgery</italic><hr/></td><td align="center" valign="bottom">USA<hr/></td><td align="center" valign="bottom">7.492<hr/></td></tr><tr><td align="left" valign="bottom"><italic>American Journal of Transplantation</italic><hr/></td><td align="center" valign="bottom">USA<hr/></td><td align="center" valign="bottom">6.394<hr/></td></tr><tr><td align="left" valign="bottom"><italic>Endoscopy</italic><hr/></td><td align="center" valign="bottom">Germany<hr/></td><td align="center" valign="bottom">5.210<hr/></td></tr><tr><td align="left" valign="bottom"><italic>Journal of Neurology, Neurosurgery & Psychiatry</italic><hr/></td><td align="center" valign="bottom">UK<hr/></td><td align="center" valign="bottom">4.764<hr/></td></tr><tr><td align="left" valign="bottom"><italic>British Journal of Surgery</italic><hr/></td><td align="center" valign="bottom">UK<hr/></td><td align="center" valign="bottom">4.606<hr/></td></tr><tr><td align="left" valign="bottom"><italic>Journal of the American College of Surgeons</italic><hr/></td><td align="center" valign="bottom">USA<hr/></td><td align="center" valign="bottom">4.549<hr/></td></tr><tr><td align="left" valign="bottom"><italic>Archives of Surgery</italic> (<italic>JAMA Surgery</italic> since 1 January 2013)<hr/></td><td align="center" valign="bottom">USA<hr/></td><td align="center" valign="bottom">4.422<hr/></td></tr><tr><td align="left" valign="bottom"><italic>Surgical Endoscopy</italic><hr/></td><td align="center" valign="bottom">Germany<hr/></td><td align="center" valign="bottom">4.013<hr/></td></tr><tr><td align="left" valign="bottom"><italic>Transplantation</italic><hr/></td><td align="center" valign="bottom">USA<hr/></td><td align="center" valign="bottom">4.003<hr/></td></tr><tr><td align="left"><italic>Surgery for Obesity and Related Diseases</italic></td><td align="center">USA</td><td align="center">3.929</td></tr></tbody></table></table-wrap><p>The <italic>American Journal of Surgical Pathology</italic> (impact factor 4.352) and <italic>Annals of Surgical Oncology</italic> (impact factor 4.166) were excluded because the author instructions of these journals did not require trial registration. JCR, Journal Citation Reports.</p><p>In a second step, MEDLINE was searched via PubMed for RCTs published in these journals between 1 June 2012 and 31 December 2012. All of the included journals are fully indexed in the MEDLINE database. For the identification of RCTs in MEDLINE (search conducted 15 February 2013), we applied the sensitivity-maximizing Cochrane Highly Sensitive Search Strategy for identifying RCTs in the PubMed format [<xref ref-type="bibr" rid="B10">10</xref>]. The search strategy was slightly modified to fit the surgical setting (Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>).</p><p>One author (JH) screened titles and abstracts, excluded clearly irrelevant references, and downloaded full-texts of all potentially relevant citations. Then, two authors (JH and MIM) independently screened the full-texts and excluded non-randomized studies. Trials recruiting participants prior to 2004 were also excluded because the ICMJE first proposed comprehensive trial registration in a statement published in September 2004, which requires registration in a public trials register for trials that started enrollment after 1 July 2005 only. All discrepancies were resolved by re-examining the full-texts and in discussion with a third author (JM).</p><p>In the next step, two authors (JH and MIM) independently searched the ICTRP (<bold>27</bold> February 2013) for information on the registration of all included RCTs. Trials were searched for either with the registration identification quoted in the publication or, if this identification was not provided, with different keywords describing the topic of the trial. We searched using the ICTRP’s standard search form, which searches within the title, primary sponsor, health condition(s), intervention(s), countries of recruitment, main identification, and secondary identification(s) of the trial data. Additionally, we used Boolean operators to broaden or narrow the search. If such a search was not successful, we took a more sensitive approach by searching first for the main author’s name, second for the most specific terms of the institutional name stated in the authors’ affiliations, or third for the country or city the trial was conducted in combined with one or two specific terms describing the trial. Trials not found through these extensive ICTRP searches were considered to be unregistered.</p><p>Reviewing the full-texts and the information given on the trials register website, we collected data regarding the following topics and parameters: sample size, country of main investigator, national versus multinational and monocenter versus multicenter setting, surgical subspecialty, study objective, and start of patient recruitment. Furthermore, we extracted information on whether trial registration was explicitly mentioned in the article, meaning whether the article included at least one full sentence describing that the trial was registered in a specific trial register. We also reviewed whether the registration number was specified in the title, abstract, or main text, since the SJEG member journals as well as the ICMJE demand specifications of the trial registration number in the abstract as evidence of registration [<xref ref-type="bibr" rid="B7">7</xref>]. Moreover, the ICMJE even recommends that authors list the trial registration number the first time a trial acronym is used in the manuscript (<ext-link ext-link-type="uri" xlink:href="http://www.icmje.org/publishing_10register.html">http://www.icmje.org/publishing_10register.html</ext-link>).</p><p>Finally, the primary trial registers were checked for study results of the registered RCTs. We defined study results as either a citation to a publication reporting the trial or, in the case that ClinicalTrials.gov was the primary register, if aggregate summary data were provided in addition to a citation. Two authors (JH and MIM) extracted the following information for the registered RCTs: primary register; link to a PubMed citation or list of publication(s) provided by the investigators; and, if ClinicalTrials.gov was the primary register: aggregate summary study data posted in the ClinicalTrials.gov results database; automatic link to a PubMed citation mapped via the ClinicalTrials.gov identifier (NCT number); and study start date and study registration date ('study first received’ in ClinicalTrials.gov) in order to identify whether trials were registered retrospectively or prospectively.</p><p>We did not explicitly assess if the RCTs registered in ClinicalTrials.gov were ACTs according to the FDAAA 801. Ethics approval was not required for this study.</p></sec><sec sec-type="results"><title>Results</title><p>There were 199 journals in the subject category 'surgery’ indexed in JCR 2011. The first ten surgery journals with the highest impact factors ranging from 7.492 to 3.929, which explicitly required trial registration in their online author instructions, were chosen (Table <xref ref-type="table" rid="T1">1</xref>).</p><p>The search for RCTs published between 1 June 2012 and 31 December 2012 retrieved 588 citations. From these, 460 clearly irrelevant references were excluded by title or abstract screening. We then evaluated the full-texts of the remaining 128 references (Additional file <xref ref-type="supplementary-material" rid="S2">2</xref>) and excluded 25 of these for the following reasons: 21 studies had started patient recruitment prior to 2004, two citations reported sub-studies of older RCTs, one reported a non-randomized trial, and one was a study not concordant with the WHO definition of a clinical trial (<ext-link ext-link-type="uri" xlink:href="http://www.who.int/ictrp/en/">http://www.who.int/ictrp/en/</ext-link>). This last study had investigated whether there were any differences in the learning outcomes of healthy participants who had trained to proficiency on low- or high-fidelity laparoscopic surgical simulators. The remaining 103 citations were further investigated. The majority of the RCTs (n = 86; 83%) had initiated patient recruitment during or after 2006, 15 RCTs (15%) had started to recruit participants in 2004 or 2005, and two RCTs (2%) did not state the start of recruitment (Figure <xref ref-type="fig" rid="F1">1</xref>).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Study flow diagram: selection process of RCTs.</bold> RCT, randomized controlled trial.</p></caption><graphic xlink:href="1745-6215-14-407-1"/></fig></sec><sec><title>Trial registration</title><p>Eighty-five of the 103 analyzed RCTs (83%) could be identified in the ICTRP (Table <xref ref-type="table" rid="T2">2</xref>; Additional file <xref ref-type="supplementary-material" rid="S3">3</xref>). Of these 85 RCTs, 15 (18%) had been registered prospectively, 45 (53%) had been registered retrospectively, and 21 (25%) had been registered within the same month as the study start date. For the remaining four studies, we were not able to find information on the study start and registration dates for the following reasons: three of the RCTs were only registered in the European Union Clinical Trials Register (EU-CTR), which does not provide the date of study registration, and one RCT was registered in a Belgian register, which is publicly not accessible. All RCTs that were prospectively registered had enrolled the first patient in or after 2006. None of the RCTs which started patient recruitment in 2004 or 2005 had been registered prospectively (Table <xref ref-type="table" rid="T2">2</xref>).</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Registration of RCTs stratified by start of patient recruitment</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th rowspan="3" align="left"> </th><th colspan="4" align="center" valign="bottom"><bold>Start of patient recruitment</bold><hr/></th></tr><tr><th align="center" valign="bottom"><bold>All</bold><hr/></th><th align="center" valign="bottom"><bold>2004 to 2005</bold><hr/></th><th align="center" valign="bottom"><bold>During or after 2006</bold><hr/></th><th align="center" valign="bottom"><bold>Unclear</bold><hr/></th></tr><tr><th align="center"><bold>n = 103 (100%)</bold></th><th align="center"><bold>n = 15 (100%)</bold></th><th align="center"><bold>n = 86 (100%)</bold></th><th align="center"><bold>n = 2 (100%)</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Registered RCTs<hr/></td><td align="center" valign="bottom">85 (83%)<hr/></td><td align="center" valign="bottom">14 (93%)<hr/></td><td align="center" valign="bottom">71 (83%)<hr/></td><td rowspan="2" align="center" valign="top">0<hr/></td></tr><tr><td align="left" valign="bottom">Registration (number) mentioned in article<hr/></td><td align="center" valign="bottom">68 (80.0%)<hr/></td><td align="center" valign="bottom">11 (71.4%)<hr/></td><td align="center" valign="bottom">57 (70.4%)<hr/></td></tr><tr><td align="left" valign="bottom">Registration (number) not mentioned in article<hr/></td><td align="center" valign="bottom">17 (20.0%)<hr/></td><td align="center" valign="bottom">3 (21.4%)<hr/></td><td align="center" valign="bottom">14 (19.7%)<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Registered prospectively<hr/></td><td align="center" valign="bottom">15 (17.6%)<hr/></td><td align="center" valign="bottom">0<hr/></td><td align="center" valign="bottom">15 (21.1%)<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Registered retrospectively<hr/></td><td align="center" valign="bottom">45 (52.9%)<hr/></td><td align="center" valign="bottom">12 (85.7%)<hr/></td><td align="center" valign="bottom">33 (46.5%)<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Study start and registration within same month<hr/></td><td align="center" valign="bottom">21 (24.7%)<hr/></td><td align="center" valign="bottom">1 (7.1%)<hr/></td><td align="center" valign="bottom">20 (28.2%)<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Unclear time relation between study start and registration<hr/></td><td align="center" valign="bottom">4 (4.7%)<hr/></td><td align="center" valign="bottom">1 (7.1%)<hr/></td><td align="center" valign="bottom">3 (4.2%)<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left">Not registered</td><td align="center">18 (17%)</td><td align="center">1 (7%)</td><td align="center">15 (17%)</td><td align="center">2 (100%)</td></tr></tbody></table><table-wrap-foot><p>RCT, randomized controlled trial.</p></table-wrap-foot></table-wrap><p>Seventeen (81%) of the 21 excluded RCTs with patient recruitment before 2004 had been registered. All of them had been retrospectively registered after enrollment of the first patient.</p><p>Sixty-eight (80%) of the 85 registered trials specified the registration identifier: 25 in the main text only, 21 in the abstract and main text, ten in the abstract only, two in the title and main text, one in the acknowledgments section, and nine below or above the list of author affiliations (two of them additionally specified the identifier in the main text and abstract, respectively). The 17 (20%) registered trials without specification of the registration number were not classified as 'registered’ unless they were found in the ICTRP searching with words extracted from the publication. One RCT which specified the registration number in the abstract and main text reported the wrong number twice. The reported number actually belonged to another RCT of the same first author, and we discovered the correct NCT identifier by searching the ICTRP. Moreover, we found four registered RCTs which were categorized as a prospective cohort study (n = 3) or prospective case–control study (n = 1).</p><p>Forty-four of the 68 (65%) RCTs specifying the registration number also mentioned trial registration explicitly in a full sentence in the abstract and/or main text.</p><p>Eighteen of the 103 included RCTs (17%) did neither report trial registration nor could they be found in the ICTRP. Upon examination of the unregistered RCTs, we noted the following differences to the registered RCTs: all these RCTs were national and all had been undertaken in a single-center setting, except for three RCTs conducted at two to three centers. In addition, it seemed that the median sample size was smaller (66 versus 126; Table <xref ref-type="table" rid="T3">3</xref>).</p><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>Study characteristics of the included RCTs</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Study characteristic</bold></th><th align="left"><bold>Registered RCTs (n = 85)</bold></th><th align="left"><bold>Unregistered RCTs (n = 18)</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">National, multinational setting<hr/></td><td align="left" valign="bottom">68 (80%), 17 (20%)<hr/></td><td align="left" valign="bottom">18 (100%), 0<hr/></td></tr><tr><td align="left" valign="bottom">Monocenter, multicenter setting<hr/></td><td align="left" valign="bottom">49 (58%), 36 (42%)<hr/></td><td align="left" valign="bottom">15 (83%), 3 (17%)<hr/></td></tr><tr><td align="left" valign="bottom">Sample size (median (range))<hr/></td><td align="left" valign="bottom">126 (12 to 66,000)<hr/></td><td align="left" valign="bottom">66 (12 to 256)<hr/></td></tr><tr><td align="left" valign="bottom">Country of main investigator<hr/></td><td align="left" valign="bottom">USA (n = 13), Germany (n = 8), Netherlands (n = 7), China (n = 6), Italy (n = 6), Norway (n = 6), UK (n = 5), Japan (n = 4), Austria (n = 3), Belgium (n = 3), Denmark (n = 3), South Korea (n = 3), Spain (n = 3), Sweden (n = 3), Switzerland (n = 3), Finland (n = 2), Egypt (n = 2), New Zealand (n = 2), and Australia, France, and India (for all n = 1)<hr/></td><td align="left" valign="bottom">South Korea (n = 5), USA (n = 5), Brazil (n = 2), UK (n = 2), and China, Finland, India, and Ukraine (for all n = 1)<hr/></td></tr><tr><td align="left">Surgical subspecialty or surgery-related discipline</td><td align="left">General surgery (n = 24), endoscopy/gastroenterology/gastrointestinal surgery (n = 14), nephrology/kidney transplantation/surgery (n = 10), bariatric surgery (n = 6), hepatology/liver transplantation/hepatobiliary surgery (n = 6), anesthesiology/surgery (n = 5), pediatrics/pediatric surgery (n = 3), psychiatry/neurology/surgery (n = 5), cardiology/heart transplantation/cardiothoracic surgery (n = 2), colorectal surgery (n = 2), and other (n = 8)</td><td align="left">General surgery (n = 9), gastroenterology/gastrointestinal endoscopy/gastrointestinal surgery (n = 3), anesthesiology (n = 2), neurosciences/neurology/psychiatry/otolaryngology (n = 2), pediatric surgery (n = 1), and reconstructive breast surgery (n = 1)</td></tr></tbody></table><table-wrap-foot><p>RCT, randomized control trial.</p></table-wrap-foot></table-wrap><sec><title>Reporting results</title><p>Sixty-one of the 85 registered RCTs were registered on ClinicalTrials.gov. One of these RCTs was still ongoing and published as a study protocol only, another one had been withdrawn before enrollment of the first patient. Both were excluded from further analysis, and the remaining 59 RCTs were included. For 7 (12%) of them, results had been posted in the results database (Table <xref ref-type="table" rid="T4">4</xref>). In our sample, the proportion of RCTs with summary data posted on ClinicalTrials.gov was smaller among the retrospectively registered trials in comparison to RCTs with prospective registration (3/45 (7%) versus 3/15 (20%); relative risk 0.33; 95% confidence interval 0.08, 1.48; <italic>P</italic> = 0.15).</p><table-wrap position="float" id="T4"><label>Table 4</label><caption><p>Results availability for registered RCTs stratified by completion date</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th rowspan="3" align="left"> </th><th colspan="5" align="center" valign="bottom"><bold>Study completion date</bold><hr/></th></tr><tr><th align="center" valign="bottom"><bold>All</bold><hr/></th><th align="center" valign="bottom"><bold>Before or during February 2011</bold><hr/></th><th align="center" valign="bottom"><bold>After February 2011 to February 2012</bold><hr/></th><th align="center" valign="bottom"><bold>After February 2012</bold><hr/></th><th align="center" valign="bottom"><bold>Unclear</bold><hr/></th></tr><tr><th align="center"><bold>n = 59 (100%)</bold></th><th align="center"><bold>n = 24 (100%)</bold></th><th align="center"><bold>n = 25 (100%)</bold></th><th align="center"><bold>n = 5 (100%)</bold></th><th align="center"><bold>n = 5 (100%)</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom"><bold>Results posted</bold><hr/></td><td align="center" valign="bottom">7 (12%)<hr/></td><td align="center" valign="bottom">5 (21%)<hr/></td><td align="center" valign="bottom">2 (8%)<hr/></td><td align="center" valign="bottom">0<hr/></td><td align="center" valign="bottom">0<hr/></td></tr><tr><td align="left" valign="bottom">Results posted only<hr/></td><td align="center" valign="bottom">2 (3%)<hr/></td><td align="center" valign="bottom">1 (4%)<hr/></td><td align="center" valign="bottom">1 (4%)<hr/></td><td align="center" valign="bottom">0<hr/></td><td align="center" valign="bottom">0<hr/></td></tr><tr><td align="left" valign="bottom">Results posted and link to publication provided by investigator<hr/></td><td align="center" valign="bottom">1 (2%)<hr/></td><td align="center" valign="bottom">0<hr/></td><td align="center" valign="bottom">1 (4%)<hr/></td><td align="center" valign="bottom">0<hr/></td><td align="center" valign="bottom">0<hr/></td></tr><tr><td align="left" valign="bottom">Results posted and automatic linkage via register identification<hr/></td><td align="center" valign="bottom">4 (7%)<hr/></td><td align="center" valign="bottom">4 (17%)<hr/></td><td align="center" valign="bottom">0<hr/></td><td align="center" valign="bottom">0<hr/></td><td align="center" valign="bottom">0<hr/></td></tr><tr><td align="left" valign="bottom"><bold>No results posted</bold><hr/></td><td align="center" valign="bottom">52 (88%)<hr/></td><td align="center" valign="bottom">19 (79%)<hr/></td><td align="center" valign="bottom">23 (92%)<hr/></td><td align="center" valign="bottom">5 (100%)<hr/></td><td align="center" valign="bottom">5 (100%)<hr/></td></tr><tr><td align="left" valign="bottom">Link to publication provided by investigator only<hr/></td><td align="center" valign="bottom">1 (2%)<hr/></td><td align="center" valign="bottom">0<hr/></td><td align="center" valign="bottom">1 (4%)<hr/></td><td align="center" valign="bottom">0<hr/></td><td align="center" valign="bottom">0<hr/></td></tr><tr><td align="left" valign="bottom">Link to publication provided by investigator and automatic linkage via register identification<hr/></td><td align="center" valign="bottom">1 (2%)<hr/></td><td align="center" valign="bottom">0<hr/></td><td align="center" valign="bottom">1 (4%)<hr/></td><td align="center" valign="bottom">0<hr/></td><td align="center" valign="bottom">0<hr/></td></tr><tr><td align="left" valign="bottom">Automatic linkage via register identification only<hr/></td><td align="center" valign="bottom">27 (46%)<hr/></td><td align="center" valign="bottom">9 (38%)<hr/></td><td align="center" valign="bottom">15 (60%)<hr/></td><td align="center" valign="bottom">1 (20%)<hr/></td><td align="center" valign="bottom">2 (40%)<hr/></td></tr><tr><td align="left">No results posted and no link provided</td><td align="center">23 (39%)</td><td align="center">10 (42%)</td><td align="center">6 (24%)</td><td align="center">4 (80%)</td><td align="center">3 (60%)</td></tr></tbody></table><table-wrap-foot><p>List of registries included in the ICTRP search portal (September 2013): Australian New Zealand Clinical Trials Registry (ANZCTR); ClinicalTrials.gov; European Union Clinical Trials Register (EU-CTR); International Standard Randomised Controlled Trial Number Register (ISRCTN); Brazilian Clinical Trials Registry (ReBec); Chinese Clinical Trial Registry (ChiCTR); Clinical Trials Registry - India (CTRI); Clinical Research Information Service (CRiS), Republic of Korea; Cuban Public Registry of Clinical Trials (RPCEC); German Clinical Trials Register (DRKS); Iranian Registry of Clinical Trials (IRCT); Japan Primary Registries Network (JPRN); Pan African Clinical Trial Registry (PACTR); Sri Lanka Clinical Trials Registry (SLCTR); and The Netherlands National Trial Register (NTR). ICTRP, International Clinical Trials Register Platform; RCT, randomized controlled trial.</p></table-wrap-foot></table-wrap><p>As mentioned in the Methods section, we did not explicitly assess whether the RCTs registered in ClinicalTrials.gov were ACTs according to the FDAAA 801. However, there are several reasons to assume that most of the included RCTs were not ACTs. First, not all of them had at least one site in the USA and were initiated or ongoing as of 27 September 2007, or later. Second, several included RCTs compared surgical procedures instead of drug or device interventions. Thus, we presume that most of the analyzed trials were not required to report summary data to ClinicalTrials.gov according to the FDAAA 801.</p><p>Twenty-five RCTs had been registered in trial registers other than ClinicalTrials.gov: International Standard Randomised Controlled Trial Number Register (ISRCTN; n = 7), EU-CTR (n = 4), The Netherlands National Trial Register (NTR; n = 3), Australian New Zealand Clinical Trials Registry (ANZCTR; n = 2), German Clinical Trials Register (DRKS; n = 2), Japan Primary Registries Network University Hospital Medical Information Network (JPRN-UMIN; n = 4), Belgian register (n = 1), Chinese Clinical Trial Registry (ChiCTR; n = 1), and Clinical Trials Registry - India (CTRI; n = 1). Of these registers, DRKS, ISRCTN, CTRI, ANZCTR, NTR, and JPRN-UMIN provide a data field for a link to publications on the study record page. The study record page of only 5 (26%) of the 19 RCTs primarily registered in these registers included a link to PubMed citations or a list of publications. EU-CTR and ChiCTR did not provide a data field for links to publications, and the Belgian register is not publicly accessible.</p></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>Eighty-five of the 103 analyzed RCTs (83%) were registered and 80% (68/85) of the registered trials specified the registration identifier, for example the NCT number, in the study report. In addition, 65% (44/68) of RCTs specifying the registration number also mentioned trial registration explicitly in a full sentence in the abstract and/or main text. Though the ICMJE trial registration policy requires trial registration before the enrollment of the first patient, only 18% of the registered trials had been registered prospectively. The majority were registered retrospectively (53%) or within the same month as the study start date (25%). This implies that with regard to the vast majority of registered trials, it cannot be excluded that initial details of study design, objective, eligibility criteria, or primary and secondary outcomes were changed after study start. For this specific reason, retrospective registration is only suboptimal. On the other hand, retrospective registration is helpful for the identification of trials, especially those still ongoing or not yet published.</p><p>The results of only 7 (12%) of the 59 RCTs registered on ClinicalTrials.gov had been submitted to the ClinicalTrials.gov results database. As mentioned before, we did not assess whether these 59 RCTs were ACTs. Thus, it remains unclear whether the legal requirements to submit aggregate summary data to ClinicalTrials.gov really pertain to the included RCTs. Nonetheless, investigators of all trials registered in ClinicalTrials.gov can voluntarily submit summary data to the results database. Though the legal requirement to report results applies only to certain interventional trials, sponsors and investigators should be encouraged to use the results database for timely dissemination of their research findings publicly [<xref ref-type="bibr" rid="B11">11</xref>].</p><p>For 28 (47%) of the trials registered on ClinicalTrials.gov there was at least a link provided to publications, which are automatically mapped to these studies by the ClinicalTrials.gov identifier (NCT number). In 29 (48%) cases there was no reporting of results or link to publications available at all.</p><p>Nguyen <italic>et al</italic>. recently published data on the public availability of trial results assessing cancer drugs in the USA. They analyzed 646 trials (including 209 RCTs) regarding results posting at ClinicalTrials.gov and/or publication of results in journals. One year after the completion of the trials, the results of only 9% of all trials (12% of the RCTs) were available at ClinicalTrials.gov [<xref ref-type="bibr" rid="B12">12</xref>]. These data are similar to our own results. Moreover, Nguyen <italic>et al</italic>. reports that, despite the FDAAA, results of almost half of the trials assessing cancer drugs were not publicly available (neither at ClinicalTrials.gov nor in journals) three years after completion of the trials [<xref ref-type="bibr" rid="B12">12</xref>].</p><p>Jones <italic>et al</italic>. recently conducted a cross-sectional analysis of 585 trials with at least 500 participants, which were prospectively registered with ClinicalTrials.gov to estimate the frequency with which results of large RCTs registered with ClinicalTrials.gov are not publicly available. Almost one third (n = 171) of the included 585 RCTs remained unpublished. Of the 171 unpublished RCTs, almost 80% (n = 133) had no results available in the ClinicalTrials.gov results database [<xref ref-type="bibr" rid="B13">13</xref>].</p><p>Comparing our findings regarding trial registration to those from disciplines not related to surgery, it seems that the awareness of the need for trial registration has grown in the surgical research community. The proportion of registered RCTs in this study by far exceeds the reported numbers from other recent trials. Milette <italic>et al</italic>. investigated the transparency of outcome reporting and trial registration of RCTs published in top psychosomatic and behavioral health journals between January 2008 and September 2009 [<xref ref-type="bibr" rid="B14">14</xref>]. Of the 63 articles reviewed, only 13 (20.6%) had been registered. A similar proportion of registered trials were reported by McGee <italic>et al</italic>. who conducted a cohort study of all RCTs in kidney transplantation published between October 2005 and December 2010 and determined trial registration and declaration of registration by authors [<xref ref-type="bibr" rid="B15">15</xref>]. Of the 307 included trials, only 74 (24%) had been registered; 44 (59%) of the registered trials declared trial registration details at least within one study report. Moreover, the authors investigated factors associated with trial registration. Trial registration was more likely if the trial was published more than once, in later years, or if it was reported in journals following the ICMJE guidelines. Furthermore, trials conducted in the USA were significantly more likely to be registered than European trials. Trial registration was also less likely for trials not declaring their funding source. Regarding the factors associated with declaration of registration details, McGee <italic>et al</italic>. found that registered trials were more likely to declare registration details in related reports if they were published in a journal complying with the ICMJE guidelines or in later years (2007 to 2010). Compared to European trials, trials conducted globally were less likely to declare registration details. Interestingly, USA trials were no more or even less likely to declare registration details than trials conducted in Europe.</p><p>Califf <italic>et al</italic>. recently examined the characteristics of clinical trials (in three different medical specialties: cardiovascular, mental health, oncology) registered in ClinicalTrials.gov. Their analysis showed that the proportion of prospectively registered trials increased over time (from 33% in October 2004 to September 2007 to 48% in October 2007 to September 2010) [<xref ref-type="bibr" rid="B16">16</xref>]. This is concordant with our results (Table <xref ref-type="table" rid="T2">2</xref>).</p><p>Reveiz <italic>et al</italic>. investigated another important aspect of trial registration: its potential influence on reporting quality [<xref ref-type="bibr" rid="B17">17</xref>]. The authors conducted a cross-sectional study of 148 RCTs from the highest ranked journals (JCR 2006) and analyzed this sample with regard to adherence to key methodological items of the Consolidated Standards of Reporting Trials (CONSORT) statement and several other secondary outcomes, <italic>inter alia</italic> trial registration. Of these, 36% of the included RCTs reported trial registration. Reporting quality was significantly better if trial registration was declared in the trial report.</p><p>Several studies have examined whether journals publishing original articles in specialties such as urology and pediatrics endorse recommendations aimed at the improvement of publication practice. Meerpohl <italic>et al</italic>. analyzed the online author instructions of 69 journals indexed in the subject category 'pediatrics’ of JCR 2007 with regard to endorsement of the Uniform Requirements for Manuscripts (URM) of the ICMJE, of five major reporting guidelines, disclosure of conflicts, and trial registration [<xref ref-type="bibr" rid="B18">18</xref>]. Only 16 (23%) of the included 69 journals either recommended or required trial registration. This means that more than three quarters of pediatric journals did not require/recommend trial registration in August 2008. One year later, Meerpohl <italic>et al</italic>. analyzed 41 pediatric open access journals with regard to good publication practice [<xref ref-type="bibr" rid="B19">19</xref>]. The authors came to the conclusion that pediatric open access journals mention certain recommendations and guidelines, for example the URM, more frequently than conventional journals, but that the endorsement was still only moderate. Trial registration, for example, was only recommended/required by approximately a third (32%) of the included journals.</p><p>Kunath <italic>et al</italic>. conducted a cross-sectional study of RCTs published in 2009 in urology-related journals indexed in JCR 2009 [<xref ref-type="bibr" rid="B20">20</xref>]. Of the 106 included RCTs, 63 (59.4%) were registered. The proportion of reports of registered trials was significantly higher in journals requiring trial registration as a requirement for publication than in journals not mentioning trial registration in their author instructions (71.4% versus 51.6%).</p><p>It is, however, not the journal editors’ main responsibility to ensure good publication practice, especially complete trial registration. Primarily, trialists are in charge of registering their trials. Reveiz <italic>et al</italic>. surveyed the corresponding authors of a random sample of 500 clinical trials published between May 2005 and May 2006 [<xref ref-type="bibr" rid="B21">21</xref>]. Of the 275 trialists who completed the questionnaire, 64% supported the registration of all 20 items of the WHO minimum data set that should be recorded for clinical trial registration, while 6% did not support any of them. Only 21% of the respondents had registered all of their trials since 2005. However, 47% declared the intention to provide all 20 items of the WHO data set to a publicly accessible register for future clinical trials. Comparing the respondents who received mixed or only industry funding with those receiving only non-industry funding, the latter were significantly more likely to intend to provide all 20 WHO data set items for future trials.</p><p>Looking into the future of trial registration and reporting, their successful implementation as integral parts of clinical research highly depends on the continuous efforts and initiatives taken by trialists, journal editors, ethic boards, and funders.</p><p>There are some limitations to this study. We studied a cohort of RCTs published between June 2012 and December 2012. Due to the moderate sample size, the generalizability of the results might be limited. RCTs could have been missed, because the PubMed search was performed only 6 weeks after the evaluated time period. Some citations might not yet have been fully indexed with MeSH terms in MEDLINE. However, the Cochrane RCT filter does not only use MeSH terms to identify RCTs, but also text words within the database’s title/abstract field, which have been validated for identifying RCTs. Thus, the chance to have missed publications reporting an RCT not yet indexed with MeSH terms is relatively low. It is also possible that we erroneously declared a trial as unregistered if it was registered within a register not included in the ICTRP search platform and the registration was not mentioned in the publication. In addition, since the analyzed cohort of RCTs was taken from the ten journals with the highest impact factors (according to JCR 2011) which explicitly required trial registration in their instructions to authors, our results might overestimate the compliancy with trial registration and therefore might not be transferable to the entirety of surgery-related journals. This likely implies a limited external validity of our results.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>Although still suboptimal, the situation is improving over time and trial registration is gaining momentum. However, complete prospective trial registration has not yet been achieved even in top surgery journals, which explicitly require trial registration in their author instructions. Furthermore, the results reporting process, for example the submission of study results to the ClinicalTrials.gov results database, is still not widely practiced. Researchers, peer reviewers, and journal editors should therefore continue to collaborate to improve trial reporting and registration.</p></sec><sec><title>Abbreviations</title><p>ACT: Applicable clinical trial; ANZCTR: Australian New Zealand Clinical Trials Registry; ChiCTR: Chinese Clinical Trial Registry; CONSORT: Consolidated Standards of Reporting Trials; CTRI: Clinical Trials Registry - India; DRKS: German Clinical Trials Register; EU-CTR: European Union Clinical Trials Register; FDAAA: Food and Drug Administration Amendments Act; FDAMA: Food and Drug Administration Modernization Act; ICMJE: International Committee of Medical Journal Editors; ICTRP: International Clinical Trials Register Platform; ISRCTN: International Standard Randomised Controlled Trial Number; JCR: Journal Citation Reports; JPRN: Japan Primary Registries Network; NTR: The Netherlands National Trial Register; RCT: Randomized controlled trial; SJEG: Surgery Journal Editors Group; UMIN: University Hospital Medical Information Network; URM: Uniform Requirements for Manuscripts; WHO: World Health Organization.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>JH conceived and designed the study, undertook data extraction, analysis, and interpretation, and drafted and revised the manuscript. MIM designed and executed the search strategies, undertook search documentation, data extraction, analysis, and interpretation, and revised the manuscript. JM conceived and designed the study, undertook data analysis and interpretation, and drafted and revised the manuscript. All authors read and approved the final manuscript.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional file 1</title><p>PubMed search strategy for identifying RCTs in 10 surgery journals.</p></caption><media xlink:href="1745-6215-14-407-S1.docx"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S2"><caption><title>Additional file 2</title><p>Extracted data of 128 references evaluated as full texts.</p></caption><media xlink:href="1745-6215-14-407-S2.xls"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S3"><caption><title>Additional file 3</title><p>Extracted data of 85 registered RCTs.</p></caption><media xlink:href="1745-6215-14-407-S3.xls"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
A prospective, randomized, placebo-controlled, double-blind, multicenter study of the effects of irbesartan on aortic dilatation in Marfan syndrome (AIMS trial): study protocol	<sec><title>Background</title><p>Cardiovascular complications are the leading cause of mortality and morbidity in Marfan syndrome (MFS), a dominantly inherited disorder caused by mutations in the gene that encodes fibrillin-1. There are approximately 18,000 patients in the UK with MFS. Current treatment includes careful follow-up, beta blockers, and prophylactic surgical intervention; however, there is no known treatment which effectively prevents the rate of aortic dilatation in MFS. Preclinical, neonatal, and pediatric studies have indicated that angiotensin receptor blockers (ARBs) may reduce the rate of aortic dilatation. This trial will investigate the effects of irbesartan on aortic dilatation in Marfan syndrome.</p></sec><sec><title>Methods/Design</title><p>The Aortic Irbesartan Marfan Study (AIMS) is an investigator-led, prospective, randomized, placebo-controlled, double-blind, phase III, multicenter trial. Currently, 26 centers in the UK will recruit 490 clinically confirmed MFS patients (aged ≥6 to ≤40 years) using the revised Ghent diagnostic criteria. Patients will be randomized to irbesartan or placebo. Aortic root dilatation will be measured by transthoracic echocardiography at baseline and annually thereafter. The primary outcome is the absolute change in aortic root diameter per year measured by echocardiography. The follow-up period will be a minimum of 36 months with an expected mean follow-up period of 48 months.</p></sec><sec><title>Discussion</title><p>This is the first clinical trial to evaluate the ARB irbesartan versus placebo in reducing the rate of aortic root dilatation in MFS. Not only will this provide useful information on the safety and efficacy of ARBs in MFS, it will also provide a rationale basis for potentially lifesaving therapy for MFS patients.</p></sec><sec><title>Trial registration</title><p>ISRCTN, <ext-link ext-link-type="uri" xlink:href="http://www.controlled-trials.com/ISRCTN90011794">90011794</ext-link></p></sec>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Mullen</surname><given-names>Michael J</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>michael.mullen@uclh.nhs.uk</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Flather</surname><given-names>Marcus D</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>M.Flather@uea.ac.uk</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Jin</surname><given-names>Xu Yu</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>xuyu.jin@msd.ox.ac.uk</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Newman</surname><given-names>William G</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>William.Newman@cmft.nhs.uk</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Erdem</surname><given-names>Guliz</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>gulizerdem@gmail.com</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Gaze</surname><given-names>David</given-names></name><xref ref-type="aff" rid="I6">6</xref><email>David.Gaze@stgeorges.nhs.uk</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Valencia</surname><given-names>Oswaldo</given-names></name><xref ref-type="aff" rid="I6">6</xref><email>Oswaldo.Valencia@stgeorges.nhs.uk</email></contrib><contrib contrib-type="author" id="A8"><name><surname>Banya</surname><given-names>Winston</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>w.banya@imperial.ac.uk</email></contrib><contrib contrib-type="author" id="A9"><name><surname>Foley</surname><given-names>Claire E</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>c.foley@rbht.nhs.uk</email></contrib><contrib contrib-type="author" id="A10"><name><surname>Child</surname><given-names>Anne</given-names></name><xref ref-type="aff" rid="I6">6</xref><email>achild@sgul.ac.uk</email></contrib>	Trials	<sec><title>Background</title><p>Cardiovascular complications are the leading cause of mortality and morbidity in Marfan syndrome (MFS), a dominantly inherited disorder caused by mutations in the gene that encodes fibrillin-1. There are approximately 18,000 patients in the UK with MFS. MFS is diagnosed clinically using the Ghent criteria which emphasizes the identification of a positive family history, ectopia lentis, aortic root dilatation Z-score >2, and a systemic score of clinical features [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>]. Twenty-five percent of cases are the result of a new mutation in the fibrillin-1 gene, and are often more seriously affected than familial cases [<xref ref-type="bibr" rid="B3">3</xref>]. Gene mutations in <italic>FBN1</italic> have been demonstrated in 92% of classically affected MFS type 1 cases [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B5">5</xref>]. Other genes capable of causing familial ascending thoracic aortic aneurysms are now being described, but these families can usually be differentiated clinically from MFS [<xref ref-type="bibr" rid="B6">6</xref>-<xref ref-type="bibr" rid="B9">9</xref>].</p><p>Aneurysmal dilatation of the aortic root is the most serious cardiovascular manifestation of MFS. This results from weakening of the tissues within the aortic wall and consequent reduced ability to contain the forces associated with cardiac ejection. The natural history of aortic root aneurysms is expansion over many years followed by dissection and rupture and premature death. In addition, myxomatous valve changes with insufficiency of the mitral and aortic valves, and progressive myocardial dysfunction can also occur and require intervention, and a wide range of non-cardiac manifestations affecting skeletal and ocular systems result in significant morbidity and mortality. The average age at death of an untreated MFS patient is 32 years [<xref ref-type="bibr" rid="B10">10</xref>].</p><p>Current treatment includes careful follow-up and prophylactic surgical intervention to replace the aneurysmal root when the risk of spontaneous dissection or rupture exceeds that of surgery. For most patients this risk is judged by the size of the aortic root, with the standard threshold at which patients are usually considered for prophylactic surgical treatment being 50 mm [<xref ref-type="bibr" rid="B11">11</xref>]. In some patients with an adverse family history, or where pregnancy is considered or where rapid dilatation is observed, surgical intervention may be considered at diameters below 50 mm.</p><sec><title>Medical treatment</title><p>The goal of medical therapy in MFS is to slow or arrest the development of clinical manifestations of MFS. With respect to the cardiovascular system, the current gold standard for medical treatment is with oral beta blockers. Beta blocker therapy has been shown in retrospective and prospective studies to reduce the rate of aortic root dilatation [<xref ref-type="bibr" rid="B12">12</xref>-<xref ref-type="bibr" rid="B15">15</xref>], and is associated with an increase in life span [<xref ref-type="bibr" rid="B16">16</xref>]. The mechanism is unknown but is likely to be mediated through reduction in left ventricular ejection force, blood pressure, and pulse pressure, all of which potentially reduce aortic wall stress. However, recent studies in children with MFS and a meta-analysis have cast doubt on the efficacy of beta blocker therapy [<xref ref-type="bibr" rid="B17">17</xref>]. Furthermore, many patients cannot tolerate beta blockers (approximately 25% to 50% of MFS patients), either because they have asthma, which affects about 20% of MFS children, or because of intolerable side effects including dizziness, nightmares, and lethargy, or can only tolerate them in small doses. Furthermore, beta blocker therapy does not alter the underlying process that results in weakness and dilatation of the aortic wall.</p></sec><sec><title>Fibrillin-1</title><p>Fibrillin-1 is the major component of extracellular myofibrils which form the backbone of the elastic tissues in the extracellular matrix. Original hypotheses of the pathogenesis of MFS were based on a simple model of aortic dilatation occurring as a mechanical consequence of abnormal elastic tissues. However, such a hypothesis does not explain many manifestations of MFS including excessive growth and abnormal alveolar septation.</p><p>Elucidating the mechanisms of aortic dilatation has been facilitated by the development of an MFS knockout mouse. The so-called mgR mouse has an identical mutation of fibrillin-1 as that seen in human MFS and the mutant allele produces structurally normal fibrillin-1 protein at 15% of the normal level. The mouse manifests all the clinical features of human MFS including the mouse equivalent of postnatally-acquired aortic disease and death by aortic dissection, as well as lung and skeletal findings [<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B19">19</xref>]. Homozygous mgR mice die between 3 and 6 months of age of dissecting aortic aneurysm. The mice show loss of structural integrity of the aortic wall with cystic medial necrosis, histologically identical to that seen in human MFS. Breach of the elastic laminae in fibrillin-1 mutant mice is believed to allow infiltration of inflammatory cells into the media, resulting in intense elastolysis associated with increased expression of matrix metalloproteinases (MMPs) [<xref ref-type="bibr" rid="B20">20</xref>].</p><p>Research in this mouse model has elucidated a much more complex role for fibrillin-1 in the regulation of extracellular activity of transforming growth factor beta (TGF-β). Abnormal fibrillin-1 leads to excess activity of TGF-β in extracellular tissues and this appears to contribute to the pathogenesis of many of the phenotypic features of MFS [<xref ref-type="bibr" rid="B21">21</xref>]. Myxomatous changes of the mitral valve in mutant mice were correlated with excess TGF-β signaling, and prevented by TGF-β antagonism <italic>in vivo</italic>[<xref ref-type="bibr" rid="B22">22</xref>]. Furthermore, increased TGF-β signaling in association with increased MMP expression was also observed in the dura and aortic wall of fibrillin-1-deficient mice [<xref ref-type="bibr" rid="B23">23</xref>]. These mice were shown to have excess immunoreactive free TGF-β, and systemic administration of a TGF-β neutralizing antibody rescued lung morphogenesis in fibrillin-1-deficient mice, and attenuated changes in the aortic wall.</p></sec><sec><title>The renin-angiotensin system and TGF-β regulation</title><p>Extracellular TGF-β is also regulated by the autocrine molecule angiotensin II. Activation of the angiotensin II receptor type 1 (AT1) can increase the production of TGF-β, which may be responsible for many of the cellular events in the tissue of patients with MFS including proliferation of vascular smooth muscle cells and levels of MMPs. By contrast, activation of the angiotensin II receptor type 2 (AT2) has beneficial effects on aortic wall homeostasis. Selective inhibition of the AT1 receptor therefore offers a therapeutic target to favorably modify the pathogenesis of tissue injury in MFS. AT1 receptor blockers (ARBs) include a number of commonly used antihypertensive medications including losartan and irbesartan. In the experimental mouse, ARB administration resulted in a clinically relevant decrease in TGF-β signaling, reduced plasma levels of free TGF-β, reduced tissue expression of TGF-β-responsive genes, and reduced levels of intracellular mediators within the TGF-β signaling cascade, such as phosphorylated Smad2.</p><p>Habashi <italic>et al</italic>. [<xref ref-type="bibr" rid="B24">24</xref>] reported that five young Marfan mice were given 0.6 g/L of losartan, consumed through their drinking water for a period of 6 to 10 months. Another group of ten MFS mice were given a placebo, and a third group of seven MFS mice were given a dose of 0.5 g/L of propranolol, a beta blocker. A fourth group of eleven wild-type mice without MFS served as a control group. The mice studied were 2 months old when therapy was started, equivalent to human teenage years. These mice already had enlarged aortas. After the mice were treated for 6 months, examination of the aorta histologically showed losartan, as opposed to placebo or propranolol, prevented elastic fiber fragmentation, and blunted TGF-β signaling in the aortic media. Additionally, echocardiographic measurements of aortic root growth in losartan MFS mice were comparable to the normal control group of mice (<italic>P</italic> = 0.55), and the absolute aortic root diameter between losartan MFS mice and the normal control group at the end of treatment was also similar (<italic>P</italic> = 0.32). Losartan MFS mice also showed significant improvement in aortic wall thickness and aortic wall architecture compared to placebo, and normalization relative to the normal control group. In comparison, propranolol-treated MFS mice showed a slower rate of aortic growth compared to the placebo group (<italic>P</italic> <0.001), but showed no effect on aortic wall thickness or aortic wall architecture compared to the placebo group, thus limiting its effect in slowing the rate of aortic growth.</p><p>Therefore it is particularly attractive to consider the use of an ARB, which both lowers blood pressure comparably with beta blocker therapy [<xref ref-type="bibr" rid="B25">25</xref>-<xref ref-type="bibr" rid="B28">28</xref>] (a known positive effect in MFS) and leads to a clinically relevant decrease of TGF-β signaling [<xref ref-type="bibr" rid="B29">29</xref>,<xref ref-type="bibr" rid="B30">30</xref>]. These data support the hypothesis that many features of MFS are probably due to failure of proper regulation of TGF-β function [<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B32">32</xref>].</p></sec><sec><title>ARBs in human MFS</title><p>Pilot data of 18 severe neonatal MFS cases [<xref ref-type="bibr" rid="B33">33</xref>,<xref ref-type="bibr" rid="B34">34</xref>] treated with losartan [<xref ref-type="bibr" rid="B31">31</xref>] indicated that placing an affected infant on an ARB was associated with a reduction in the rate of aortic root dilatation. Preliminary data indicated the average change in aortic root diameter pre-losartan treatment was 3.5 mm/yr and following losartan treatment was 0.5 mm/yr. This compares with an average change in aortic root diameter of 1.67 mm/yr in patients treated with beta blockers alone. This pilot data suggests that a clinical trial of an ARB may demonstrate effects more specific than, and probably additive to, beta blockers [<xref ref-type="bibr" rid="B35">35</xref>-<xref ref-type="bibr" rid="B37">37</xref>].</p><p>In the Aortic Irbesartan Marfan Study (AIMS), we wish to translate the pre-clinical, neonatal, and pediatric work into a randomized clinical trial (RCT) to investigate whether the ARB irbesartan can reduce aortic dilatation in patients with MFS compared to placebo. All patients will receive standard treatments including beta blocker therapy if tolerated. We believe this will provide additional information regarding the effects of combined therapy as well as the effects in patients intolerant of beta blockers. Furthermore, we propose to examine the effects of irbesartan in a wider age range, extending the upper adult age limit to 40 years, since patients are often first diagnosed in adulthood.</p><p>A study funded by the National Institute of Health (NIH) of losartan versus beta blocker is being carried out in the USA. This study has recruited 604 patients and is also evaluating effects on aortic dilatation [<xref ref-type="bibr" rid="B38">38</xref>]. The USA study has a different design to the one described in this protocol, since the AIMS trial is comparing irbesartan versus placebo, but the two trials will provide complementary information on this important question.</p><p>It is widely accepted that aortic root dilatation is the hallmark of serious cardiovascular complications in MFS. Furthermore, aortic root dilatation is usually the major factor considered in referring patients for surgery. Therefore, while being a surrogate outcome measure, clinical outcome and the decision to intervene are directly related to aortic dilatation. We have considered conducting a clinical outcome study in MFS (evaluating the effects of irbesartan on rates of death, aortic surgery, or other serious cardiovascular complications) but this study would probably take 10 years or more to complete in order to assess the true long-term effects of treatment. For this reason we have selected the rate of aortic root dilatation as the primary outcome measure. We believe that the main benefit of ARB treatment will be in the prevention of aortic complications when applied as a prophylactic measure in patients with MFS who are either in their growing years or in adulthood prior to the development of severe dilatation. We consider that this is too long to wait to introduce a potentially lifesaving treatment for young patients at risk of severe complications. It is important to be aware of the level of evidence needed to introduce a new treatment for MFS, but the carefully considered opinion of the MFS experts collaborating in this study is that a significant reduction in aortic dilatation would provide important clinical benefits to patients, and that this measure is a robust surrogate for clinical outcomes. Previous studies that showed the benefits of beta blockers in MFS used the same outcome of aortic dilatation as the primary outcome measure. We also know that there is increasing use of ARBs among patients with MFS even though there is no clear evidence of efficacy or indeed safety for this indication, and thus there is an urgent need to complete studies of the efficacy of ARBs in MFS.</p></sec><sec><title>Feasibility</title><p>MFS patients require lifelong monitoring and many need major surgery. As a healthcare burden comparison there are about 8,000 patients with Cystic Fibrosis in the UK who generally require a greater level of ongoing healthcare support than MFS patients, but there are important similarities in that both conditions are inherited, affect young people, and reduce life expectancy.</p><p>MFS patients currently suffer great fear about their long-term prognosis and if effective, irbesartan would provide a lifesaving treatment option which could extend life span into the normal range. There is tremendous support from the medical community and lay MFS population to perform this trial.</p></sec><sec><title>Aim of study</title><p>To investigate whether the angiotensin II receptor antagonist irbesartan reduces the rate of aortic dilatation in MFS compared to placebo.</p></sec></sec><sec><title>Methods/Design</title><p>This trial is an investigator-led, prospective, randomized, placebo-controlled, double-blind, phase III, multicenter study. Patients will be randomized to two groups:</p><p>1. Irbesartan group</p><p>2. Placebo group.</p><sec><title>Patient population</title><sec><title>Inclusion criteria</title><p>1. Clinically confirmed MFS using the revised Ghent diagnostic criteria (2010)</p><p>2. Provision of informed consent</p><p>3. From ≥6 to ≤40 years of age inclusive.</p></sec><sec><title>Exclusion criteria</title><p>1. Previous cardiac or aortic surgery</p><p>2. Planned cardiac or aortic surgery at the time of randomization</p><p>3. Aortic root Z-score ≤0</p><p>4. Aortic diameter ≥4.5 cm</p><p>5. Hemodynamically significant, severe valvular disease (at the judgement of the treating clinician)</p><p>6. Heart failure, defined as left ventricular ejection fraction <40%</p><p>7. Therapeutic use of angiotensin-converting-enzyme (ACE) inhibitors/angiotensin II receptor antagonist (patients on ACE inhibitors/angiotensin II receptor antagonists who discontinue this treatment are required to have a 3-month wash-out period prior to entry)</p><p>8. Previous recorded adverse reaction to the trial medication (irbesartan) or any ACE inhibitor/angiotensin II receptor antagonist</p><p>9. Female patients who are pregnant, planning pregnancy, or not using reliable contraception</p><p>10. Known impaired renal function defined as estimated creatinine clearance <60 mL/min, or serum creatinine >150 μmol/L.</p></sec></sec><sec><title>Concomitant treatments and procedures</title><p>Patients should generally continue all their concomitant routinely indicated treatments including beta blockers. Beta blocker use is not mandated by this protocol and should be used at the judgement of the treating physician.</p><p>Therapeutic use of ACE inhibitors or other angiotensin II receptor antagonists during the trial is not permitted. Patients are eligible for the trial if they have a 3-month wash-out period (no ACE inhibitors/angiotensin II receptor antagonists) prior to entry.</p></sec><sec><title>Screening, randomization, and unblinding</title><sec><title>Screening</title><p>Potentially eligible patients will be screened at participating centers throughout the UK. All patients diagnosed clinically as having MFS within the participating hospital clinics will be screened for eligibility. Those who are identified as potentially suitable will be approached to see if they wish to participate. Written informed consent will be requested before the patient is enrolled into the study. Family members who are affected can also be screened. Patients will be jointly supervised by the responsible cardiologist and geneticist at each site.</p></sec><sec><title>Randomization</title><p>Randomization will be carried out by an internet-based randomization service. Investigators will be required to confirm that the patient is eligible. Patients will be stratified at randomization according to age (6 to 18 years and >18 years of age), beta blocker use, and center.</p></sec><sec><title>Unblinding</title><p>Unblinding the allocation code can only be undertaken in exceptional circumstances via the electronic case record form (eCRF), when knowledge of the allocation is essential for treating the patient, for example, suspected unexpected serious adverse reaction (SUSAR) or other serious adverse event (SAE). The Clinical Trials and Evaluation Unit (CTEU) at the Royal Brompton and Harefield NHS Foundation Trust, London, UK, will be contacted before breaking the code. In all cases, the date and reason for breaking the code will be documented. Unblinding will occur at the individual patient-level only.</p></sec></sec><sec><title>Trial intervention</title><p>Study treatment will be in three phases:</p><p>1. Run-in phase: 75 mg once daily (OD) open-label irbesartan for 4 weeks before randomization</p><p>2. Treatment phase: 150 mg (OD) active/placebo for 4 weeks before uptitration in appropriate patients</p><p>3. Treatment phase: 300 mg (OD) active/placebo maximum tolerated dose for remaining treatment period.</p><p>The proposed target doses are as follows: 300 mg OD for patients >50 kg and 150 mg OD for patients ≤50 kg. Patients will be monitored at regular intervals in the baseline phase for tolerability to study medication including general clinical examination, blood pressure, electrolytes, and renal function. Compliance and tolerability will also be monitored by the study teams at 3-month intervals by telephone, and should there be any issues the patient will return to clinic for review. Indications for stopping the study drug would include any apparent serious side effects, hypotension not amenable to a reduction in study drug, pregnancy, or significant impairment of renal function.</p></sec><sec><title>Study visits</title><sec><title>Visit 0, run-in phase (month 1)</title><p>Eligible patients screened from outpatient clinics will be invited to consent to the study. If the patient consents, they will undergo a clinical examination including blood pressure check and electrocardiography (ECG). Patients will be dispensed with a 1-month supply of 75 mg open-label irbesartan. This is to establish tolerance and compliance to irbesartan prior to the baseline study visit. Patients will also undergo the baseline echocardiogram and have study-specific bloods taken for renal function, mutation analysis (if not already taken as part of routine clinical care), and TGF-β sub-study.</p></sec><sec><title>Visit 1, baseline (month 2)</title><p>Eligible patients who tolerate the open-label run-in phase will have a clinical examination and baseline characteristics recorded (height, weight, blood pressure, heart rate). Patients will also undergo a compliance to medication check, clinical evaluation including ECG, medications review, liver function tests, full blood count, urea and electrolytes, and renal function. Baseline medication will remain unchanged (beta blocker, other antihypertensive, or nil).</p><p>Patients will then be randomized into the trial using the interactive voice recognition system (IVRS). A unique identification number will be allocated to the patient, which will match the number on the study drug held at the site pharmacy. Once allocated, the study drug will be dispensed by the pharmacy. Patients will be provided with the 150 mg dose of irbesartan/placebo for 1 month. Children ≤50 kg will continue with the 150 mg dose and will not be uptitrated.</p></sec><sec><title>Visit 2, uptitration or maintenance visit (month 3)</title><p>Patients >50 kg will be uptitrated to the 300 mg dose of irbesartan/placebo at month 3 if tolerated. Patients will undergo a compliance to medications check, clinical evaluation including blood pressure check, medications review, liver function tests, full blood count, urea and electrolytes, and renal function tests.</p><p>Patients who remain on the 150 mg dose will undertake the visit procedures described above, although they will not be dispensed the 300 mg dose.</p></sec><sec><title>Visit , titration visit (if necessary)</title><p>Patients who do not tolerate the maximum 300 mg dose for whatever reason will be downtitrated to the 150 mg dose and continue in the trial. At this visit, the patients will have a clinical evaluation including blood pressure, medications review, liver function tests, full blood count, urea and electrolytes, and renal function tests, before the 150 mg dose is dispensed.</p></sec></sec><sec><title>Telephone checks (3-month intervals)</title><p>Subsequent to visit 2, there will be 3-month interval telephone calls between the research team and patient to check compliance to the medication and tolerability up to month 60 (5 years). Should the patient have any problems they will return to the clinic for further review by the research team.</p></sec><sec><title>Annual visits (month 12, 24, 36, 48, and 60)</title><p>Patients will also have an annual follow-up at month 12, 24, 36, 48, and 60 (depending on entry to the trial) as per routine clinical care to undergo a compliance to medication check, clinical evaluation including blood pressure, ECG, medications review, liver function tests, full blood count, urea and electrolytes, renal function, and drug dispensing. Patients will also undergo an annual echocardiogram for analysis. At year 1, an annual study-specific blood sample will be taken for analysis for the TGF-β sub-study.</p><p>Study procedures and follow-up are described in Table <xref ref-type="table" rid="T1">1</xref>.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Study-related investigations and follow-up</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="left" valign="bottom"><bold>Procedure</bold><hr/></th><th align="center" valign="bottom"><bold>Visit 0</bold><hr/></th><th align="center" valign="bottom"><bold>Visit 1</bold><hr/></th><th align="center" valign="bottom"><bold>Visit 2</bold><hr/></th><th align="center" valign="bottom"><bold>Visit </bold><hr/></th><th align="center" valign="bottom"><bold>Telephone call</bold><hr/></th><th align="center" valign="bottom"><bold>Visit 3</bold><hr/></th><th align="center" valign="bottom"><bold>Telephone call</bold><hr/></th><th align="center" valign="bottom"><bold>Visit 4</bold><hr/></th><th align="center" valign="bottom"><bold>Telephone call</bold><hr/></th><th align="center" valign="bottom"><bold>Visit 5</bold><hr/></th><th align="center" valign="bottom"><bold>Telephone call</bold><hr/></th><th align="center" valign="bottom"><bold>Visit 6</bold><hr/></th><th align="center" valign="bottom"><bold>Telephone call</bold><hr/></th><th align="center" valign="bottom"><bold>Visit 7</bold><hr/></th></tr><tr><th align="left"> </th><th align="center"><bold>Month 1, run-in phase, 75 mg open-label</bold></th><th align="center"><bold>Month 2, baseline, 150 mg active/placebo for 1 month</bold></th><th align="center"><bold>Month 3, uptitration to 300 mg active/placebo or maintenance at 150 mg active/placebo</bold></th><th align="center"><bold>Month 4, titration visit (if necessary)</bold></th><th align="center"><bold>Month 6, 9</bold></th><th align="center"><bold>Year 1 (month 12)</bold></th><th align="center"><bold>Month 15, 18, 21</bold></th><th align="center"><bold>Year 2 (month 24)</bold></th><th align="center"><bold>Month 27, 30, 33</bold></th><th align="center"><bold>Year 3, (month 36)</bold></th><th align="center"><bold>Month 39, 42, 45</bold></th><th align="center"><bold>Year 4 (month 48)</bold></th><th align="center"><bold>Month 51, 54, 57</bold></th><th align="center"><bold>Year 5 (month 60)</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Screening<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Compliance check<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td></tr><tr><td align="left" valign="bottom">Clinical evaluation<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td></tr><tr><td align="left" valign="bottom">Informed consent<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Randomization<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Blood pressure<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td></tr><tr><td align="left" valign="bottom">Echocardiogram<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td></tr><tr><td align="left" valign="bottom">ECG<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td></tr><tr><td align="left" valign="bottom">Medications<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td></tr><tr><td align="left" valign="bottom">Liver function<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td></tr><tr><td align="left" valign="bottom">Full blood count<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td></tr><tr><td align="left" valign="bottom">Urea and electrolytes<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td></tr><tr><td align="left" valign="bottom">Renal function<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td></tr><tr><td align="left" valign="bottom">Study drug given<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Mutation analysis (if necessary)<hr/></td><td align="center" valign="bottom">√<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left">Blood sample (TGF-β and other biomarkers)</td><td align="center">√</td><td align="center"> </td><td align="center"> </td><td align="center"> </td><td align="center"> </td><td align="center">√</td><td align="center"> </td><td align="center"> </td><td align="center"> </td><td align="center"> </td><td align="center"> </td><td align="center"> </td><td align="center"> </td><td align="center"> </td></tr></tbody></table><table-wrap-foot><p>Visit * only required if patient does not tolerate 300 mg dose. Patient will be downtitrated to 150 mg OD. ECG, electrocardiography; OD, once daily; TGF-β, transforming growth factor beta.</p></table-wrap-foot></table-wrap></sec><sec><title>Participant follow-up</title><p>Patients will also be followed up via the NHS Information Centre Medical Research Information Service for a minimum of 20 years after the end of study follow-up.</p></sec><sec><title>Pregnancy</title><p>Women of childbearing potential (able to have children) are required to use reliable forms of contraception if taking part in the study (including barrier and/or oral methods). Women of childbearing potential will be excluded if planning pregnancy during the trial period or sexually active and not using a reliable form of contraception.</p></sec><sec><title>Patient withdrawal</title><sec><title>Temporary discontinuation of investigational medicinal product (IMP)</title><p>There should be no reason to discontinue the study drug unless in exceptional circumstances such as side effects, for example renal impairment or significant hypotension. In such cases, the investigator should downtitrate or discontinue the study medication as required. Reintroduction of the study medication should be monitored and uptitration undertaken as required.</p></sec><sec><title>Permanent discontinuation of IMP</title><p>Patients may permanently withdraw from treatment with the investigational medicinal product (IMP) if they decide to do so, at any time and for any reason, or this may be the investigator’s decision. If there is considered to be a concern about the batch of IMP, recall procedures will be in place and measures will be taken to ensure the safety of all trial participants.</p></sec></sec><sec><title>Withdrawal from trial procedures and incomplete follow-up</title><p>Patients are free to withdraw consent from trial procedures at any time. Investigators must ascertain the reasons for the withdrawal including discontinuation of study drug, withdrawal from study investigations and/or follow-up, withdrawal due to adverse events, failure to attend, non-compliance, withdrawal of consent, or other reasons. The withdrawal form must be faxed to the CTEU within 5 working days, unless withdrawal is due to a SAE, in which case the investigator will follow SAE reporting procedures.</p><p>Withdrawal from trial procedures may result in incomplete patient follow-up and failure to capture outcome data. In these cases as much data as possible will be collected, up until the point of withdrawal. Patients may choose to withdraw from trial procedures and request that further data are not collected.</p></sec><sec><title>Enrolment and participating centers</title><p>Based on 26 centers recruiting, we expect an approximate recruitment rate of one patient/month/center, which will enable 490 patients to be recruited over a 2-year enrolment window. The study will run for 66 months (5.5 years) split into four periods as follows:</p><p>1. Start-up: 6 months</p><p>2. Enrolment: 24 months (with potential to extend into the third year if necessary)</p><p>3. Follow-up: 36 months minimum (patients enrolled at the start of the study will be followed up for a maximum of 60 months, thus the mean follow-up period is 48 months)</p><p>4. Closeout: 6 months.</p></sec><sec><title>Outcome measures</title><sec><title>Primary outcome measure</title><p>The primary outcome measure will be the absolute change in aortic root diameter per year measured by echocardiography. Echo measurements will be taken by transthoracic echocardiograms (TTEs) performed at baseline and annually thereafter in order to assess the annual change of aortic dilatation.</p></sec><sec><title>Secondary outcome measures</title><p>The secondary outcome measures are as follows:</p><p>1. Change in Z-score per year, where the Z-score is calculated on aortic root and body surface area (BSA)</p><p>2. Clinical events and requirement for surgery including aortic dissection confirmed by transesophageal echocardiography (TEE), MRI, or CT, aortic dissection requiring emergency surgery, aortic dissection requiring elective surgery, aortic dilatation requiring elective or emergency surgery, death (all causes and classified by suspected cause), cerebrovascular accident, cardiovascular death, aortic regurgitation requiring surgery, or death during surgery for any of the above</p><p>3. Left ventricular function determined by volumes and ejection fraction</p><p>4. Left ventricular mass measurements</p><p>5. Assessment of valvular function</p><p>6. Cardiac rhythm and voltage</p><p>7. Height, weight, arm span, and lower segment measurements</p><p>8. Fibrillin-1 mutation analysis will be performed for those patients whose mutation status is unknown.</p></sec></sec><sec><title>Echocardiography</title><p>Aortic dilatation will be measured by echocardiography. All annual echocardiograms will be analyzed by an established echo core laboratory. Patients will be seen in centers that have routine access to standard M-mode, two-dimensional (2-D) echocardiography. Transthoracic M-mode, 2-D, and Doppler echocardiograms will be performed by experienced technicians, according to a standardized protocol. Each center will be trained on the standardized protocol prior to commencing the trial. Aortic root diameter will be measured at the annulus, in the sinuses of Valsalva at the tip of the open cusps at 90° to the direction of flow, the sinotubular junction, ascending aorta, aortic arch, and descending aorta.</p><p>Each echocardiogram will be sent electronically to the echocardiographic core laboratory at the John Radcliffe Hospital, Oxford, UK, where a single experienced investigator will supervise the reading and interpretation of all echocardiograms according to a standardized protocol to reduce variability due to observer variation. The echo core laboratory will be blinded to the study drug allocation to eliminate any bias in echo measurement. Quality control processes will be developed as part of the echo protocol. The echocardiographic protocol is included in Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>.</p></sec><sec><title>Fibrillin-1 mutation analysis at baseline</title><p>There is a 92% chance of finding a mutation in classical MFS. For patients who have not already had fibrillin-1 mutation screening, but who are considered clinically affected with MFS, a 5 mL EDTA blood sample will be collected and stored for assay. Samples will be analyzed and funded as per usual local arrangements.</p><p>Samples from units which do not have funding for these tests should be sent directly to the Sonalee Laboratory, St. George’s, University of London, London, UK [<xref ref-type="bibr" rid="B3">3</xref>]. Samples will be entered in the research program assay, and research laboratory reports will be issued.</p><p>A separate genetic sub-study will be undertaken to investigate the correlation between the response to medication and the site and type of mutation which determine the phenotype [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B39">39</xref>,<xref ref-type="bibr" rid="B40">40</xref>]. Patients will consent to this sub-study separately from the main trial.</p></sec><sec><title>Additional investigations and sub-studies</title><p>To increase knowledge about the underlying disease and the potential mechanism of action of ARBs, several sub-studies are proposed. The main proposed sub-studies are:</p><p>1. Genetic sub-studies: a) fibrillin-1 mutation analysis; and b) pharmacogenetics sub-study</p><p>2. Total circulating TGF-β1.</p></sec><sec><title>Data collection</title><sec><title>Electronic case record form (eCRF)</title><p>Trial data will be captured on a web-based eCRF. The eCRF will be designed in accordance with the requirements of the trial protocol and will comply with regulatory requirements. Access to the eCRF will be password-protected.</p></sec><sec><title>Pharmacovigilance</title><p>The Royal Brompton and Harefield NHS Foundation Trust (Sponsor) has delegated responsibility for pharmacovigilance to the trial coordinating center, the CTEU of the Royal Brompton and Harefield NHS Foundation Trust. The CTEU will be responsible for recording all reported SAEs from investigational trial sites, and expedited reporting of SUSARs in accordance with statutory regulations.</p></sec><sec><title>Adverse event (AE)</title><p>An adverse event (AE) is defined as any untoward medical occurrence in a patient or clinical trial subject administered a medicinal product and which does not necessarily have a causal relationship with this treatment. An AE can therefore be any unfavorable and unintended sign (including an abnormal laboratory finding), symptom, or disease temporally associated with the use of an IMP, whether or not considered related to the IMP.</p></sec><sec><title>Adverse reaction (AR)</title><p>Adverse reactions (ARs) are all untoward and unintended responses to an IMP related to any dose administered. All AEs judged by either the reporting investigator or the sponsor as having reasonable causal relationship to a medicinal product qualify as ARs.</p><p>In the event an AR is reported during the trial, investigators will assess the severity of the AE using the following criteria, detailed on the AE report form in the eCRF:</p><p>1. Mild: awareness of signs or symptoms, but easily tolerated; are of minor irritant type; causing no loss of time from normal activities; symptoms would not require medication</p><p>2. Moderate: discomfort severe enough to cause interference with usual activities</p><p>3. Severe: inability to do work or usual activities; signs and symptoms may be of systemic nature or require medical evaluation and/or treatment.</p></sec></sec><sec><title>Unexpected adverse reaction (UAR)</title><p>An unexpected adverse reaction (UAR) is an AR, the nature or severity of which is not consistent with the applicable product information (summary of product characteristics, SmPC). When the outcome of the AR is not consistent with the applicable product information this AR should be considered as unexpected. Side effects documented in the SmPC which occur in a more severe form than anticipated are also considered to be unexpected.</p></sec><sec><title>Expected drug-related AR</title><p>Expected drug-related ARs will be referred to the SmPC as provided and the summary in Additional file. Symptomatic hypotension defined as systolic blood pressure <90 mmHg in combination with dizziness and/or syncope on standing is considered an expected AR and will be reported on a specific AE form. Any other AR will be reported on a generic AE form on the eCRF.</p></sec><sec><title>Reporting ARs</title><p>Investigators will report all ARs on the AE report form in the eCRF including information of the event, details of date of onset, frequency, severity, and potential relationship to treatment, outcomes, and action taken. Investigators will make a clinical judgement as to the appropriate action required depending on the severity of the reaction. This could include monitoring the patient over a period of time, interrupting the drug regime, discontinuing the patient from the trial, or continuing with the trial as specified. Investigators will submit AE reports to the CTEU after each patient visit. The CTEU will maintain a database of all ARs. AEs will be reviewed at regular intervals by the Data Monitoring Committee (DMC) for signal and trend analysis.</p></sec><sec><title>Serious adverse events (SAEs)/reactions</title><p>SAEs or reactions are defined as any untoward medical occurrence or effect that at any dose results in death, is life threatening, requires hospitalization or prolongation of existing inpatient hospitalization, results in persistent or significant disability or incapacity, or is a congenital anomaly or birth defect.</p><p>Should a study participant become pregnant while undertaking the trial, or aid in the conception of a child while they are participating in the trial, the pregnancy and resulting child should be followed up for a period of no less than 18 months. In this trial should a child be followed up and diagnosed with MFS, this would not be considered unexpected due to the nature of the syndrome.</p></sec><sec><title>Expected SAEs (as a result of the underlying disease)</title><p>Expected SAEs as a result of the underlying disease include: admission or procedure for MFS including treatment for cardiovascular, musculoskeletal, ocular, and thoracic complications; aortic dissection; aortic regurgitation requiring surgery; emergency or elective aortic root and/or valve replacement surgery; cerebrovascular accident; and cardiovascular death, sudden death, or death during surgery.</p><p>Expected SAEs will be reported as per the usual data capture requirements for the study and are not subject to expedited reporting.</p><p>Other SAEs which are not expected irrespective of causality will be subject to SAE reporting requirements. In the event of an SAE, investigators will report details on the SAE form on the eCRF including date of event, admissions, diagnosis details, date of discharge, or death. SAE reports must be completed within 24 hours of the investigator’s knowledge of the SAE. Investigators will be able to submit follow-up SAE reports should further information become available. Investigators will be expected to assess and assign causality and expectedness of each event on the form using the definitions described below. The CTEU will review all SAE reports. The Chief Investigator/deputy will review the SAE reports and inform the CTEU of the assessment.</p></sec><sec><title>Definitions for assessment of causality</title><p>The definitions for assessment of causality include:</p><p>1. Unrelated: there is no evidence of any causal relationship</p><p>2. Unlikely: there is little evidence to suggest there is a causal relationship (for example the event did not occur within a reasonable time after administration of the trial medication). There is another reasonable explanation for the event (for example the patient’s clinical condition, other concomitant treatment)</p><p>3. Possible: there is some evidence to suggest a causal relationship (for example because the event occurs within a reasonable time after administration of the trial medication). However, the influence of other factors may have contributed to the event (for example the patient’s clinical condition, other concomitant treatments)</p><p>4. Probable: there is evidence to suggest a causal relationship and the influence of other factors is unlikely</p><p>5. Definitely: there is clear evidence to suggest a causal relationship and other possible contributing factors can be ruled out</p><p>6. Not assessable: there is insufficient or incomplete evidence to make a clinical judgement of the causal relationship.</p></sec><sec><title>Suspected unexpected serious adverse reactions (SUSARs)</title><p>A serious adverse reaction (SAR) can be considered unexpected when the AR is not consistent with the applicable product information or expected SAEs listed above. All SUSARs related to an IMP, which occur during the trial, are subject to expedited reporting. Where applicable, if an event is considered a SUSAR, the patient should be unblinded from the study allocation.</p></sec><sec><title>Reporting of SUSARs</title><p>A full and detailed account of the SAE must be recorded on the SAE report. The SAE report must be completed within 24 hours. A medical summary should also be faxed to the CTEU within 24 hours. The Chief Investigator/deputy will review the report and summary and inform the CTEU of the assessment.</p></sec><sec><title>Expedited reporting of SUSARs</title><p>All SUSARs will be reported to the Medicines and Healthcare Products Regulatory Agency (MHRA) and NHS Research Ethics Committee (REC) by the CTEU. All SUSAR reports will be unblinded prior to submission. A SUSAR which is fatal or life threatening will be reported to the MHRA and the main REC by the CTEU as soon as possible and within 7 days of knowledge of the event. A SUSAR which is not fatal or life threatening must be reported to the MHRA and the main REC as soon as possible and within 15 days of knowledge of the event. The CTEU will inform all relevant parties of any reported SUSARs within 15 working days.</p></sec><sec><title>Annual reporting</title><p>The CTEU will submit annual safety reports of all suspected SARs in accordance with regulatory requirements to the MHRA and the main REC. Annual safety reports will be submitted to the MHRA on the date of the original clinical trials authorization. Annual progress reports will also be submitted to the main REC. There is no requirement for local trial sites to submit progress reports to local RECs.</p></sec><sec><title>Statistical considerations</title><sec><title>Sample size calculation</title><p>The primary outcome is absolute change in aortic root diameter per year. The sample size calculation was based on estimates obtained from a database of MFS patients maintained by AC, co-investigator and lead Geneticist. Information was extracted from this database for all patients who met the following criteria: age at first echo between 6 and 40 years; at least two serial echo measurements; and time between the first and last valid measurement between 0.9 and 5.1 years.</p><p>This provided a database of 254 patients who had a median (interquartile range) follow-up time of 3.4 (2.4, 4.5) years. The data were cleaned and reviewed and during this time the average rate of aortic dilatation was approximately 1 mm per year, with a standard deviation of 1.8 mm. Table <xref ref-type="table" rid="T2">2</xref> shows the number of patients required to test for a difference between an annual dilatation rate of 1 mm on placebo against hypothesized rates on irbesartan. This is based on achieving 80% power and testing at the 5% significance level.</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Estimated annual dilatation rates over a mean follow-up of 3.4 years</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="center"><bold>Number of patients in placebo group</bold></th><th align="center"><bold>Number of patients in irbesartan group</bold></th><th align="center"><bold>Annual growth on placebo (mm)</bold></th><th align="center"><bold>Annual growth on irbesartan (mm)</bold></th><th align="center"><bold>SD of annual growth (mm)</bold></th></tr></thead><tbody valign="top"><tr><td align="center" valign="bottom">142<hr/></td><td align="center" valign="bottom">142<hr/></td><td align="center" valign="bottom">1<hr/></td><td align="center" valign="bottom">0.4<hr/></td><td align="center" valign="bottom">1.8<hr/></td></tr><tr><td align="center" valign="bottom">169<hr/></td><td align="center" valign="bottom">169<hr/></td><td align="center" valign="bottom">1<hr/></td><td align="center" valign="bottom">0.45<hr/></td><td align="center" valign="bottom">1.8<hr/></td></tr><tr><td align="center" valign="bottom">204<hr/></td><td align="center" valign="bottom">204<hr/></td><td align="center" valign="bottom">1<hr/></td><td align="center" valign="bottom">0.5<hr/></td><td align="center" valign="bottom">1.8<hr/></td></tr><tr><td align="center" valign="bottom">252<hr/></td><td align="center" valign="bottom">252<hr/></td><td align="center" valign="bottom">1<hr/></td><td align="center" valign="bottom">0.55<hr/></td><td align="center" valign="bottom">1.8<hr/></td></tr><tr><td align="center">318</td><td align="center">318</td><td align="center">1</td><td align="center">0.6</td><td align="center">1.8</td></tr></tbody></table></table-wrap><p>By inspection of this table, to detect a 0.5 mm change in dilatation rate would require 204 patients in each group. Allowing for a 20% drop-out (including missing data and non-compliance) we aim to recruit 245 patients in each group, making a total of 490 patients. Expansion rates according to treatment group would be regularly reviewed by the independent Data and Safety Monitoring Committee (DSMC) who would advise the Steering Committee if there was clear evidence of benefit before the scheduled end of the study, or alternatively may advise if there is no real possibility of finding a difference (performing a 'futility’ analysis) either because irbesartan does not have the expected effects, or if aortic expansion rates are slower than expected in the control group. We have selected a proportional reduction of 50% of dilatation rate per annum (0.5 mm) largely on pragmatic grounds. If irbesartan only offered a very modest reduction in rate we are much less certain that this could be translated into a clinical benefit. We are also measuring a surrogate outcome (aortic dilatation) but one which is closely related to adverse clinical events in this population.</p></sec></sec><sec><title>Statistical analysis</title><p>The primary analysis in this study is the comparison of the rate of annual aortic dilatation in those patients treated with irbesartan compared to placebo. The annual rate of dilatation will be calculated by estimates of mean values of the annual echocardiograms adjusting the time of follow-up to a 'common’ start point, that is, the baseline measurement. An independent samples <italic>t</italic>-test will be used to test for a difference in the rate between the irbesartan and placebo groups, assuming that the measurement of annual dilation follows a normal distribution. Analysis of covariance (ANCOVA) will be used if the analysis of the primary outcome is adjusted for any other variables. The primary analysis will be carried out according to the intention-to-treat principle in which all randomized patients will be included according to their initial randomized allocation irrespective of whether they continue to take the assigned treatment or not. Patients who are not followed up to the end of the study for whatever reason will have the last echo measurement included in the analysis. Patients will be followed from the time of randomization to the end of the study period. Thus individual patients will have a variable follow-up period. No formal method for imputing missing aortic diameter values is proposed as the comparison will be performed on the mean of group data. Secondary analyses include comparison of other measurements using appropriate comparative and descriptive statistics.</p></sec><sec><title>Regulatory and ethical considerations</title><sec><title>Regulatory framework and approval</title><p>This study is a randomized trial of an IMP (licensed product in new conditions of use; new dosing schemes/new target population), and as such will comply with the European Clinical Trials Directive and the Medicines for Human Use (Clinical Trials) Regulations 2004. A clinical trial authorisation (CTA) has been granted from the MHRA. The study is registered in the European Clinical Trials Database with a European Union Drug Regulating Authorities Clinical Trials (EudraCT) number.</p></sec><sec><title>Ethical approval</title><p>The trial will comply with the Declaration of Helsinki (<ext-link ext-link-type="uri" xlink:href="http://www.wma.net/">http://www.wma.net/</ext-link>) on research involving human subjects. The study protocol, patient information sheet, and consent form have been approved by the NHS National Research Ethics Service (NRES) and subsequently the research and development departments of each participating center for site-specific approval. The AIMS trial is also approved on the National Institute for Health Research (NIHR) portfolio.</p></sec></sec><sec><title>Monitoring</title><sec><title>Initiation visit</title><p>Before the study commences, each trial site will receive a training visit from the CTEU where required. The purpose of these visits will be to ensure that the local research team (local principal investigator, co-investigators, study coordinator, and pharmacists) fully understand the protocol, eCRF, and the practical procedures for the study.</p></sec></sec><sec><title>Interim monitoring visits</title><p>At regular intervals during the study, the CTEU will perform monitoring visits to each trial site. The purpose of these visits is to ensure compliance with the protocol and that ethical and regulatory requirements are met. Source data verification (SDV) and checking of essential documents will be performed. Monitors will also visit the pharmacy departments to review study procedures, storage, and accountability of the IMP.</p><p>Monitoring visits also provide an opportunity for further training if required (for example new staff). Central review of study data will also be performed throughout the study by the data management team at the CTEU.</p><sec><title>Closeout visit</title><p>At the end of the study, each trial site will receive a closeout visit from the CTEU to resolve any outstanding edit queries or AEs and to verify the archiving procedures for study documentation.</p></sec></sec><sec><title>Trial organization and committees</title><sec><title>Study management</title><p>The study will be sponsored by the Royal Brompton and Harefield NHS Foundation Trust. The Chief Investigator will be MM at the Royal Brompton Hospital, London, UK. AC at St George’s Hospital will be the lead Geneticist and XYJ at the John Radcliffe Hospital will be the lead Echocardiologist for the study. All participating clinicians have extensive experience of running MFS diagnostic clinics. A Trial Steering Committee (TSC), DMC, and Trial Management Group will be convened to oversee the trial. Central coordination of this clinical trial will be provided by the CTEU.</p></sec></sec><sec><title>Trial Steering Committee (TSC)</title><p>The main role of the TSC is to monitor and supervise the progress of the trial. The composition of the TSC will comply with Medical Research Council (MRC) guidelines with an independent Chair and lay representation as well as the Chief Investigator and main co-investigators. The TSC will meet regularly throughout the study.</p></sec><sec><title>Data Monitoring Committee (DMC)</title><p>All members of the DMC are independent of the trial. The DMC will meet prior to the start of the trial and then one and two thirds of the way through the trial, or as required thereafter. The DMC will be expected to develop, in agreement with the investigators, a charter outlining their responsibilities and operational details.</p></sec><sec><title>Study coordination</title><p>The study will be coordinated and managed by the CTEU, a dedicated clinical trials department within the Royal Brompton Hospital. In addition to providing overall project coordination, the CTEU will assist in preparing the final protocol, the investigators’ manuals, design the eCRF, provide the randomization service and design, and instigate the data management system. The CTEU will ensure that the trial runs according to the pre-agreed timetable, recruitment targets are met, eCRFs are completed accurately, compliance with relevant ethical and regulatory standards, and that all aspects of the study are performed to the highest quality. The CTEU will also assist in the training of investigators and coordinators at the start-up of the study and for performing monitoring and pharmacovigilance procedures throughout.</p></sec><sec><title>Investigators’ responsibilities</title><p>Investigators are required to ensure compliance to the protocol and all statutory regulations and guidelines, eCRFs, and manual of operations. Investigators are required to allow access to study documentation or source data on request for monitoring visits and audits performed by the CTEU, sponsor, or any regulatory authorities.</p></sec><sec><title>End of trial</title><p>The end of trial will be declared when the last patient recruited completes the last follow-up visit, that is, echocardiogram at 36 months follow-up visit.</p></sec><sec><title>Investigational medicinal products</title><sec><title>Manufacture</title><p>The study drug (irbesartan and placebo) will be purchased from the commercial supplier, Sanofi-Aventis, Guildford, Surrey, UK, which holds the manufacturing license to produce the IMP.</p><p>Brecon Pharmaceuticals Ltd, Hay-on-Wye, UK, will undertake to prepare, pack, and label the IMPs and distribute as required throughout the trial. The IMP supply will be labeled in accordance with regulatory requirements and specifications and will be approved by the MHRA as part of the application for CTA.</p></sec></sec><sec><title>Storage and dispensing</title><p>The study drug patient kit for the 1-month run-in phase, baseline phase, and uptitration phase (month 1 to 3) will be stored in a secure area of the pharmacy, under the conditions described in the respective SmPC.</p><p>The IMP supply will be dispensed by the local pharmacy which will be responsible for maintaining a record of accountability.</p><p>Continuation packs of the study drug will be dispensed direct to patient homes on a 3-month supply basis by Brecon Pharmaceuticals Ltd. Drug accountability records will be supplied as required.</p></sec><sec><title>Publication policy and dissemination of results</title><p>The results from the trial will be submitted for publication in a major journal irrespective of the outcome. The TSC will be responsible for approval of all manuscripts arising from the study prior to submission for publication. Sub-studies of center-specific data may only be carried out with the knowledge and approval of the TSC. Sub-study publications must not be published prior to the publication of the main study.</p><p>All publications and presentations will make appropriate acknowledgement of the contribution of the collaborative group. At the end of the study, patients will be able to request a copy of the results of the study from the investigator at that site.</p></sec></sec><sec><title>Trial status</title><p>The first patient was enrolled in March 2012, and recruitment is ongoing.</p></sec><sec><title>Abbreviations</title><p>2-D: Two-dimensional; ACE: Angiotensin-converting-enzyme; AE: Adverse event; AIMS: Aortic Irbesartan Marfan Study; ANCOVA: Analysis of covariance; AR: Adverse reaction; ARB: Angiotensin receptor blocker (also known as angiotensin II receptor antagonist); AT1: Angiotensin II receptor type 1; AT2: Angiotensin II receptor type 2; BMS: Bristol-Myers Squibb; BSA: Body surface area; CT: Computed tomography; CTA: Clinical trial authorisation; CTEU: Clinical Trials and Evaluation Unit; DMC: Data Monitoring Committee; DSMC: Data and Safety Monitoring Committee; ECG: Electrocardiography; eCRF: Electronic case record form; EDTA: Ethylenediaminetetraacetic acid; EudraCT: European Union Drug Regulating Authorities Clinical Trials; IMP: Investigational medicinal product; IVRS: Interactive voice recognition system; MFS: Marfan syndrome; MHRA: Medicines and Healthcare Products Regulatory Agency; MMP: Matrix metalloproteinase; MRC: Medical Research Council; MRI: Magnetic resonance imaging; NHS: National Health Service; NIH: National Institute of Health; NIHR: National Institute for Health Research; NRES: National Research Ethics Service; OD: Once daily; RCT: Randomized clinical trial; REC: Research Ethics Committee; SAE: Serious adverse event; SAR: Serious adverse reaction; SDV: Source data verification; SmPC: Summary of product characteristics; SUSAR: Suspected unexpected serious adverse reaction; TGF-β: Transforming growth factor beta; TEE: Transesophageal echocardiography; TSC: Trial Steering Committee; TTE: Transthoracic echocardiogram; UAR: Unexpected adverse reaction.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests</p></sec><sec><title>Authors’ contributions</title><p>MM (Chief Investigator) developed the study, participated in the design and co-ordination, and helped to draft the manuscript; grant holder (British Heart Foundation). AC developed the study, participated in the design and co-ordination, and helped to draft the manuscript; grant holder (British Heart Foundation). MF developed the study, participated in the design and co-ordination, and helped to draft the manuscript; grant holder (British Heart Foundation). XYJ developed the study, participated in the design and coordination, leads the echo core laboratory, and helped to draft the manuscript; grant holder (British Heart Foundation). WN participated in the design of the pharmacogenetics sub-study and helped to draft the manuscript. GE participated in the design of the study and helped to draft the manuscript. DG participated in the design of the TGF-β sub-study and sample core laboratory. OV assisted in the design of the study and sample core laboratory. WB participated in the statistical design and critical review of the manuscript. CF participated in the design and co-ordination of the study and helped to draft the manuscript. All authors read and approved the final manuscript.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional file 1</title><p>Appendices.</p></caption><media xlink:href="1745-6215-14-408-S1.doc"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
A community-based intervention for primary prevention of cardiovascular diseases in the slums of Nairobi: the SCALE UP study protocol for a prospective quasi-experimental community-based trial	<sec><title>Background</title><p>The burden of cardiovascular disease is rising in sub-Saharan Africa with hypertension being the main risk factor. However, context-specific evidence on effective interventions for primary prevention of cardiovascular diseases in resource-poor settings is limited. This study aims to evaluate the feasibility and cost-effectiveness of one such intervention—the “Sustainable model for cardiovascular health by adjusting lifestyle and treatment with economic perspective in settings of urban poverty”.</p></sec><sec><title>Methods/Design</title><p><italic>Design</italic>: A prospective quasi-experimental community-based intervention study.</p><p><italic>Setting</italic>: Two slum settlements (Korogocho and Viwandani) in Nairobi, Kenya.</p><p><italic>Study population</italic>: Adults aged 35 years and above in the two communities.</p><p><italic>Intervention</italic>: The intervention community (Korogocho) will be exposed to an intervention package for primary prevention of cardiovascular disease that comprises awareness campaigns, household screening for cardiovascular diseases risk factors, and referral and treatment of people with high cardiovascular diseases risk at a primary health clinic. The control community (Viwandani) will continue accessing the usual standard of care for primary prevention of cardiovascular diseases in Kenya.</p><p><italic>Data</italic>: Demographic and socioeconomic data; anthropometric and clinical measurements including blood pressure. Population-based data will be collected at the baseline and endline—12 months after implementing the intervention. These data will be collected from a random sample of 1,610 adults aged 35 years and above in the intervention and control sites at both baseline and endline. Additionally, operational (including cost) and clinic-based data will be collected on an ongoing basis.</p><p><italic>Main outcomes</italic>: (1) A positive difference in the change in the proportion of the intervention versus control study populations that are at moderate or high risk of cardiovascular disease; (2) a difference in the change in mean systolic blood pressure in the intervention versus control study populations; (3) the net cost of the complete intervention package per disability-adjusted life year gained.</p><p><italic>Analysis</italic>: Primary outcomes comparing pre- and post-, and operational data will be analyzed descriptively and “impact” of the intervention will be calculated using double-difference methods. We will also conduct a cost-effectiveness analysis of the intervention using World Health Organization guidelines.</p></sec><sec><title>Discussion</title><p>The outcomes of the study will be disseminated to local policy makers and health planners.</p></sec><sec><title>Trial registration</title><p>Current controlled trials <ext-link ext-link-type="uri" xlink:href="http://www.controlled-trials.com/ISRCTN84424579">ISRCTN84424579</ext-link></p></sec>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Oti</surname><given-names>Samuel O</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>soti@aphrc.org</email></contrib><contrib contrib-type="author" id="A2"><name><surname>van de Vijver</surname><given-names>Steven JM</given-names></name><xref ref-type="aff" rid="I2">2</xref><xref ref-type="aff" rid="I1">1</xref><email>svijver@aphrc.org</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Kyobutungi</surname><given-names>Catherine</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>ckyobutungi@aphrc.org</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Gomez</surname><given-names>Gabriela B</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>g.gomez@aighd.org</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Agyemang</surname><given-names>Charles</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>c.o.agyemang@amc.uva.nl</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Moll van Charante</surname><given-names>Eric P</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>e.p.mollvancharante@amc.uva.nl</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Brewster</surname><given-names>Lizzy M</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>l.m.brewster@amc.uva.nl</email></contrib><contrib contrib-type="author" id="A8"><name><surname>Hendriks</surname><given-names>Marleen E</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>m.hendriks@aighd.org</email></contrib><contrib contrib-type="author" id="A9"><name><surname>Schultsz</surname><given-names>Constance</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>c.schultsz@aighd.org</email></contrib><contrib contrib-type="author" id="A10"><name><surname>Ettarh</surname><given-names>Remare</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>rettarh@gmail.com</email></contrib><contrib contrib-type="author" id="A11"><name><surname>Ezeh</surname><given-names>Alex</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>aezeh@aphrc.org</email></contrib><contrib contrib-type="author" id="A12"><name><surname>Lange</surname><given-names>Joep</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>j.lange@aighd.org</email></contrib>	Trials	<sec><title>Background</title><p>The burden of cardiovascular diseases (CVD) is rising in sub-Saharan Africa (SSA) where up to 12.5% of deaths are attributable to CVD (SSA) [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>]. Hypertension is the leading risk factor for CVD worldwide, and it is becoming even more pronounced in SSA [<xref ref-type="bibr" rid="B3">3</xref>]. For example, the average blood pressure of people in Kenya has risen from approximately 125 mmHg in 1990 to around 130 mmHg in 2010 [<xref ref-type="bibr" rid="B4">4</xref>]. This is in contrast to countries in North America and Western Europe where the weighted average blood pressure has decreased by approximately 3 mmHg in the same period [<xref ref-type="bibr" rid="B5">5</xref>]. This is partly because countries in SSA are mostly in an earlier phase of the epidemiological transition [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref>] in which relatively low levels of behavioral CVD risk factors are increasing rapidly compared to their counterparts elsewhere in the world [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B9">9</xref>]. On the other hand, countries in SSA also score relatively poorly in terms of availability of and access to medication for treating CVD and their risk factors [<xref ref-type="bibr" rid="B10">10</xref>].</p><p>Moreover, the health and economic impact of CVD in SSA and other low-resource settings is disproportionately higher than elsewhere [<xref ref-type="bibr" rid="B11">11</xref>]. Not only do people in SSA who suffer from CVD have a higher chance of disability or death, they are also more likely to have developed CVD earlier in life—during their most economically productive years [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B13">13</xref>]. At the same time, most countries in SSA are still struggling with a high burden of infectious diseases such as malaria and HIV/AIDS. The so-called risk of a “double burden of disease” due to infectious and non-communicable diseases poses a serious threat to the weak health systems in such resource-poor settings [<xref ref-type="bibr" rid="B14">14</xref>]. Hence, there is an urgent need to implement and evaluate cost-effective interventions for primary prevention of CVD in such settings. Generally, primary prevention of CVD could involve lifestyle interventions targeting the common behavioral risk factors for CVD—tobacco use, alcohol misuse, unhealthy diet and lack of adequate physical activity [<xref ref-type="bibr" rid="B15">15</xref>]. Other primary prevention strategies target the physiological risk factors for CVD including drug therapy for the treatment and control of high blood pressure, glucose and cholesterol [<xref ref-type="bibr" rid="B15">15</xref>]. While there is strong evidence of the benefits of lifestyle modification efforts in individuals at ‘high risk’ , the evidence of such interventions when implemented at a population level (including those at ‘low risk’) is less convincing [<xref ref-type="bibr" rid="B16">16</xref>].</p><p>Modeling studies have estimated that scaling up the coverage of appropriate drug therapy will be very cost-effective in reducing the burden of CVD in low-resource settings [<xref ref-type="bibr" rid="B17">17</xref>]. Specifically, it appears that improving availability of appropriate medication for people with hypertension may play a crucial role in slowing down the rising trends of CVD mortality in SSA [<xref ref-type="bibr" rid="B18">18</xref>]. However, in order to be successful, the scaling up of antihypertensive medication and indeed other cost-effective interventions for primary prevention of CVD needs to overcome certain barriers at both the population and individual levels. First, the level of awareness about hypertension and other CVD risk factors is low, and in most countries in SSA screening opportunities at the population level are limited [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>]. At the individual level, access to quality treatment and follow-up care for hypertension and CVD in general remains poor [<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B22">22</xref>]. Primary health care facilities in SSA often lack essential medicines and technologies for diagnoses and treatment of hypertension and other CVD risk factors [<xref ref-type="bibr" rid="B23">23</xref>]. In addition, standard treatment guidelines or protocols for CVD are usually lacking and/or not implemented [<xref ref-type="bibr" rid="B24">24</xref>].</p><p>Therefore, health systems in most countries in SSA might not be prepared to deal with a hypertension or CVD epidemic even though cost-effective interventions are available. This is more so among the urban poor who are resident in vast slums across the sub-continent. There is a paucity of information on CVD and its risk factors in slums in SSA. Yet, more than 60 percent of urban populations in the sub-continent reside in slums or informal settlements [<xref ref-type="bibr" rid="B25">25</xref>]. For instance, in Nairobi (Kenya), up to 70 percent of the urban population resides in vast slums across the city [<xref ref-type="bibr" rid="B26">26</xref>]. These slums are typically underserved by social amenities including access to quality health care. Some evidence suggests that such populations fare worse than their non-slum and even rural counterparts on most health measures including high prevalence of CVD and low levels of awareness, treatment and control of hypertension [<xref ref-type="bibr" rid="B27">27</xref>].</p><p>The “Sustainable model for Cardio-vascular health by Adjusting Lifestyle and treatment with Economic perspective in settings of Urban Poverty” (acronym: SCALE UP) study is designed to implement a comprehensive intervention package of primary prevention strategies for CVD risk reduction in a slum setting in Nairobi, Kenya, and to evaluate its feasibility and cost-effectiveness. Specifically, the intervention package integrates approaches that aim to raise awareness and improve detection of hypertension at the population level. Additionally, the intervention aims to provide access to standardized quality treatment and follow-up for hypertensive patients with the overall objective of reducing their CVD risk profile in such a low-resource setting.</p></sec><sec><title>Methods/Design</title><sec><title>Design</title><p>SCALE UP is a community-based intervention aimed at reducing cardiovascular risk in people free from cardiovascular disease. It is designed to allow for a before-after comparison of cardiovascular risk between a control and an intervention setting.</p></sec><sec><title>Setting</title><p>Since 2002, the African Population and Health Research Center (APHRC) has been operating the Nairobi Urban Health Demographic Surveillance System (NUHDSS). Details about the NUHDSS have been provided elsewhere [<xref ref-type="bibr" rid="B28">28</xref>]. In brief, the Demographic Surveillance Area (DSA) covers two socio-demographically similar slums (Korogocho and Viwandani), each located about 5 to 10 km from Nairobi (Kenya). There are approximately 72,000 individuals resident in 25,000 households almost equally distributed in both slums. High levels of poverty, unemployment and lack of social amenities, including limited access to quality primary health care, characterize both slums. Specifically, there are only two public primary health facilities located on the outskirts of either slum. However, there are numerous private health providers in the slums, the majority of which are unlicensed and unregulated. Most of the private facilities operate largely for profit and rarely provide professional quality care. However, slum residents seem to prefer these services to the public ones for a number of reasons, including easier access, more approachable staff and flexible working hours (APHRC unpublished observations).</p></sec><sec><title>Intervention community</title><p>Korogocho slum, which has eight villages, each with between 3,000–5,000 residents, will be the intervention site. This slum has one centrally located private health facility with a reliable track record of providing primary health services. This facility is known as <italic>Provide International Clinic</italic>. It is well known in the slum and, although most patients still make out-of-pocket payments for services received, the clinic offers non-profit services at highly subsidized (through donor funding) costs to residents. There are no physicians or medical officers present at this clinic. Nurses and clinical officers (health personnel with a diploma in clinical medicine) provide consultations. However, the facility does not typically provide primary preventive services for CVD such as screening and treatment of hypertension. The location of this clinic and the absence of primary preventive services for CVD guided our choice of intervention site.</p></sec><sec><title>Control community</title><p>Viwandani slum will be the control site. There are seven villages in Viwandani with approximately 2,000–4,000 residents each. Unlike Korogocho, Viwandani does not have a centrally located health facility. The main health facility serving Viwandani is located on the outskirts of the slum. This facility is publicly owned and represents the usual standard of care for CVD that is available to underserved slum communities in Nairobi, thus making Viwandani the appropriate ‘control’ site in comparison with Korogocho. The clinic in Viwandani operates a weekly CVD clinic where approximately 30 patients with uncomplicated hypertension and/or diabetes from Viwandani slums are seen on each clinic day. Like the facility in Korogocho, the Viwandani clinic is also run by nurses and clinical officers.</p></sec><sec><title>Study population</title><sec><title>Inclusion criteria</title><p>Adults aged 35 years and above living in the slums of Korogocho and Viwandani who give informed consent to participate in the study. The main reason for this age cutoff is because the group above 35 years represents 21% of the total population and accounts for 71% of all known hypertensive cases, according to a CVD risk factor survey conducted in the DSA in 2008 (APHRC unpublished data). Also, due to financial constraints, the study could not be extended to an unrestricted age group. Persons with diagnosed hypertension and/or on antihypertensive therapy will be included in the study.</p></sec><sec><title>Exclusion criteria</title><p>The following will be excluded from the study: pregnant women, persons with self-reported pre-existing CVD (myocardial infarction, stroke, heart failure and angina) and all those unable to provide informed consent such as the mentally incapacitated.</p></sec></sec><sec><title>The intervention package</title><p>APHRC and the Amsterdam Institute for Global Health and Development (AIGHD) have developed an intervention package for primary prevention of CVD in urban slums based on findings from earlier studies by APHRC on CVD risk factors in this specific setting [<xref ref-type="bibr" rid="B29">29</xref>], literature review [<xref ref-type="bibr" rid="B30">30</xref>] and input from various experts and local stakeholders. The intervention package is composed of four components, which will be described below (see Figure <xref ref-type="fig" rid="F1">1</xref>).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Overview of interventions in the SCALE UP intervention package.</bold><sup><italic>1</italic></sup><italic>CHW</italic>: <italic>Community health worker</italic>.</p></caption><graphic xlink:href="1745-6215-14-409-1"/></fig><sec><title>(1) Raising awareness and (2) improving access to screening</title><p>The first two components of the intervention package include door-to-door household visits by community health workers (CHWs) to raise awareness about CVD risk and to conduct screening of each eligible adult aged 35 years and above in the intervention slum in order to determine their individual CVD risk profile. CHWs are typically community-based volunteers who have an interest in community health. They are usually well known by the community and most have received some form of basic training in community health activities such as peer health education for HIV/AIDS prevention. The SCALE UP study will recruit CHWs to represent each of the eight villages in the intervention slum. These CHWs will then be trained and equipped to perform the door-to-door visits and screening exercise in their respective villages in the intervention slum.</p><p>Two weeks prior to the commencement of the screening exercise, there will be public awareness campaigns organized by the CHWs in each village of the intervention slum. This awareness campaigns will take place at community gatherings (known locally as <italic>barazaas</italic>). At these <italic>barazaas</italic>, CHWs together with village leaders will inform the audience about the burden of CVD in the community and the need to participate in the door-to-door screening. Additionally, CHWs will visit religious gatherings, usually held at local churches and mosques, to provide the same information. Such gatherings are usually well attended by respected members of the community who will then pass the information on to others. Finally, radio jingles announcing the door-to-door screening will be aired daily on the local radio station (Korogocho FM) over a 2-week period from the start of the intervention.</p><p>The SCALE UP study investigators will train the CHWs to assess the study participants’ level of engagement in risky lifestyle behavior including tobacco use, alcohol misuse, physical activity levels and dietary habits. CHWs will also be trained in performing basic anthropometric and clinical measurements including height, weight, waist and hip circumference, blood pressure and blood glucose. They will be provided with the appropriate equipment for the clinical measurements (see Table <xref ref-type="table" rid="T1">1</xref>). Also, each CHW will be trained to provide brief counselling assistance (BCA) of healthy lifestyle modification using the six As approach –<italic>Ask</italic>, <italic>Advice</italic>, <italic>Assist</italic>, <italic>Arrange</italic>, <italic>Agree and Affirm</italic>[<xref ref-type="bibr" rid="B31">31</xref>]. Traditional BCA does not include the sixth A (Affirm) as a separate entity. However, the SCALE UP team has included this component to emphasize the need for CHWs to encourage study participants to continue with any healthy lifestyle behavior in which they reported being currently engaged. All eligible and consenting study participants will receive BCA during the door-to-door visit. If the eligible adult(s) are not at home, the CHWs will try to visit another time, with a maximum of two attempts.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>List of screening equipment</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Equipment</bold></th><th align="left"><bold>Units</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">SECA 201 circumference measurement tape<hr/></td><td align="left" valign="bottom"><italic>cm</italic><hr/></td></tr><tr><td align="left" valign="bottom">SECA 874 flat scale electronic<hr/></td><td align="left" valign="bottom"><italic>kg</italic><hr/></td></tr><tr><td align="left" valign="bottom">SECA 214 stadiometer transportable<hr/></td><td align="left" valign="bottom"><italic>cm</italic><hr/></td></tr><tr><td align="left" valign="bottom">OMRON M6 blood pressure machine<hr/></td><td align="left" valign="bottom"><italic>mmHg</italic><hr/></td></tr><tr><td align="left">ACCUCHECK glucometer and test strips</td><td align="left"><italic>mmol</italic>/<italic>l</italic></td></tr></tbody></table></table-wrap></sec><sec><title>(3) Facilitating access to quality treatment</title><p>The third component of the intervention package includes facilitating access to quality treatment for hypertension. During the door-to-door screening all persons with elevated blood pressure (≥140 mmHg systolic and/or ≥90 mmHg diastolic) will be referred to the intervention clinic—Provide International Clinic—by the CHWs. A project supervisor will independently visit each person referred by a CHW to perform a confirmatory blood pressure check before that person proceeds to the intervention clinic. Those people below 35 years who are interested in knowing about their blood pressure and/or CVD risk will be referred to a central screening point in the slum. The results of this additional screening will not be considered as part of this study.</p><p>Based on previous experience in the slums, the SCALE UP team included two incentives in the intervention package to encourage referred study participants to seek care at the intervention clinic. First, each referred participant will receive a voucher for a first free consultation at the clinic. At the same time, CHWs will receive a cash reward of approximately 3 US dollars (US$ 3.00) for each of their referred participants who attends the clinic for the first time. Hopefully, this cash reward will be a reasonable incentive for each CHW to follow up those whom they have referred and encourage them personally to attend the clinic for the first time.</p><p>In addition to improving access to treatment, it is also important to ensure that the treatment is of high quality. Thus, as part of the intervention package, the SCALE UP team will build the capacity of the intervention clinic to provide primary care services for CVD risk management. To this end, the intervention clinic will be equipped with basic and essential diagnostic equipment for CVD including a validated digital blood pressure measurement device, measurement tape, weighing scale, height measurement board, glucometer and blood glucose test strips. These devices have been selected in accordance with WHO essential technologies and tools for implementing non-communicable disease (NCD) interventions in primary health care using non-physician health workers [<xref ref-type="bibr" rid="B32">32</xref>].</p><p>Selected non-physician health workers including clinical officers and nurses from the intervention clinic will be trained on how to implement a standard protocol for management of hypertension at the primary health care level in low-resource settings. This protocol was developed by the SCALE UP study team in collaboration with senior cardiologists with experience in primary care from the University of Nairobi and the Division of Non-Communicable Diseases in the Kenya Ministry of Health (see Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>). The various treatment thresholds for hypertension given in the protocol are based on a non-laboratory-based CVD risk prediction assessment method [<xref ref-type="bibr" rid="B33">33</xref>]. Using this method, all participants with high blood pressure who enroll at the intervention clinic will be classified as low (≤10%), moderate (>10-20%) or high (>20%) risk based on their individual 10-year CVD risk (fatal and non-fatal) profile. Treatment for hypertension will range from lifestyle modification to use of locally available antihypertensive medication depending on the risk profile of the participants. The treatment threshold and target blood pressure are specified in Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>.</p><p>In addition to referring participants with high blood pressure to the intervention clinic, CHWs will also be instructed to refer other potentially moderate to high CVD risk study participants. Specifically, CHWS will be instructed to perform random blood glucose tests on all study participants who are older than 55 years of age AND have any of the following: systolic blood pressure ≥140 mmHg, diastolic blood pressure ≥90 mmHg OR abnormal waist circumference (>102 cm for men and >88 cm for women). Participants whose random blood glucose is ≥11.1 mmol/l will be referred to the intervention clinic for follow-up. At the clinic, random blood glucose will be repeated, and if still ≥11.1 mmol/l, the participant will be managed in accordance with the Kenya national clinical guidelines for management of diabetes mellitus [<xref ref-type="bibr" rid="B34">34</xref>]. As diabetes management is not a primary objective of the SCALE UP study, the intervention clinic will only be stocked with metformin—a relatively affordable oral hypoglycemic medication. Participants requiring other antidiabetic medication including insulin, based on the national guidelines, will be referred to the nearest district hospital.</p></sec><sec><title>(4) Promoting long-term adherence</title><p>The final component of the intervention package aims to promote long-term adherence among participants enrolled in the intervention clinic. For logistic reasons, all participants at the intervention clinic will be required to visit the clinic at least once every month for the entire duration of the study. Based on previous experience in the study area, longer follow-up periods tend to increase the chances of patient loss to follow-up. Beyond the regular follow-up interval, the adherence component of the intervention package will also include two subcomponents:</p><p>First, there will be an incentive-driven support group system for all participants enrolled into the intervention clinic. Each support group will have about 10–30 members drawn from participants living within the same villages. CHWs from respective villages will coordinate the activities of the support groups. This subcomponent of the intervention package seeks to leverage on the group dynamics of these support groups to promote adherence among participants. To this end, there will be a group incentive that rewards the entire group for achieving a collective level of adherence to clinic appointments at a level of 80% or more for a consecutive period of 6 months. If achieved, the entire group will receive a rebate in the monthly cost of their medication equivalent to approximately one-third of the usual cost. The estimated usual cost of medication for hypertension in the intervention clinic is Ksh 150 (US$ 1.8) per month. Additionally, the CHW will also receive a cash incentive to follow up every individual participant in the support group and encourage him/her to adhere to the clinic appointments. If participants remain adherent over the first 6 months of clinic enrollment, the CHW will receive a cash bonus of approximately US$ 1.8 per participant. Such an incentive is crucial because a previous study in the study areas found that almost 70% of hypertensive patients who drop out of primary health care clinics do so within the first 6 months (APHRC unpublished observations). The use of incentives to improve patient adherence has been tested in other settings, though the evidence is mixed [<xref ref-type="bibr" rid="B35">35</xref>].</p><p>There are other expected benefits that participants will enjoy for being part of the support group. The support group will hopefully be a forum where participants can share in the experiences of living with hypertension and learn from each other on how best to cope with this condition. Also, highly motivated participants will be selected by CHWs to become peer-educators. There will be train-the-trainer sessions where these participants will be trained by local experts on how to adopt healthy lifestyle changes such as healthy cooking classes and physical activity sessions, to mention a few.</p><p>The second subcomponent of the intervention package aimed at promoting long-term adherence is the use of the mobile phone Short Message Service (SMS). Unpublished data from the study area show that more than 80% of the adult population reports owning a mobile phone, and the remainder report having a close neighbor or other family member who owns a mobile phone through which they can be reached. Studies from other chronic conditions such as HIV have shown encouraging levels of success in the use of SMS to improve adherence [<xref ref-type="bibr" rid="B36">36</xref>]. In the SCALE UP study, an SMS will be sent every week to remind participants about their next clinic appointments, to take their medication and to provide them with healthy lifestyle tips.</p></sec></sec><sec><title>Data management</title><p>To measure the health effect of the whole intervention package at the population level, there will be two cross-sectional surveys (before and after, 12 months apart). Data will be collected on demographic and socioeconomic variables, behavioral risk factors such as tobacco and alcohol use, anthropometric measurements such as height and weight, and clinical measurements such as blood pressure and random blood glucose. Four cadres of field staff will be involved in data collection. These include field interviewers, CHWs, field assistants and supervisors.</p><p>Field interviewers will be trained by the SCALE UP study investigators to collect the demographic, socioeconomic and behavioral risk factor data using structured interviews during the cross-sectional surveys in both the intervention and control sites. In the intervention site, the anthropometric and clinical measurement will be performed by CHWs. In the control site, however, measurements will be performed by trained field assistants rather than CHWs. This is because CHWs are considered to be part of the intervention. Each interview is estimated to last approximately 30 min, followed by the physical and clinical measurements. The field staff will be instructed to follow standard procedures for all measurements as outlined in the WHO STEPS manual [<xref ref-type="bibr" rid="B37">37</xref>]. Specifically, while taking the blood pressure, field staff will be required to ensure that the respondent remains seated for about 5 min, with no talking, holding the monitoring device on the upper arm and holding it at heart level against his/her chest. The blood pressure will be measured three times, using the left arm. To minimize observer bias, validated digital equipment will be used, the OMRON M6® (Digital Automatic Blood Pressure Monitor). Note that referral of participants for further management will be based on the average of the three blood pressure measurements (systolic ≥140 and/or diastolic ≥90 mmHg).</p><p>At the intervention clinic, additional information will be collected on clinic attendees to monitor their progress and evaluate the effect of the intervention on blood pressure control and overall CVD risk profile. Baseline interviews will be conducted with each clinic attendee by trained field interviewers. Immediately after an interview, nurses at the clinic will record physical and clinical measurements for each clinic attendee. These measurements will be repeated each time the participant attends the intervention clinic over a period of at least 12 months. At the end of this period, an endline interview will be conducted with each clinic attendee. Table <xref ref-type="table" rid="T2">2</xref> summarizes the data to be collected over the intervention period.</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Data collection schedule</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Data</bold></th><th align="left"><bold>Baseline</bold></th><th align="left"><bold>Ongoing</bold></th><th align="left"><bold>Endline</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom"><bold> <italic>Population based</italic> </bold><hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Socio-demographic characteristics<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">CVD behavioral risk factors<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Physical measurements (weight, height, waist and hip circumference, blood pressure)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Blood testing (glucose)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"><bold> <italic>Clinic based</italic> </bold><hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Physical measurements (weight, height, waist and hip circumference, blood pressure)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Lifestyle modification advice<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Drug prescriptions and side effects<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Morisky adherence score<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"><bold> <italic>Operational data</italic> </bold><hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Costs<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left">Timesheets</td><td align="left"> </td><td align="left">X</td><td align="left"> </td></tr></tbody></table></table-wrap><p>All data will be entered onto electronic data collection forms pre-loaded into Mecer® Netbooks and stored in an SQL database managed by the data unit of APHRC. To ensure data quality, the field staff will be trained in interviewing techniques and how to use the Netbooks. Data collection forms will be translated into Swahili and back to English for consistency. Interviews will be conducted in Swahili—the main lingua franca in the study areas. Supervisors will be recruited to conduct spot checks of at least 5% of randomly selected interviews conducted by each field staff they supervise. They will also perform a random check of 5% of data collection forms completed by field staff. Forms will be checked for errors, missing information and inconsistent responses, and, where necessary, field staff will be required to revisit a study participant in order to clarify any erroneous information.</p><p>At the intervention clinic, qualified data entry clerks will be trained to double-capture all clinic data collected by the nurses and clinical officers during the patient consult. Clinic data will also be electronically captured using Netbooks.</p><p>Finally, cost data will be collected on an on-going basis in order to feed into the cost-effectiveness analysis of the intervention package. This will be done through an adapted checklist to collect information on costs and related time spent in all aspects of implementing the intervention package.</p><sec><title>Primary outcomes</title><p>The primary outcomes of the SCALE UP study are:</p><p>1. The difference in change in the proportion of the study populations (intervention and control slums) that are at high risk of CVD (defined as >10% risk of developing cardiovascular event in the next 5 years based on the method for assessment of cardiovascular disease risk by Gaziano et al. [<xref ref-type="bibr" rid="B33">33</xref>]).</p><p>2. The difference in change in mean systolic blood pressure in the study populations (intervention and control slums).</p><p>3. The change in mean systolic blood pressure among participants attending the local clinic (intervention slum only).</p><p>4. The net cost of the intervention package per disability-adjusted life year gained (intervention slum only).</p></sec><sec><title>Secondary outcomes</title><p>1. Prevalence of hypertension in the intervention and control slums.</p><p>2. Proportion of hypertensive respondents who are on treatment, and under control, in the intervention and control slums.</p><p>3. Proportion of high-risk participants who sought first time treatment after screening and referral.</p><p>4. Prevalence of behavioral and biological CVD risk factors: smoking, physical exercise, diet, alcohol intake, body mass index, waist circumference and waist-to-hip ratio in the intervention and control slums.</p></sec></sec><sec><title>Sample size considerations</title><p>In order to detect a 5% reduction at endline in the proportion of adults aged 35 years and above who are at moderate or high risk of CVD [<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B39">39</xref>] in the intervention population versus no change in the control population (assuming both populations have similar start prevalence at 25%), we need 2,927 respondents in both intervention and control sites, using an alpha of 0.05 and power (1-beta) of 0.90. Taking into account a non-response rate of 10%, we will approach 3,220 individuals per cross-sectional study—that is, 1,610 per site at baseline and endline surveys, respectively.</p><p>The sampling frame will be based on the most recently updated NUHDSS database. This database contains details of about 72,000 individuals including names, locations, gender, dates of birth and residential status in both slums. In the control site, we will use computer randomization (STATA® statistical software) to select the 1,610 individuals aged 35 years and older per site for each cross-sectional survey.</p><p>In the intervention site, the same computer randomization process will be followed. However, unlike the control site, the 1,610 individuals to be included in the cross-sectional survey analysis will be collected retrospectively. In other words, the intervention package will be delivered to all adults aged 35 years or older in the intervention site—that is, 6,780 individuals according to the DSS database (as at 15 June 2012).</p><p>At the clinic level, we calculated that in order to detect a 10 mmHg reduction in blood pressure (at 20 mmHg standard deviation, alpha of 0.05 and 1-β on 0.9), about 44 participants are needed. However, it is projected that approximately 1,350 participants (out of 6780) will be referred from the door-to-door visit. This number is derived from a 20% prevalence of hypertension among adults aged 35 years and older in the study area [<xref ref-type="bibr" rid="B27">27</xref>]. We estimate that roughly half of these 1,350 participants, being 675, will continue visiting the clinic for treatment. Hence, this number of people is more than sufficient for the analysis of our main primary outcome at the clinic level.</p></sec><sec><title>Analysis</title><p>Due to the fact that the slums are part of a large demographic surveillance site and therefore not randomized, we will use the double-difference method [<xref ref-type="bibr" rid="B40">40</xref>] to evaluate the primary outcomes in the intervention versus control sites. In this approach, Impact = (Y<sub>p(t>0)</sub> – Y<sub>p(t=0)</sub>) – (Yc<sub>(t>0)</sub> – Yc<sub>(t=0)</sub>) where Y<sub>p</sub> is the primary outcome in the intervention group and Y<sub>c</sub> is the outcome in the control group. For the double-difference method to work, it is essential that there are at least two pre-intervention data points. In addition, having many preintervention data points allows for the detection of shifts or interruptions in trends (if any) after the introduction of an intervention. A cross-sectional study conducted in the NUHDSS from 2008–2009 provides one time point of pre-intervention data on CVD risk in the intervention and control slums [<xref ref-type="bibr" rid="B27">27</xref>]. Note that cardiovascular risk reduction will be calculated by entering the outcomes in a non-laboratory-based CVD risk assessment chart [<xref ref-type="bibr" rid="B33">33</xref>]. Regression analyses will also be performed to investigate the association of each risk factor with the main outcomes such as blood pressure. Specifically, multivariate analyses will be used to adjust for known or perceived confounding variables while comparing outcomes between intervention and control sites. Prior to this, descriptive statistics will be applied to compare characteristics of the intervention and control sites. For behavioral risk factor analysis, we will use an interpretive descriptive approach with matrix comparisons between groups (such as sex, age group and site).</p><p>Cost-effectiveness analysis will be conducted according to the WHO framework for cost-effectiveness analysis [<xref ref-type="bibr" rid="B41">41</xref>]. This framework will consider intervention effectiveness data based on changes in blood pressure and overall predicted cardiovascular risk at the population level, as well as cost data on the intervention. The overall cost-effectiveness of the intervention package will be calculated in terms of DALYs averted per US dollar.</p><p>Intervention cost will be estimated using a micro-costing approach where feasible. Micro-costing is a process of systematically identifying and measuring resource utilization using a process tracking system and interviews with the local program team [<xref ref-type="bibr" rid="B42">42</xref>]. According to the Panel on Cost-Effectiveness in Health and Medicine, the theory and process of valuing costs through a micro-costing methodology rest on a three-step approach: identification, measurement and valuation of resources used [<xref ref-type="bibr" rid="B43">43</xref>]. For other non-specific costs, gross costing methods will be considered. Once resource utilization has been measured, the component-specific costs of the intervention can be computed by multiplying the quantity of each type of resource consumed by unit costs. The component-specific costs can be summed up to get the total costs of the intervention [<xref ref-type="bibr" rid="B42">42</xref>]. Finally, the outcome of this analysis will be the average cost for CVD risk reduction per participant per year. Data on costs and timings will be collected from the preparation of the intervention until the end of the intervention period. A yearly discount rate of 3% will be used for long-term modeling and projections [<xref ref-type="bibr" rid="B44">44</xref>]. Estimations will be extended to project the cost-effectiveness of the intervention package were it implemented on a national scale.</p><p>Finally, in order to determine the scalability and feasibility of the intervention package, a comprehensive process evaluation will be conducted involving analysis of operational data as well as qualitative sub-studies with beneficiaries and other relevant stakeholders such as local policy makers.</p><p>Additionally, since the package is to be implemented in a private health sector setting, it will be important to examine essential aspects of the intervention package that are needed to translate the package to the public sector.</p></sec><sec><title>Ethical approval</title><p>Ethical approval for this study was obtained from the Kenyan Medical Research Institute (KEMRI), reference KEMRI/RES/7/3/1 no. Non-SSC 399, dated 11 June 2012, renewable annually. Informed consent will be applied to all participants, and the overall study complies with the Declaration of Helsinki principles.</p></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>Although it is evident that a CVD epidemic is on the rise in most countries in SSA [<xref ref-type="bibr" rid="B10">10</xref>], there is limited evidence on the feasibility, cost-effectiveness and scalability of comprehensive primary prevention programs for CVD in these settings, in particular in the slum population. There is a scarcity of studies in SSA on community-based intervention for reduction of cardiovascular risk [<xref ref-type="bibr" rid="B30">30</xref>], and very few of them have looked into the cost-effectiveness [<xref ref-type="bibr" rid="B45">45</xref>] or scalability [<xref ref-type="bibr" rid="B46">46</xref>]. Considering the possibility of a double burden of disease that many countries in SSA are facing, it is essential to provide insight into the health impact and feasibility of primary prevention programs in order to enable policymakers and other stakeholders to make effective choices within their limited resources.</p><p>Despite the facts that the rise of CVD in SSA has been strongly linked with urbanization, and a sizable majority of the urban population in SSA is resident in slums, there are no specific studies or programs designed for this extremely vulnerable population. It is important to share knowledge and experiences on how this growing population at risk can be supported to reduce their cardiovascular burden. Therefore, it is important to implement programs in these challenging living circumstances and evaluate their feasibility and health effect. The SCALE UP study is unique in this regard.</p><p>In our intervention package we focus specifically on screening and treatment of hypertension as this is the main modifiable risk factor to achieve CVD risk reduction in SSA [<xref ref-type="bibr" rid="B47">47</xref>]. It is our hope that the combination of raised awareness through access to screening, improved access to quality treatment and the promotion of adherence will reduce hypertension rates in the study area to the extent that it is significantly detectable at population level. However, due to the depth of data that we will collect as part of the study, we will also be able to assess the effect of our intervention on behavioral risk factors for CVD such as tobacco use, excessive alcohol intake, poor dietary behavior and physical inactivity, as well as biological CVD risk proxies such as body composition (BMI) and blood glucose levels.</p><p>As mentioned previously, the content of our intervention arose from a theoretical cost-effectiveness analysis. Based on earlier studies done by APHRC and literature review. we made estimations of the different interventions possible in raising awareness, screening, treatment and promoting adherence in a slum setting. The final package is based on the theoretical effectiveness of each individual intervention component. Should our analysis prove the SCALE UP intervention be cost-effective, we will then work with local stakeholders and policymakers toward integrating the package in the larger health sector. Additionally, this vision of the scalability of the project not only implies the roll out of primary preventive services for CVD to more slums, but also provides the opportunity for the integration of other essential preventive services such as HIV testing or childhood vaccinations. Therefore, we anticipate and hope that this program will stimulate strengthening of weak health systems and structures in the ever-expanding slum settlements in SSA and beyond.</p><p>Our study is limited in the generalizability of the setting where the intervention package is tested. The DSA has been under surveillance for the past 10 years and may not be typical of other slum settings where the NUHDSS infrastructure does not exist. Another limitation of our study is that the intervention package cannot be evaluated in terms of its individual components. In other words, it will be impossible to tease out which parts of the intervention were most effective relative to others. However, we are confident that the logical composition of the complete intervention package will make it practical to implement as a whole in other similar settings should the multi-component intervention have an overall beneficial effect.</p><p>In conclusion, it is our hope that the outcomes of this study will inform policy makers and health professionals at various levels about the feasibility and cost-effectiveness of implementing community-based cardiovascular risk prevention programs in low- and middle-income countries, and specifically for the urban poor.</p></sec><sec><title>Trial status</title><p>Participant inclusion started in August 2012 and participant recruitment and follow-up at the intervention clinic will continue until December 2013.</p></sec><sec><title>Abbreviations</title><p>AIDS: Acquired immune deficiency syndrome; APHRC: African Population Health Research Center; AIGHD: Amsterdam Institute for Global Health and Development; BP: Blood pressure; CHW: Community health worker; CVD: Cardiovascular diseases; DALY: Disability adjusted life year; DSA: Demographic surveillance area; LMIC: Low- and middle-income country; NCD: Non-communicable diseases; NUHDSS: Nairobi Urban Health Demographic Surveillance System; RBS: Random blood sugar; SMS: Short message service; SSA: Sub-Saharan Africa; WHO: World Health Organization.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>SO and SV both conceptualized the study. SO, SV, GBG, CK, JL, LB, MH, CS, EM, RE, CA and AE all participated in the study design and contributed to the writing of the study protocol, drafting and editing of this manuscript. JL is the overall study principal investigator, and CK is the co-principal investigator. All authors read and approved the final manuscript.</p></sec><sec><title>Authors’ information</title><p>SO and SV are Senior Research Officers at APHRC, work as project managers of the SCALE UP project and are involved in a joint PhD program at the University of Amsterdam. CK and JL are part of the steering group of the SCALE UP study. GBG, CA, LB and EM give support from the AIGHD and University of Amsterdam. RE supports the project as a senior researcher at APHRC.</p><p>First authorship is shared by Samuel Oti and Steven van de Vijver.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional file 1</title><p>SCALE UP guidelines for management of hypertension in primary care.</p></caption><media xlink:href="1745-6215-14-409-S1.docx"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
Platelet-rich plasma (PRP) in chronic epicondylitis: study protocol for a randomized controlled trial	<sec><title>Background</title><p>Tendinopathy is a difficult problem to manage and can result in significant patient morbidity. Currently, the clinical use of platelet-rich plasma (PRP) in painful tendons is widespread but its efficacy remains controversial.</p></sec><sec><title>Methods/Design</title><p>This study is a single-center, randomized double-blind controlled trial. Eighty patients will be allocated to have ultrasound (US)-guided needling combined with a leukocyte-depleted (that is, pure) PRP or lidocaine each alternate week for a total of two interventions. Outcome data will be collected before intervention, and at 6 weeks, 3, 6, and 12 months after intervention. Main outcome measure: Changes in pain and activity levels, as assessed by Disabilities of the Arm, Shoulder and Hand (DASH-E, Spanish version) score, at 6 months. We will compare the percentage of patients in each group that achieve a successful treatment defined as a reduction of at least 25% in the DASH-E score. Secondary outcome measures include changes in DASH-E at 3 and 12 months, changes in pain as assessed by the visual analogue scale (VAS) at the 6-week, 3-, 6-, and 12-month follow-up, changes in sonographic features and neovascularity, and percentage of patients in each group with adverse reactions at 3, 6, and 12 months.</p></sec><sec><title>Discussion</title><p>The results of this study will provide insights into the effect of pure PRP in tendon and may contribute to identifying the best protocol for PRP application in tendinopathies.</p></sec><sec><title>Trial registration</title><p>ClinicalTrials.gov: <ext-link ext-link-type="uri" xlink:href="http://clinicaltrials.gov/show/NCT01945528">NCT01945528</ext-link>.</p></sec>	<contrib contrib-type="author" id="A1"><name><surname>Martin</surname><given-names>Jose I</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>joseignacio.martingomez@osakidetza.net</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Merino</surname><given-names>Josu</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>josu.merinoperez@osakidetza.net</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Atilano</surname><given-names>Leire</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>leire.atilanosantos@osakidetza.net</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Areizaga</surname><given-names>Luis M</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>luismaria.areizagahernandez@osakidetza.net</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Gomez-Fernandez</surname><given-names>Maria C</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>gomefert@hotmail.com</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Burgos-Alonso</surname><given-names>Natalia</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>natalia.burgosalonso@osakidetza.net</email></contrib><contrib contrib-type="author" corresp="yes" id="A7"><name><surname>Andia</surname><given-names>Isabel</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>iandia2010@hotmail.com</email></contrib>	Trials	<sec><title>Background</title><p>Tendon disorders (tendinopathies) are noteworthy in sports and occupational settings due to repetitive trauma and overuse; besides they are prevalent among individuals of all ages, and also part of the ageing process. The term ‘tendinopathy’ describes painful conditions affecting tendons associated with repetitive strain, overuse, ageing, degeneration, or poor biomechanics [<xref ref-type="bibr" rid="B1">1</xref>]. Tendinopathies worsen quality of life by causing pain and impairing mobility, decreasing the ability to perform daily activities, and compromising an active lifestyle.</p><p>Current research has produced several biological hypothesis based on histopathological, biochemical, and clinical findings that show cell apoptosis, angiofibroblastic features, or abnormal biochemical adaptations, largely suggesting that a failed healing response underlies the condition [<xref ref-type="bibr" rid="B2">2</xref>].</p><p>At present, minimally invasive interventions capable of boosting the healing response or counteracting degenerative changes in tendinopathy are being investigated. Among the emerging technologies, one investigational biological therapy, platelet-rich plasma (PRP), has been recently explored in several clinical studies [<xref ref-type="bibr" rid="B3">3</xref>]; in particular, several controlled clinical studies have examined the effect of PRP in epicondylitis [<xref ref-type="bibr" rid="B4">4</xref>-<xref ref-type="bibr" rid="B9">9</xref>]. PRP therapies are multitargeted approaches able to release a large pool of signals, producing an instructional biological microenvironment for local and migrating cell activities. Moreover, PRPs modulate inflammation and angiogenesis largely because of their ability to secrete high levels of growth factors and chemokines [<xref ref-type="bibr" rid="B10">10</xref>].</p><p>Different PRP formulations can be obtained depending on the preparation protocol, that is, single or double spinning. Most double spinning and also the buffy coat-based methods (single spinning) produce PRP with a high concentration of platelets and leukocytes relative to peripheral blood. These products are named L-PRP in contrast to pure PRP that contains a moderate concentration of platelets and absence or non-relevant concentration of leukocytes, and are generally obtained after a soft single spinning procedure [<xref ref-type="bibr" rid="B11">11</xref>]. The majority of clinical studies published up to now (>90%) have examined the efficacy of L-PRP injections with controversial results [<xref ref-type="bibr" rid="B12">12</xref>]. Assuming that leukocyte-released proteases may compromise the stability of platelet-released growth factors, better efficacy might be achieved with pure PRP injections. Currently, most published controlled studies have used corticosteroids as comparators. Instead, we propose lidocaine injections to avoid corticosteroids interference with the healing mechanisms.</p><p>We will compare the clinical outcomes and sonographic features of US-guided tenotomy combined with pure PRP with the outcome of US-guided tenotomy combined with lidocaine. This study protocol aims to evaluate the potential of pure PRP associated with needling for the treatment of epicondylitis.</p></sec><sec><title>Methods/Design</title><sec><title>Study design</title><p>B-PRPtendon is a patient and assessor blinded superiority-type randomized controlled trial; the study will be conducted at Hospital Universitario Cruces (HUC). The research protocol is approved by the Ethics Committee of HUC and authorized by the Spanish Agency of Medicines. A total of 80 patients will be randomly allocated into one of two groups. The two groups are: US-guided percutaneous needling tenotomy combined with PRP injection each alternate week for a total of two interventions; and US-guided needling tenotomy combined with lidocaine injection each alternate week for a total of two interventions.</p><p>The study will run for 2 years. Recruitment will be for 12 months with final follow-up at 1 year post treatment (Table <xref ref-type="table" rid="T1">1</xref>).</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Outcome assessments</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left" valign="bottom"> <hr/></th><th colspan="8" align="center" valign="bottom"><bold>Study period</bold><hr/></th></tr><tr><th align="left"> </th><th align="left"><bold>Enrollment</bold></th><th align="left"><bold>Allocation</bold></th><th align="left"><bold>Treatment</bold></th><th align="left"> </th><th colspan="4" align="center"><bold>Follow-up</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom"><bold>Timepoints</bold><hr/></td><td align="left" valign="bottom">−15 days<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"><bold>0</bold><hr/></td><td align="left" valign="bottom"><bold>2 weeks</bold><hr/></td><td align="left" valign="bottom"><bold>6 weeks</bold><hr/></td><td align="left" valign="bottom"><bold>3 months</bold><hr/></td><td align="left" valign="bottom"><bold>6 months</bold><hr/></td><td align="left" valign="bottom"><bold>12 months</bold><hr/></td></tr><tr><td align="left" valign="bottom">Eligibility screen<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td rowspan="4" colspan="2" align="left" valign="bottom"> <hr/></td><td rowspan="4" align="left" valign="bottom"> <hr/></td><td rowspan="4" align="left" valign="bottom"> <hr/></td><td rowspan="4" align="left" valign="bottom"> <hr/></td><td rowspan="4" align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Informed consent<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Laboratory tests<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Allocation<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"><bold>Interventions</bold><hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td colspan="9" align="left" valign="bottom"><bold>Assessments</bold><hr/></td></tr><tr><td align="left" valign="bottom"><bold>DASH</bold><hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td colspan="2" align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"><bold>VAS</bold><hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td colspan="2" align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom"><bold>Complications</bold><hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td colspan="2" align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left"><bold>Ultrasound</bold></td><td align="left"> </td><td align="left"> </td><td colspan="2" align="left">X</td><td align="left"> </td><td align="left">X</td><td align="left">X</td><td align="left">X</td></tr></tbody></table></table-wrap></sec><sec><title>Study population</title><p>Patients will be identified, and recruited in HUC and from primary care settings of Bizkaia. Primary care physicians and research nurses will introduce the trial to the patient and refer them to an orthopedic investigator for screening and potential recruitment<italic>.</italic></p><p>Patients will be treated and followed up in HUC at 6 weeks, 3, 6, and 12 months post treatment, and will be required to attend an appointment for sonographic assessments at 3, 6, and 12 months.</p><p>The inclusion criteria are: tendinopathy present in either lateral or medial elbow; patients will have failed conservative treatment; baseline elbow pain >3/10 during resisted wrist extension; history of at least two periods of elbow pain lasting >10 days; symptoms lasting at least 3 months or longer; body mass index (BMI) between 20 and 35; commitment to comply with all study procedures; and the patient must give written informed consent.</p><p>Patients may not enter the study if any of the following apply: presence of full tendon tear; BMI >35; systemic autoimmune rheumatologic disease (connective tissue diseases and systemic necrotizing vasculitis); poorly controlled diabetes mellitus (glycosylated hemoglobin above 9%); blood disorders (thrombopathy, thrombocytopenia, anemia with Hb <9); patient receiving immunosuppressive treatments; received local steroid injection within 3 months of randomization; received non-steroidal anti-inflammatory, opioids, or oral corticosteroids within 15 days before inclusion in the study; severe heart disease; patients unable to comply with scheduled visits, for work, or spend long periods away from their habitual residence; patients with active cancer or cancer diagnosed in the last 5 years; analytical diagnosis of hepatitis B, C, or HIV infection; pregnant or lactating; or people who are taking a drug in clinical investigation. Initial patient selection is conditioned to the negative results in the analytical tests for hepatitis B, C, or HIV infection.</p></sec><sec><title>PRP preparation</title><p>Peripheral venous blood is collected into three 9 mL tubes containing 3.8% (wt/vol) sodium citrate. The anticoagulated blood is centrifuged at 1,200 rpm for 6 min and PRP is collected taking care to avoid contamination with the buffy coat containing the leukocytes. Plasma is kept at room temperature until intervention; the delay between blood extraction and plasma administration will not be >4 h. To avoid blood lipids in the PRP, patients will fast or follow a fat-free diet during the 6 h prior to blood extraction.</p><p>Just preceding PRP administration, 10% calcium chloride will be added, at a final concentration of 22.6 mM (50 μL per 1 mL of PRP), and the 5 mL Luer Lok syringe is filled with the activated PRP.</p></sec><sec><title>Procedures</title><p>Interventions are performed by two radiologists with extensive clinical experience in musculoskeletal intervention procedures. Prior to needling and administration of PRP or lidocaine, an exploratory echography is performed to identify clefts of hypoechogenicity and/or changes in vascularity, and baseline sonographic characteristics are recorded, as described in Table <xref ref-type="table" rid="T2">2</xref>[<xref ref-type="bibr" rid="B13">13</xref>].</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Sonographic assessments</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th colspan="2" align="left"><bold>Echotexture grading scale</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">Normal<hr/></td></tr><tr><td align="left" valign="bottom">1a<hr/></td><td align="left" valign="bottom">Hypoechogenicity in less than one-third of the tendon<hr/></td></tr><tr><td align="left" valign="bottom">1b<hr/></td><td align="left" valign="bottom">Hypoechogenicity in between one-third and two-thirds of the tendon<hr/></td></tr><tr><td align="left" valign="bottom">1c<hr/></td><td align="left" valign="bottom">Hypoechogenicity in more than two-thirds of the tendon<hr/></td></tr><tr><td align="left" valign="bottom">2<hr/></td><td align="left" valign="bottom">Partial-thickness tear<hr/></td></tr><tr><td align="left" valign="bottom">3<hr/></td><td align="left" valign="bottom">Full-thickness tear<hr/></td></tr><tr><td align="left" valign="bottom"><bold>Neovascularization grading scale</bold><hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">No neovascularization<hr/></td></tr><tr><td align="left" valign="bottom">1<hr/></td><td align="left" valign="bottom">Neovessels on the tendon surface<hr/></td></tr><tr><td align="left" valign="bottom">2<hr/></td><td align="left" valign="bottom">1 or 2 intratendinous neovessels<hr/></td></tr><tr><td align="left" valign="bottom">3<hr/></td><td align="left" valign="bottom">3 or more intratendinous neovessels<hr/></td></tr><tr><td align="left" valign="bottom"><bold>Tendon size (mm)</bold><hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left"><bold>Calcifications</bold></td><td align="left"> </td></tr></tbody></table></table-wrap></sec><sec><title>Needle tenotomy with PRP (or lidocaine)</title><p>Ultrasound-guided percutaneous needle tenotomy with PRP (or lidocaine) will be performed each alternate week for a total of two interventions. Blood will be drawn from all the patients’ unaffected arm and PRP prepared as described above. Using a single skin portal, local anesthetic (2 mL of 1% lidocaine HCl 10 mg/mL) will be injected into the subcutaneous tissue of the lateral or medial elbow using a 20 G needle. Once the needle is in place, the 5 mL Luer Lok syringe loaded with the treatment is attached. Multiple longitudinal and transversal penetrations of the tendon are performed, and 3 to 5 mL of PRP (or lidocaine) is delivered in multiple depots during needling fenestrations. The injected volume is adapted to the morphometric characteristics of each patient. The same protocol with lidocaine is performed in the control group.</p></sec><sec><title>Study assessments</title><p>Study assessments will include the Spanish version [<xref ref-type="bibr" rid="B14">14</xref>] of the patient-reported outcome measure Disabilities of the Arm, Shoulder and Hand (DASH-E, Spanish version of the DASH, © Institute for Work & Health 2006), pain outcome as measured by the visual analogue scale (VAS) and changes in tendon structures and vascularity as assessed by Doppler sonography (Table <xref ref-type="table" rid="T2">2</xref>).</p><p>DASH is a patient-filled questionnaire based on the patient’s subjective assessment of symptoms and abilities to perform activities of daily living on the last week. The range of available scores is 0 (best) to 100 (worst). The questionnaire will be administered to patients at the baseline and during their follow-up visits at 6 weeks, 3, 6, and 12 months.</p><p>The DASH disability/symptom score is calculated as follows: DASH score = ((sum of n responses/n)–1) × 25, where <italic>n</italic> is equal to the number of completed responses.</p></sec><sec><title>Primary outcome measures</title><p>Successful treatment is defined as a reduction of >25% in the DASH-E score at 6 months post-treatment. The primary outcome measure is the percentage of patients that achieve a successful treatment. We will examine if the therapeutic success rates of the PRP and control groups are statistically different.</p></sec><sec><title>Secondary outcome measures</title><p>Secondary outcomes include: percentage of patients that achieve a successful treatment at 6 weeks, 3, and 12 months; pain reduction as measured by changes in pain rating on a visual analogue scale (VAS) with respect to baseline; changes in echogenicity and vascularity as assessed by Doppler sonography at 3, 6, and 12 months.</p><p>Frequency, severity, intensity, and duration of adverse events will be recorded and the ratio of adverse events in both groups compared.</p><p>A summary of the study design is shown in Figure <xref ref-type="fig" rid="F1">1</xref>.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p>Study flow chart.</p></caption><graphic xlink:href="1745-6215-14-410-1"/></fig></sec><sec><title>Sample size</title><p>The minimum sample size required to achieve scientifically valid results is 80. This sample size provides an 80% potency to detect any significant difference between the success rates in both groups (<italic>P</italic>1 = 0.93 and <italic>P</italic>2 = 0.65) with a level of significance of 5%, being each arm formed by 40 patient. We have assumed that the relative improvement with the PRP intervention is 1.43, assuming that the differences between PRP and lidocaine would be similar to the differences reported with corticoids [<xref ref-type="bibr" rid="B8">8</xref>], and a patient loss of approximately 20%. However, this is a randomized study which will help to gain insights into feasibility of recruitment, and into the number of patients who become lost to follow-up.</p></sec><sec><title>Randomization</title><p>Randomization will be performed in blocks of four, and equal allocation ratio will be achieved by means of a free informatics tool, EPIDAT3.1. Treatment assignments will be conducted by an independent researcher at Primary Care Investigation Unit of Bizkaia (UIAPB) who will not interfere with the study.</p></sec><sec><title>Blinding</title><p>All patients will be blinded to the treatment to which they are allocated, thus peripheral blood is drawn even if they are assigned to the control group. All outcome assessors (orthopedists and radiologists) will be blinded, but the radiologists applying the treatment will not.</p></sec><sec><title>Adverse events</title><p>The patients will be instructed to record any discomfort and/or any adverse reaction or event, whether or not it is related to the intervention. The investigator will evaluate and record the seriousness, intensity, expectedness, and causality relationship, and a written communication will be sent to the sponsor. Pertinent adverse events will be notified to the licensing authorities (Ethics Committee of HUC and Spanish Agency of Medicines) by fax or email.</p></sec><sec><title>Statistical analysis</title><p>Categorical variables will be presented as rates and percentages. Demographic and clinical baseline characteristics will be assessed to confirm comparability between groups. If any clinically relevant data are unbalanced, we will perform an adjusted analysis. Categorical variables will be presented as frequencies and percentages, and we will use the mean and standard deviation for data with a normal distribution and median and interquartile range for non-parametric data. Potential efficacy analysis will be performed as intention to treat, comparing the percentage of patients that have achieved therapeutic success in each group using the chi-squared test with <italic>P</italic> < 0.05 deemed statistically significant. All statistical analysis will be performed using the SAS 9.2 version.</p></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>The present study had three principal objectives: first, to investigate if pure PRP interventions reduce symptoms and improve function by using the patient self-reported DASH-E questionnaire; second, to examine the clinical efficacy of needling combined with PRP injections in pain; and third, to identify the potential structural changes in the tendon after PRP treatment.</p><p>Multiple commercial protocols to prepare PRP are currently available on the market. Depending on the specific preparation protocol the qualitative and quantitative composition of PRP varies, and most likely so do the biological effects. Current experimental research postulates different efficiency among PRP formulations [<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B16">16</xref>], mainly leukocyte-rich PRP (L-PRP) and leukocyte-depleted PRP (pure PRP). In fact, experimental research has shown that L-PRP is more pro-inflammatory when injected in rabbits [<xref ref-type="bibr" rid="B15">15</xref>], and it increases the levels of metalloproteases when assayed in tenocyte cultures compared to pure PRP [<xref ref-type="bibr" rid="B17">17</xref>].</p><p>On the basis of these preliminary experimental results, and the clinical results in the conservative management of knee OA in which L-PRP injections induced more transient post-injection swelling and pain than pure PRP [<xref ref-type="bibr" rid="B18">18</xref>], we expect fewer adverse reactions using pure PRP (leukocyte-depleted).</p><p>To date, all controlled clinical trials in epicondylitis have been performed with L-PRP [<xref ref-type="bibr" rid="B4">4</xref>-<xref ref-type="bibr" rid="B9">9</xref>]. In this pilot study, we aim to examine the efficacy of pure PRP, and potential comparative efficacy studies (L-PRP <italic>vs.</italic> pure PRP) would be performed in a subsequent step forward. Moreover, when compared to published clinical protocols performed with L-PRP, we might anticipate improved function and reduced pain using pure PRP associated with the needling intervention.</p><p>Additionally, there is no consensus about the frequency and number of PRP treatments in chronic injuries. Thus, whether two interventions would be more efficient than a single PRP application remains to be clarified. Previous studies in tendinopathy have not found any structural change after one PRP injection pointing out that one single intervention may be insufficient to induce structural changes [<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>]. In this pilot study we will explore whether two injections could modify the structural characteristics of the injured tendon. In addition, the optimal volume of PRP has not been defined so far thus we will adjust the injected volume (3–5 mL) to the individual anatomical characteristics.</p><p>The results of this study will provide insights into the effect of pure PRP in tendons and may contribute to identifying the best protocol for PRP application in tendinopathies.</p></sec><sec><title>Trial status</title><p>Recruitment commencement is planned by January 2014 and the expected average enrollment rate is four patients every month. Data collection will continue for one year after the last recruited patient.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>MJI, AL, MJ, and ALM designed the study protocol; G-FMC, B-AN, and AI wrote the clinical protocol and obtained authorization from the Ethics Committee and Spanish Agency of Drugs. All authors read and approved the final manuscript.</p></sec>
Protocol for the New Medicine Service Study: a randomized controlled trial and economic evaluation with qualitative appraisal comparing the effectiveness and cost effectiveness of the New Medicine Service in community pharmacies in England	<sec><title>Background</title><p>Medication non-adherence is considered an important cause of morbidity and mortality in primary care. This study aims to determine the effectiveness, cost effectiveness and acceptability of a complex intervention delivered by community pharmacists, the New Medicine Service (NMS), compared with current practice in reducing non-adherence to, and problems with, newly prescribed medicines for chronic conditions.</p></sec><sec><title>Methods/design</title><p>Research subject group: patients aged 14 years and above presenting in a community pharmacy for a newly prescribed medicine for asthma/chronic obstructive pulmonary disease (COPD); hypertension; type 2 diabetes or anticoagulant/antiplatelet agents in two geographical regions in England.</p><p>Design: parallel group patient-level pragmatic randomized controlled trial.</p><p>Interventions: patients randomized to either: (i) current practice; or (ii) NMS intervention comprising pharmacist-delivered support for a newly prescribed medicine.</p><p>Primary outcomes: proportion of adherent patients at six, ten and 26 weeks from the date of presenting their prescriptions at the pharmacy; cost effectiveness of the intervention versus current practice at 10 weeks and 26 weeks; in-depth qualitative understanding of the operationalization of NMS in pharmacies.</p><p>Secondary outcomes: impact of NMS on: patients’ understanding of their medicines, pharmacovigilance, interprofessional and patient-professional relationships and experiences of service users and stakeholders.</p><p>Economic analysis: Trial-based economic analysis (cost per extra adherent patient) and long-term modeling of costs and health effects (cost per quality-adjusted-life-year) will be conducted from the perspective of National Health Service (NHS) England, comparing NMS with current practice.</p><p>Qualitative analysis: a qualitative study of NMS implementation in different community settings, how organizational influences affect NMS delivery, patterns of NMS consultations and experiences of professionals and patients participating in NMS, and patients receiving current practice.</p><p>Sample size: 250 patients in each treatment arm would provide at least 80% power (two-tailed alpha of 0.05) to demonstrate a reduction in patient-reported non-adherence from 20% to 10% in the NMS arm compared with current practice, assuming a 20% drop-out rate.</p></sec><sec><title>Discussion</title><p>At the time of submission of this article, 58 community pharmacies have been recruited and the interventions are being delivered. Analysis has not yet been undertaken.</p></sec><sec><title>Trial registration</title><p>Current controlled trials: <ext-link ext-link-type="uri" xlink:href="http://www.controlled-trials.com/ISRCTN23560818">ISRCTN23560818</ext-link></p><p>Clinical Trials US (clinicaltrials.gov): <ext-link ext-link-type="uri" xlink:href="http://clinicaltrials.gov/show/NCT01635361">NCT01635361</ext-link></p></sec>	<contrib contrib-type="author" id="A1"><name><surname>Boyd</surname><given-names>Matthew</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>matthew.boyd@nottingham.ac.uk</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Waring</surname><given-names>Justin</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>Justin.waring@nottingham.ac.uk</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Barber</surname><given-names>Nick</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>nicholas.barber@ucl.ac.uk</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Mehta</surname><given-names>Rajnikant</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>Rajnikant.mehta@nottingham.ac.uk</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Chuter</surname><given-names>Antony</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>chuter@me.com</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Avery</surname><given-names>Anthony J</given-names></name><xref ref-type="aff" rid="I6">6</xref><email>tony.avery@nottingham.ac.uk</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Salema</surname><given-names>Nde-Eshimuni</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>Ndeshi.salema@nottingham.ac.uk</email></contrib><contrib contrib-type="author" id="A8"><name><surname>Davies</surname><given-names>James</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>james.davies.11@ucl.ac.uk</email></contrib><contrib contrib-type="author" id="A9"><name><surname>Latif</surname><given-names>Asam</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>Asam.latif@nottingham.ac.uk</email></contrib><contrib contrib-type="author" id="A10"><name><surname>Tanajewski</surname><given-names>Lukasz</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>lukasz.tanajewski@nottingham.ac.uk</email></contrib><contrib contrib-type="author" corresp="yes" id="A11"><name><surname>Elliott</surname><given-names>Rachel A</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>rachel.elliott@nottingham.ac.uk</email></contrib>	Trials	<sec><title>Background</title><p>Favorable outcomes in long-term conditions depend on self-management by patients, including appropriate medicines use. About 25% of medicines prescribed for long-term conditions are not taken as directed [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>], and 15% of people receiving new medicines take few, if any, doses [<xref ref-type="bibr" rid="B3">3</xref>]. Many have problems with their medicines and have information needs, but often fail to discuss these concerns with their prescriber. Furthermore, prescribers do not ask about, and so are generally unaware of, patients’ behavior regarding following instructions, experimentation and self-medication with other therapies [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B5">5</xref>]. Prescribers may overestimate adherence [<xref ref-type="bibr" rid="B6">6</xref>], and be reluctant to voice suspicions about non-adherence [<xref ref-type="bibr" rid="B7">7</xref>]. Harm caused by non-adherence includes poor health-related quality of life, increased hospitalizations and premature mortality [<xref ref-type="bibr" rid="B8">8</xref>-<xref ref-type="bibr" rid="B11">11</xref>]. Wider burden includes cost to patients, healthcare providers and society. Improving adherence in asthma, type 2 diabetes and hypertension could save the English National Health Service (NHS England) £290 million, annually [<xref ref-type="bibr" rid="B12">12</xref>].</p><sec><title>Definition of adherence</title><p>In this study we have taken the definition of adherence from the World Health Organization: ‘the extent to which a person’s medication-taking behavior, following a diet, and/or executing lifestyle changes, corresponds with agreed recommendations from a healthcare provider’ [<xref ref-type="bibr" rid="B2">2</xref>]. In this study we operationalize adherence behavior by defining a patient as non-adherent if any scheduled doses were missed at various time points as a result of different adherence measures deployed.</p></sec><sec><title>Pharmacist-provided interventions to improve medicines use</title><p>Pharmacist-provided interventions exist to facilitate improved medicines usage and health outcomes [<xref ref-type="bibr" rid="B13">13</xref>-<xref ref-type="bibr" rid="B26">26</xref>]. They are also viewed by some as a means to enhance pharmacist professional status [<xref ref-type="bibr" rid="B27">27</xref>]. Some interventions are effective [<xref ref-type="bibr" rid="B13">13</xref>-<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B25">25</xref>,<xref ref-type="bibr" rid="B26">26</xref>], but not all [<xref ref-type="bibr" rid="B20">20</xref>,<xref ref-type="bibr" rid="B22">22</xref>-<xref ref-type="bibr" rid="B24">24</xref>]. There is little evidence around cost effectiveness of clinical community pharmacy interventions [<xref ref-type="bibr" rid="B28">28</xref>-<xref ref-type="bibr" rid="B30">30</xref>]. Most studies have methodological limitations: absence of a control, exclusion of pharmacist employment cost, use of intermediate outcomes, exclusion of health benefits, and absence of incremental analysis [<xref ref-type="bibr" rid="B28">28</xref>,<xref ref-type="bibr" rid="B31">31</xref>]. Interventions to improve medicines adherence have been disappointing in producing sustained behavior change [<xref ref-type="bibr" rid="B32">32</xref>], consistent health benefits, [<xref ref-type="bibr" rid="B33">33</xref>] or demonstrable cost effectiveness [<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B34">34</xref>]. This can be attributed to poor study design, not using evidence on reasons for non-adherence, poor intervention development, lack of understanding of intervention complexity and its effects, and lack of integration into service delivery [<xref ref-type="bibr" rid="B35">35</xref>-<xref ref-type="bibr" rid="B37">37</xref>].</p><p>Implementing clinical services run by community pharmacists has been hampered by insufficient integration into patient pathways; poorly developed relationships between pharmacists and general practitioners (GPs); lack of access to patient information; inadequate methods for targeting services; and pharmacists’ lack of willingness to provide the service [<xref ref-type="bibr" rid="B38">38</xref>]. Medicines Use Reviews (MURs) are a recent example of a community pharmacy-based intervention. However, there has been variable uptake of MURs, mostly in chain pharmacies [<xref ref-type="bibr" rid="B39">39</xref>-<xref ref-type="bibr" rid="B41">41</xref>]. MURs have been carried out to variable standards by pharmacists, [<xref ref-type="bibr" rid="B42">42</xref>,<xref ref-type="bibr" rid="B43">43</xref>], partly due to variable understanding of what constitutes an MUR [<xref ref-type="bibr" rid="B44">44</xref>].</p></sec><sec><title>The New Medicine Service</title><p>The New Medicine Service (NMS) [<xref ref-type="bibr" rid="B45">45</xref>] in England is based on evidence derived from our research that studied doctor-patient communication concerning medicines [<xref ref-type="bibr" rid="B46">46</xref>-<xref ref-type="bibr" rid="B49">49</xref>]. This and subsequent work established that problems with newly prescribed medicines appeared rapidly, were widespread and that a significant proportion of patients on a long-term medication quickly become non-adherent [<xref ref-type="bibr" rid="B50">50</xref>]. A pharmacist’s intervention could significantly reduce reported problems and non-adherence in a cost effective manner [<xref ref-type="bibr" rid="B51">51</xref>,<xref ref-type="bibr" rid="B52">52</xref>]. Within the NHS community pharmacy contract, the NMS is classified as an Advanced Service, whereby community pharmacists can opt to provide the service from their pharmacy [<xref ref-type="bibr" rid="B45">45</xref>]. This comes at a time when efficient medicines use could not be more important, in the face of economic pressures on the public sector budget. England’s ageing population now receives 50% more prescriptions items per capita for conditions such as heart disease, stroke, diabetes, COPD and asthma than in 1990 [<xref ref-type="bibr" rid="B53">53</xref>]. This proposed study is therefore timely as it is essential to evaluate whether NMS is effective, cost effective and acceptable to patients and healthcare providers. Supplementary policy-relevant outcomes from this service are opportunities to intervene regarding lifestyle and potential improved tracking of medicines-related adverse events by community pharmacists and patients [<xref ref-type="bibr" rid="B54">54</xref>].</p></sec><sec><title>Aims of the study</title><p>The aims of this study are to:</p><p>•evaluate effectiveness and cost effectiveness of the NMS, from an NHS perspective, to inform decisions about continuation of the service</p><p>•explore operation of the NMS to determine acceptability to patients, pharmacists and GPs, indicators of successful implementation, generalizability and replicability across four therapeutic groupings in multiple pharmacy settings</p></sec><sec><title>Specific objectives</title><p>This study has two work streams: an RCT and a qualitative investigation/appraisal. The purpose of the RCT is to evaluate effectiveness and cost effectiveness of the NMS from an NHS perspective to inform decisions about continuation of the service.</p><p>Data collected from the RCT will be used to:</p><p>•evaluate the impact of the NMS on patient medicines-taking behavior, patient outcomes, and cost effectiveness from an NHS perspective</p><p>•determine patients’ understanding of their medicines and the extent to which they are informed and supported in their medicines-related behavior</p><p>•examine whether NMS encourages pharmacovigilance by community pharmacists and patients</p><p>•inform and support future implementation and support development of outcome and quality measures for community pharmacy</p><p>The purpose of the qualitative workstream is to understand how NMS is implemented and experienced as a situated complex healthcare intervention. This involves investigation and analysis of:</p><p>•the implementation of NMS in different community pharmacy settings to determine the influence of local organizing factors</p><p>•the organization and configuration of NMS in different community pharmacies to determine variation in NMS operation and delivery</p><p>•the delivery of the NMS as a situated social practice, including interactions between multiple stakeholders (GP-pharmacist; pharmacist-patient; patient-pharmacist)</p><p>•the experiences and reflections of professionals and patients involved in the NMS</p><p>Evidence generated will allow better understanding of results obtained from both technology appraisal and NMS audit, through providing detailed case evidence of how and why the NMS is implemented in practice, including variations in implementation and uptake.</p></sec></sec><sec><title>Methods/design</title><sec><title>Trial design</title><p>This study will involve community pharmacies in England. It is a multi-center, pragmatic RCT involving a parallel group design. The health technology appraisal will assess the effectiveness and cost effectiveness of NMS via a patient-level pragmatic RCT of NMS versus current practice combined with modeling of long-term economic impact. Results generated from the technology appraisal will be combined with NMS audit data to provide wider estimates of effectiveness and cost effectiveness.</p></sec><sec><title>Qualitative study design</title><p>The qualitative workstream comprises three distinct research activities:</p><p>a) pharmacy profiling will generate data on the organizational variables associated with the implementation, configuration and delivery NMS. Profiling will involve ethnographically informed observations and semi-structured interviews within participating community pharmacies, including non-participatory observations of work organization and pharmacy practice, guided tours, shadowing of pharmacy staff and interviews with pharmacy staff over a three to five day period. Observations will be guided by a standardized profiling tool to acquire common data set from across different sites.</p><p>b) patient tracking will generate data on the real-time delivery of NMS as a situated social interaction. Based upon pharmacy profiles, it involves producing a descriptive understanding of the ‘expected’ patient pathway to identify key interaction, communication and decision-making points followed by focused observations of, and interviews with, pharmacist and patient ‘before’, ‘during’ and ‘after’ the NMS consultation. A range of methods of recording data will be offered to participants, including video and audio recording or direct observation by research staff. Participant patients to be tracked will be recruited via the patient pool from the RCT.</p><p>c) stakeholder interviews will generate additional qualitative understanding on the organization and experience of NMS from the perspective of community pharmacy staff including pharmacists, dispensers and technicians. Patients who decline the invitation for an NMS in the pharmacy will also be invited to take part in short semi-structured interview to incorporate their views. Additionally, one-to-one interviews will be conducted with patients whose voices are seldom heard, and pharmacist and GP interviews will also be conducted to explore the professional perspective of the NMS.</p></sec><sec><title>Eligibility for pharmacies to join the trial</title><p>Community pharmacies providing NMS in East Midlands and South Yorkshire (EMSY) and Greater London (GL) are eligible. Pharmacy selection will take into account known variables that influence organizational structures, workflow and integration, including:</p><p>•ownership: independent; small multiple; large multiple; supermarket</p><p>•proximity to GP: co-location; less than 500 meters; 500 meters to one km; over one km</p><p>•setting: urban; suburban; rural</p><p>•economic deprivation: based on Economic Deprivation Index</p></sec><sec><title>Inclusion criteria</title><p>Community pharmacies will be eligible to take part in the RCT if they meet the following criteria:</p><p>•they provide care for patients starting a new medicine for one of four therapeutic groups: asthma/chronic obstructive pulmonary disease (COPD), type 2 diabetes, antiplatelets/anticoagulants or hypertension</p><p>•they are able to understand the participant study documents</p><p>•they are able and willing and able to provide consent</p><p>•they are accredited to provide the New Medicines Service</p></sec><sec><title>Exclusion criteria</title><p>Community pharmacies will not be eligible to take part in the RCT if:</p><p>•they are outside of the EMSY and GL area</p><p>•they currently do not provide NMS or undertake an insufficient number to ensure recruitment to the study (< two per week).</p><p>•the pharmacy does not have a regular pharmacist (working on most days)</p><p>•the pharmacist is unable or unwilling to provide consent</p></sec><sec><title>Eligibility for patients to join the trial</title><sec><title>Patient inclusion criteria</title><p>The study will include:</p><p>•community dwelling patients eligible for the NMS (starting a new medicine for asthma/COPD, type 2 diabetes, antiplatelet/anticoagulation or hypertension)</p><p>•participants who are able to understand and consent to the NMS and also the study and who are willing to provide written consent/assent.</p></sec><sec><title>Patient exclusion criteria</title><p>•those not eligible for the NMS such as</p><p>•patients collecting a repeat prescription for a medicine (that is, not new)</p><p>•patients collecting a medicine where the only change from the previous medicine involves a dosage or formulation change only</p><p>•participants who are unable to understand patient/participant study documents</p><p>•participants who are unable and unwilling to provide written consent/assent</p><p>•those patients aged 13 and under</p></sec></sec><sec><title>Recruitment</title><sec><title>Community pharmacies</title><p>Pharmacies will be recruited using a pragmatic convenience sample to enable a representative sample across the four eligibility criteria (ownership, proximity to the GP, setting, and economic deprivation). Local pharmaceutical committees and other regional and national pharmacy bodies in the GL and EMSY have been approached to assist in finding suitable research sites, as were the superintendent pharmacists of a range of multiple-owned pharmacy organizations.</p></sec><sec><title>Patients for RCT</title><p>The initial approach to participants will be from the community pharmacy that is providing the patient with NMS. Patients who have consented to the NMS will be invited to take part in the study. It will be explained to patients that taking part in the study is optional and if they decide not to take part they will be offered the NMS service as normal.</p><p>The study-designated pharmacist will inform the patient or their nominated representative (other individual or other body with appropriate jurisdiction), of all aspects pertaining to participation in the study. If the patient agrees to take part, the pharmacist will randomize the patient to receive the NMS or allocated to receive current practice. Patients could, if they were eligible for the NMS, volunteer for the study once they have seen the study poster displayed in the pharmacy. If a study-designated pharmacist is not present in the pharmacy at the time the patient presents with a prescription for a new medicine eligible for the NMS, the patient cannot be recruited into the study.</p></sec><sec><title>Participants for the qualitative analysis</title><sec><title>Pharmacy profiles</title><p>Twenty pharmacies will be recruited for pharmacy profiling from those sites recruited to the RCT. These community pharmacies will be selected across geographical areas with the intention of investigating differences in pharmacy type, location, ownership, organization, and staffing.</p></sec><sec><title>Patient tracking</title><p>Participants will be identified and their NMS consultation recorded from a range of pharmacies and patient characteristics. Written informed consent will be taken before undertaking observation and interviews. The first two patients recruited into the study in each pharmacy will not be considered for tracking to allow the study pharmacist opportunity to familiarize themselves with the study paperwork.</p><p>Semi-structured interviews will be undertaken with community pharmacy and GP staff. A purposive sampling strategy will be used to identify and recruit a range of participants linked to the participating pharmacies: pharmacists, pharmacy technicians, counter assistants, GPs, nurse prescribers and patients in both geographical regions. Written informed consent will be obtained before interviews will take place.</p></sec></sec></sec><sec><title>Interventions</title><sec><title>Current practice</title><p>Those patients randomized to this arm will receive current practice where the patient receives their prescription, with advice as clinically necessary from the pharmacist at point of supply. It is not routine for patients to have any follow-up in this arm. Patients in this arm are not restricted from contacting the pharmacist for advice should they wish to.</p></sec><sec><title>NMS intervention</title><p>Those patients randomized to this arm will receive the NMS. The NMS will be offered by community pharmacists to people starting a new medicine for asthma/COPD, type 2 diabetes, hypertension or antiplatelet/anticoagulant treatment. The NMS can be summarized as patient engagement, intervention and follow-up (Figure <xref ref-type="fig" rid="F1">1</xref>). This is described in the NMS service specification [<xref ref-type="bibr" rid="B45">45</xref>].</p><fig id="F1" position="float"><label>Figure 1</label><caption><p>NMS intervention.</p></caption><graphic xlink:href="1745-6215-14-411-1"/></fig></sec></sec><sec><title>Study pharmacist training</title><p>All pharmacists who would be recruiting and consenting patients to be in the study will be required to attend a one day face-to-face training session at one of the host institutions. Training will be delivered by the study team and includes an overview of the study, study aims and objectives, research governance (including principles of Good Clinical Practice (GCP)), conducting the study in your pharmacy and interactive practice sessions.</p><p>Pharmacist study training will be followed up by an abridged training session in each pharmacy. This training provides all staff members with an overview of the study and enables them to assist the pharmacist in identifying potential study participants.</p></sec><sec><title>Allocation of trial intervention</title><p>Following the consenting process, patients will be randomized into their respective study groups. The randomization sequence will be generated by the study statistician (RM).</p><p>Patients will be 1:1 randomized into one of the two study arms stratified by drug/disease group within each pharmacy using the SAS statistical software (version 9) (SAS UK Headquarters Wittington House Henley Road Medmenham Marlow, Buckinghamshire SL7 2 EB). Block randomization at the pharmacy level is necessary to avoid imbalances due to individual pharmacy variability. Separate randomization sequences were produced for patients 16 years and over and for patients aged 14 and 15 years old. The separation by age is required as study documents for the 14 and 15 year-old group include patient assent and parental consent instead of the usual normal patient consent forms. It is anticipated that there will be an unequal distribution across the four groups; therefore, stratification by drug/disease group is necessary, each with a different baseline adherence and/or effect size. The intervention arm will receive the NMS as per service specification while the control arm will receive care as per current practice. Concealment of sequence allocation will be achieved as pharmacists will randomly allocate patients to the two study arms using disease and age group specific, sequentially numbered tamper-proof opaque sealed envelopes containing details of allocation group. Periodic checks will be undertaken by a researcher to assess the integrity of study sites adhering to the randomization procedures. The sequence of treatment allocations will remain concealed until analysis is completed. Exceptions to this will include the need to reveal the randomization code because the patient has been identified as suitable for the qualitative work stream. As patients are registered into the study, checks will be conducted to ensure compliance with the registration protocol.</p></sec><sec><title>Outcome measures</title><p>A successfully implemented NMS has five levels (see Figure <xref ref-type="fig" rid="F2">2</xref>) [<xref ref-type="bibr" rid="B55">55</xref>].</p><fig id="F2" position="float"><label>Figure 2</label><caption><p>Intended outcomes of NMS intervention.</p></caption><graphic xlink:href="1745-6215-14-411-2"/></fig><p>A summary of outcome measures, at which time points they are collected, and the method of collection is provided in Table <xref ref-type="table" rid="T1">1</xref>.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Summary of outcome measures</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Outcome measure</bold></th><th align="left"><bold>Time point recorded</bold></th><th align="left"><bold>Method of recording</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Health status<hr/></td><td align="left" valign="bottom">0 weeks<hr/></td><td rowspan="4" align="left" valign="bottom">Questionnaire<hr/></td></tr><tr><td align="left" valign="bottom">EuroQol-5 dimension-3<hr/></td><td align="left" valign="bottom">6 weeks<hr/></td></tr><tr><td align="left" valign="bottom">Level instrument<hr/></td><td align="left" valign="bottom">10 weeks<hr/></td></tr><tr><td align="left" valign="bottom">(EQ-5D-3 L)<hr/></td><td align="left" valign="bottom">26 weeks<hr/></td></tr><tr><td align="left" valign="bottom">Adherence<hr/></td><td align="left" valign="bottom">6 weeks<hr/></td><td align="left" valign="bottom">Questionnaire<hr/></td></tr><tr><td align="left" valign="bottom">Morisky’s medication<hr/></td><td align="left" valign="bottom">10 weeks<hr/></td><td rowspan="2" align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Adherence scale 8-item version (MMAS-8)<hr/></td><td align="left" valign="bottom">26 weeks<hr/></td></tr><tr><td align="left" valign="bottom">Adherence<hr/></td><td align="left" valign="bottom">6 weeks<hr/></td><td rowspan="3" align="left" valign="bottom">Questionnaire<hr/></td></tr><tr><td align="left" valign="bottom">Visual analogs scale<hr/></td><td align="left" valign="bottom">10 weeks<hr/></td></tr><tr><td align="left" valign="bottom">(VAS)<hr/></td><td align="left" valign="bottom">26 weeks<hr/></td></tr><tr><td align="left" valign="bottom">Adherence<hr/></td><td align="left" valign="bottom">6 weeks<hr/></td><td rowspan="3" align="left" valign="bottom">Telephone interview<hr/></td></tr><tr><td align="left" valign="bottom">NMS service question<hr/></td><td align="left" valign="bottom">10 weeks<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">26 weeks<hr/></td></tr><tr><td align="left" valign="bottom">Medicines<hr/></td><td align="left" valign="bottom">6 weeks<hr/></td><td rowspan="3" align="left" valign="bottom">Questionnaire<hr/></td></tr><tr><td align="left" valign="bottom">Understanding<hr/></td><td align="left" valign="bottom">10 weeks<hr/></td></tr><tr><td align="left" valign="bottom">Beliefs About Medicines Questionnaire (BMQ)<hr/></td><td align="left" valign="bottom">26 weeks<hr/></td></tr><tr><td align="left" valign="bottom">Consultation satisfaction<hr/></td><td rowspan="3" align="left" valign="bottom">6 weeks<hr/></td><td rowspan="3" align="left" valign="bottom">Questionnaire<hr/></td></tr><tr><td align="left" valign="bottom">Medical interview satisfaction scale-15<hr/></td></tr><tr><td align="left" valign="bottom">(MISS-15)<hr/></td></tr><tr><td align="left">Health resource use</td><td align="left">0 to 26 weeks inclusive</td><td align="left">Self-completed diary</td></tr></tbody></table></table-wrap></sec><sec><title>Primary outcome measures</title><p>The primary endpoint for the RCT is to establish the impact of the NMS on patient medicines-taking behavior through measurement of self-reported adherence. The adherence measure in the interview schedule for the NMS intervention [<xref ref-type="bibr" rid="B56">56</xref>] and follow-up interview is a simple question: ‘People often miss taking doses of their medicines, for a wide range of reasons. Have you missed any doses of your new medicine, or changed when you take it? (Prompt: when did you last miss a dose?)’</p><p>Adherence in our original work was measured using self-report [<xref ref-type="bibr" rid="B57">57</xref>], using a similar question to that in the NMS interview. The patient was defined as non-adherent if any doses were missed in the previous seven days, partly because simply asking patients about their adherence will identify at least 50% of those with low adherence, with a specificity of 87% [<xref ref-type="bibr" rid="B58">58</xref>]. We will validate the NMS adherence measure by using an existing validated scale alongside it (Morisky Eight Item Medication Adherence Scale (MMAS-8)) [<xref ref-type="bibr" rid="B59">59</xref>].</p></sec><sec><title>Secondary outcome measures</title><p>We are collecting data on a number of secondary outcome measures relating to number of medicines-related problems reported by patients, and resolved; adverse events and NHS contacts associated with adverse events; health status; resource use and costs; length of initial and subsequent NMS consultations; NHS contact (primary and secondary care) and non-medical costs.</p></sec><sec><title>Composite outcomes</title><p>In the original work, most ‘withdrawals’ were that the patients in the intervention arm were referred back to the GP by the pharmacist due to side effects, lack of effect or patient non-adherence’ [<xref ref-type="bibr" rid="B51">51</xref>]. In this study, these events are being recorded and will be combined into a composite outcome.</p><p>Adherence will be reported as adherence in the group eligible to be adherent (that is, still meant to be taking the medicine). We will also report patients referred back to the GP (whether a new medicine is prescribed or not) as a separate outcome. This information will be available from the six week call for both intervention and current practice arm.</p><p>A composite outcome that combines proportion of adherent patients and patients appropriately referred back to the GP will be derived and summary statistics presented.</p></sec><sec><title>Data for the economic analysis</title><sec><title>Costs of the intervention</title><p>Costs will be incurred at the patient level, in delivering the intervention. Variable costs refer to items where the quantity of resources used is determined only by the need for them as inputs to individual patient care. Variable resource use associated with the interventions (time spent, costs of telephone calls, printing and posting) will be recorded for each patient. Fixed costs are those costs that are not affected by patient activity in the short term. UK standard costs will be used for unit costs. This may somewhat over- or under-estimate local unit costs, but allows explicit comparison of costs and local adjustments can be made. Unit costs associated with the intervention will be obtained from the Personal Social Services Research Unit (PSSRU) [<xref ref-type="bibr" rid="B60">60</xref>], Department of Health reference tables and other reference costs.</p></sec><sec><title>Clinical and economic impact of non-adherence</title><p>The trial will not observe patient outcomes and NHS costs resulting from non-adherence to their newly prescribed medicine. Rather, outcomes will be derived from published evidence on the link between adherence improvement and impact on health. The NMS study is not designed to calculate the impact of the intervention on patient health outcomes, either in terms of sample size or length of follow-up. Use of proxy measures such as number of primary and secondary care contacts (hospital admissions, accident and emergency visits and outpatient visits) may be subject to difficulties if considered as patient outcomes. This is because the intervention may lead to increased NHS contact in the short term. Thus, we will estimate the long-term effect of the observed adherence improvements on patient outcomes and NHS costs. Data on natural history of diseases, treatment effectiveness, resource use, and health status (utility) will be obtained from published literature to populate the model.</p><p>We will develop at least one treatment pathway model for each of the four treatment groups targeted by NMS, encompassing the consequences of being adherent or non-adherent (see Figure <xref ref-type="fig" rid="F3">3</xref>). A common generic approach will be used to develop the models. We will undertake the economic analysis from the perspective of the English NHS in terms of the direct costs of providing an intervention to improve medicines adherence in chronic illnesses and the long-term costs of managing the conditions for adherent and non-adherent patients (to estimate cost of non-adherence).</p><fig id="F3" position="float"><label>Figure 3</label><caption><p>Decision-analytic model for NMS economic evaluation.</p></caption><graphic xlink:href="1745-6215-14-411-3"/></fig><p>Markov models will be designed and populated, incorporating the measures of uncertainty around the point estimates, to conduct probabilistic sensitivity analysis. The UK Treasury recommended 3.5% discount rate for both costs and outcomes will be used.</p><p>Literature review will be used to obtain published utility weights to allow quality adjusted like year (QALY) generation and cost utility analysis. The baseline EuroQol 5D-3 L (EQ-5D) in our trial cohort will be incorporated into the QALY generation.</p><p>We will model the effect of the observed adherence improvements on patient outcomes and NHS costs. Probability, resource use and health status (utility) data will be obtained from published literature to populate the model.</p></sec></sec><sec><title>Data for the qualitative study</title><p>It is anticipated that data for the qualitative study will be obtained from 20 pharmacy profiles, 12 patient tracking, 24 patient interviews (12 for each NMS and current practice arm), 24 pharmacist interviews and eight GP interviews. Interviews will incorporate a range of patient and professional views and will be digitally recorded and transcribed. All accompanying field notes will be retained.</p></sec><sec><title>Adverse events</title><p>The nature of the study presents a low risk to participants. In the unlikely event of a suspected adverse event, these will be reported to the study principal investigator and appropriate action taken on a case-by-case basis. All adverse events will be reported to the study sponsor.</p></sec><sec><title>Sample size</title><p>The estimated change in prevalence of non-adherence behavior (primary outcome) is expected to fall from 20% to 10%. A sample size of 250 patients/arm is sufficient to detect this change with 80% power, 5% significant level (two-tailed) at a projected 20% drop-out rate. Therefore, the total number of patients to be recruited for the RCT is 500 patients.</p><p>Data obtained a few months post-NMS implementation indicated that on average, two NMS consultations were initiated per pharmacy per week. It was initially estimated that on average there will be 52 eligible patients per pharmacy in six months. Based on the assumption that 50% of eligible NMS patients consent to be part of the study, there will be 26 in each pharmacy over the required six months.</p></sec><sec><title>Revision of clinical site recruitment strategy</title><p>To deliver the sample as quickly as possible, patients were to be recruited from 20 pharmacies which, assuming the above drop-out rate, would deliver 520 patients within six months. However, of 24 pharmacies initially recruited, at least ten have not delivered patients to the study due to lack of both NMS service uptake and patients declining to be in the study. Therefore, we are currently replacing these pharmacies with ‘NMS-active’ pharmacies. The pharmacy profiling has provided insights into the reasons that affect service uptake and facilitators to recruitment to the study. This has enabled us to develop a site suitability survey to better identify more appropriate and potentially more successful pharmacy sites for recruitment (see Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Table S1).</p></sec><sec><title>Compliance</title><p>We recognize that it can be a challenge to encourage community pharmacies and patients using community pharmacies to engage in interventions. We believe that the risks of non-compliance will be minimized by providing pharmacies with clear information on what the study involves, providing access to members of the research team to answer queries and address problems experienced by the pharmacies.</p></sec><sec><title>Likely rate of loss to follow-up</title><p>In the original work, 20% patients withdrew from the study due to a range of reasons [<xref ref-type="bibr" rid="B51">51</xref>]. The main reason was that the patients in the intervention arm were referred back to the GP by the pharmacist due to side effects, lack of effect or patient non-adherence. In this study, these events are being recorded and will be combined into a composite outcome. Loss to follow-up for other reasons is expected to be less than 10%.</p></sec><sec><title>Withdrawal of patients from the study</title><p>A patient will be considered to have withdrawn from the study if the study team receives any notification that the patient wishes to withdraw. This notification might come from the patient themselves, the patient’s representative or from the patient’s pharmacist. We have accounted for withdrawal in our sample size calculations to minimize the effect on the analysis.</p><p>As the NMS is a nationally available service which all patients have the right to receive, some patients allocated to the current practice study arm may decide post-randomization that they still wish to receive the NMS. Should this occur this will be counted as a withdrawal at randomization. If a patient ‘loops’ through the NMS due to referral back to their GP or addition/change of medication (which may mean they received a second new medicine so will start at the beginning of the NMS for this medicine), this patient will continue in the RCT/study as normal. The interaction with the GP and/or change in medication will be noted by the researchers.</p></sec><sec><title>Patient/public involvement</title><p>AC is co-investigator on the project and is an equal member of the team. By attending monthly meetings in person, he has been able to challenge, input and advise the project at every step, bringing his experience and expertise of both the world outside academia and his time living with a long-term condition, to the work. AC was also involved in delivering training to the NMS study pharmacists and in organizing the stakeholder day. We have had further input from a group of patient representatives in development of:</p><p>•the NMS evaluation study application submitted to the DOH</p><p>•the content of materials used for patient recruitment (mainly information sheets and consent forms)</p><p>•data collection forms used during the follow-up telephone calls</p><p>•the diary for collecting health care resource use data</p><p>•the presentation used for the NMS study pharmacist training day</p><p>•training of the NMS study pharmacist</p><p>•frequently asked questions for the study website</p><p>We have a website that provides information to interested parties. (<ext-link ext-link-type="uri" xlink:href="http://www.nottingham.ac.uk/">http://www.nottingham.ac.uk/</ext-link>).</p></sec><sec><title>Statistical analysis</title><sec><title>Descriptive analyses</title><p>Continuous data will be explored using means and standard deviations (SD) if approximately normally distributed and medians and inter-quartile ranges (IQR) if non-normally distributed. Categorical data will be described using frequencies and percentages.</p></sec><sec><title>Comparing baseline characteristics between treatment arms</title><p>The following characteristics will be described by treatment arm: patient age, gender geographic location, disease group and diagnosed person years; (Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Table S2).</p></sec><sec><title>Comparisons between treatment arms</title><p>An intention-to-treat analysis will be used such that patients will be analyzed in the arms they were allocated to, regardless of whether they received the intervention or not [<xref ref-type="bibr" rid="B61">61</xref>,<xref ref-type="bibr" rid="B62">62</xref>].</p><p>Data analysis will be conducted by the research team and by the study statistician (RM). To compare adherence rates, differences in categorical variables will be analyzed using the chi-squared test or Fisher’s exact test as appropriate. The relationship between non-adherence and treatment group will be investigated using logistic regression models to adjust for interaction with chronic disease and other potential confounders. The significance of the variables in the model will be assessed using the Wald chi-squared test and determination of odds ratios (ORs) with associated 95% confidence intervals (CIs). Goodness of fit to the model will be assessed using the Hosmer and Lemeshow test.</p><p>Data collected from the study will be compared with anonymized data from available NMS national data sets. This will facilitate generalizability of the findings.</p><p>Statistical significance will be assessed at the 5% (two-sided) level. All statistical analyses will be conducted using SPSS16 (SPSS IBM United Kingdom Limited, PO Box 41, North Harbour, Portsmouth, Hampshire, PO6 3 AU)and STATA 11.0 (STATA: Timberlake Consultants Limited B3 Broomsleigh Business Park Worsley Bridge Road, London SE26 5BN United Kingdom) [<xref ref-type="bibr" rid="B63">63</xref>].</p></sec></sec><sec><title>Primary outcome</title><p>The primary outcomes will report on the impact of the intervention on adherence to prescribed medicines.</p><p>If any patients are lost to follow-up, a sensitivity analysis will be undertaken [<xref ref-type="bibr" rid="B61">61</xref>,<xref ref-type="bibr" rid="B62">62</xref>].</p></sec><sec><title>Secondary outcome measures</title><p>We acknowledge the potential for type 1 errors associated with significance testing for multiple end points. We will therefore consider our analyses of secondary outcome measures to be partly exploratory in nature, and partly confirmatory of our findings for the primary outcome measures.</p><p>•process measures: parameters from NMS audit data set</p><p>•cognitive and behavioral outcome (self-reported adherence (NMS question and MMAS-8))</p><p>•number of medicines-related problems reported by patients, and resolved</p><p>•adverse events and NHS contacts associated with adverse events</p><p>•health status (EQ-5D)</p><p>•resource use and costs</p><p>•length of initial and subsequent NMS consultations</p><p>•NHS contact (primary and secondary care)</p><p>•non-medical costs</p></sec><sec><title>Within trial analysis of costs</title><p>Costs calculated in the trial analysis will be the cost of intervention for each patient enrolled in the trial for both treatment arms. These data will be presented separately for the two treatment arms. Comparisons between treatment arms at patient level will be made using a two-sample <italic>t</italic>-test on the original dataset, or on a bootstrapped dataset, depending on the normality of the distribution of costs [<xref ref-type="bibr" rid="B64">64</xref>].</p></sec><sec><title>Sub-group analyses</title><p>Sub-group analyses [<xref ref-type="bibr" rid="B65">65</xref>] will only be undertaken for primary outcome measures. Analyses will be undertaken to assess whether the effect of the intervention varies by disease type, age, gender, pharmacy type, pharmacy location, time since diagnosis, number of other medicines prescribed, and deprivation index. Treatment arm and the (continuous) covariate of interest will be added into the regression model [<xref ref-type="bibr" rid="B65">65</xref>]. Where there is evidence of non-linearity, this will be investigated and appropriate transformations will be performed. Significance will be assessed based on likelihood ratio tests with a <italic>P</italic> value of < 0.05 taken as significant and <italic>P</italic> values between 0.05 and 0.1 described as there being ‘some evidence’ for an interaction.</p></sec><sec><title>Missing data</title><p>A complete case analysis will be undertaken with a range of approaches for undertaking sensitivity analyses to assess the robustness of the findings with respect to missing data.</p></sec><sec><title>Economic analysis</title><p>A standard approach to economic analysis will be applied [<xref ref-type="bibr" rid="B66">66</xref>]. We propose to undertake the economic analyses from the perspective of the NHS in terms of the direct costs of providing an intervention to improve medicines adherence in primary care, and the costs of managing diseases for adherent and non-adherent patients.</p></sec><sec><title>Comparators and key parameters under investigation</title><p>The evaluation will compare the NMS intervention with current practice. The main outcome of the economic analysis is cost per QALY gained. We will examine the differences in overall NHS costs and in QALYs gained between NMS intervention and current practice patient groups. Additionally, trial-based economic analysis will be conducted and cost per extra adherent patient will be estimated.</p></sec><sec><title>Sample size for the economic analysis</title><p>The study cannot be powered to detect differences in costs because there is no prior study upon which to base a power calculation.</p></sec><sec><title>Time horizon (follow-up period)</title><p>Adherence rates in both groups will be followed up for six months following the completion of the intervention. The Markov models will follow up patients for long-term horizon (to capture all the relevant cost and outcome consequences for each disease group). Life time horizon will be considered.</p></sec><sec><title>Modeling the effect of non-adherence in each disease group</title><p>Each disease-specific Markov model will be populated with transition probability, cost and health status data to generate the outcomes and costs in a cohort of adherent patients and in a cohort of non-adherent patients.</p></sec><sec><title>Estimating the cost effectiveness of the NMS intervention</title><sec><title>Trial-based cost-effectiveness analysis (short-term economic evaluation)</title><p>Trial-based economic evaluation will be conducted. The adherence measures, intervention costs, and total costs incurred during the trial horizon will be generated at patient level. Then, the adherence rates in the NMS and current practice arms and cost per extra adherent patient will be calculated. Both deterministic and probabilistic incremental economic analyses will be carried out using the adjusted cost and outcome data, in combination with the NMS intervention costs.</p><p>The incremental cost per extra adherent patient generated by the NMS intervention over current practice will be calculated using the following equation:</p><p><disp-formula><mml:math id="M1" name="1745-6215-14-411-i1" overflow="scroll"><mml:mtable columnalign="left"><mml:mtr><mml:mtd><mml:mi>ICER</mml:mi><mml:mo>=</mml:mo><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi>Cost</mml:mi><mml:mi>NMS</mml:mi></mml:msub><mml:mo>–</mml:mo><mml:msub><mml:mi>Cost</mml:mi><mml:mrow><mml:mi>Current</mml:mi><mml:mspace width="0.12em"/><mml:mi>practice</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mfenced><mml:mo stretchy="true">/</mml:mo></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mspace width="4em"/><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi mathvariant="normal">r</mml:mi><mml:mi>NMS</mml:mi></mml:msub><mml:mo>–</mml:mo><mml:msub><mml:mi mathvariant="normal">r</mml:mi><mml:mrow><mml:mi>Current</mml:mi><mml:mspace width="0.12em"/><mml:mi>practice</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mfenced><mml:mo>,</mml:mo></mml:mtd></mml:mtr></mml:mtable></mml:math></disp-formula></p><p>where r<sub>NMS</sub>, r<sub>Current practice</sub> is the proportion of adherent patients in the NMS arm, and current practice arm, respectively. Statistical analysis is not appropriate to test the robustness of ICER. It is not possible to generate 95% confidence intervals around ICERs because the ratio of two distributions does not necessarily have a finite mean, or therefore, a finite variance [<xref ref-type="bibr" rid="B67">67</xref>]. Therefore, generation of a bootstrap estimate of the ICER sampling distribution to identify the magnitude of uncertainty around the ICERs is required. Bootstrapping with replacement will be employed, utilizing Microsoft Excel (Microsoft Campus, Thames Valley Park, Reading, Berkshire, RG6 1WG), using a minimum of 5,000 iterations. These incremental costs and outcomes will be plotted on cost effectiveness plane, uncertainty around ICER will be investigated and cost effectiveness acceptability curves (CEACs) [<xref ref-type="bibr" rid="B68">68</xref>,<xref ref-type="bibr" rid="B69">69</xref>] will be constructed.</p></sec><sec><title>Cost-utility analysis (long-term economic evaluation)</title><p>Combining the NMS trial results with the disease-specific models will allow us to estimate the costs and outcomes associated with the NMS intervention versus current practice.</p><p>The adherence rates for each disease group occurring in the NMS and current practice arms will be combined with the disease-specific Markov models (see Figure <xref ref-type="fig" rid="F3">3</xref>). The incremental costs and outcomes associated with each disease group will be estimated based on trial combined with economic model. This will allow us to generate the incremental effect of the NMS intervention on the costs and outcomes for each disease group, and overall (for the population for which NMS service is targeted).</p><p>Incremental economic analyses will be carried out using the adjusted cost and outcome data (observed in the trial) in combination with the NMS intervention costs, and long-term costs and health effects (estimated using Markov models). This will generate the estimates of the overall costs and health outcomes, measured by QALY gained, for the NMS and the current practice arms.</p><p>The incremental cost per extra QALY generated by the NMS intervention over current practice will be calculated using the following equation:</p><p><disp-formula><mml:math id="M2" name="1745-6215-14-411-i2" overflow="scroll"><mml:mtable columnalign="left"><mml:mtr><mml:mtd><mml:mi>ICER</mml:mi><mml:mo>=</mml:mo><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi>Cost</mml:mi><mml:mi>NMS</mml:mi></mml:msub><mml:mo>–</mml:mo><mml:msub><mml:mi>Cost</mml:mi><mml:mrow><mml:mi>Current</mml:mi><mml:mspace width="0.12em"/><mml:mi>practice</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mfenced><mml:mo stretchy="true">/</mml:mo></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mspace width="4em"/><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi>QALY</mml:mi><mml:mi>NMS</mml:mi></mml:msub><mml:mo>–</mml:mo><mml:msub><mml:mi>QALY</mml:mi><mml:mrow><mml:mi>Current</mml:mi><mml:mspace width="0.12em"/><mml:mi>practice</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mfenced><mml:mo>.</mml:mo></mml:mtd></mml:mtr></mml:mtable></mml:math></disp-formula></p></sec><sec><title>Sensitivity and scenario analysis</title><p>Sensitivity analysis is required to assess the level of uncertainty in the data collected within the trial and subsequent internal robustness of the results.</p><p>Several deterministic one-way sensitivity analyses will be conducted (testing the impact of key uncertain parameters on the cost-utility result). Scenario analyses will be proposed and discussed, alternative key assumptions will be tested.</p><p>Probabilistic sensitivity analysis (PSA) will be conducted for base-case scenario, main alternative scenarios (and one-way sensitivity analyses).</p><p>Monte Carlo simulation will be applied for sampling incremental costs and QALY, using Tree Age Pro (TreeAge Software, Inc., 888-TreeAge -or- +1 413-458-0104, One Bank Street Williamstown, MA, 01267 USA) (at least 1,000 samples). Uncertainty around input parameters will be modeled in a standard way: appropriate probability distributions will be assumed for cost, utility, probabilities and ratios [<xref ref-type="bibr" rid="B70">70</xref>].</p><p>Cost effectiveness acceptability curves (CEACs) [<xref ref-type="bibr" rid="B68">68</xref>] will be constructed to express the probability that the cost per QALY gained (y-axis) is cost effective as a function of the decision-maker’s ceiling cost effectiveness ratio (λ) (x-axis) for base-case, sensitivity and scenario analyses [<xref ref-type="bibr" rid="B69">69</xref>].</p><p>The incremental net monetary benefit (INB) will be estimated from the incremental costs and QALYS for NMS compared with current practice using the formula:</p><p><disp-formula><mml:math id="M3" name="1745-6215-14-411-i3" overflow="scroll"><mml:mtable columnalign="left"><mml:mtr><mml:mtd><mml:mi>INB</mml:mi><mml:mfenced open="(" close=")"><mml:mi>λ</mml:mi></mml:mfenced><mml:mo>=</mml:mo><mml:mi>λ</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi>QALY</mml:mi><mml:mi>NMS</mml:mi></mml:msub><mml:mo>–</mml:mo><mml:msub><mml:mi>QALY</mml:mi><mml:mrow><mml:mi>Current</mml:mi><mml:mspace width="0.12em"/><mml:mi>practice</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mfenced></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mspace width="4em"/><mml:mo>−</mml:mo><mml:mfenced open="(" close=")"><mml:mrow><mml:msub><mml:mi>Cost</mml:mi><mml:mi>NMS</mml:mi></mml:msub><mml:mo>–</mml:mo><mml:msub><mml:mi>Cost</mml:mi><mml:mrow><mml:mi>Current</mml:mi><mml:mspace width="0.12em"/><mml:mi>practice</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:mfenced></mml:mtd></mml:mtr></mml:mtable></mml:math></disp-formula></p><p>The incremental net benefit approach will be used due to well-known problems associated with incremental cost effectiveness ratios (ICERs) when bootstrap replicates cover all four quadrants of the cost effectiveness plane [<xref ref-type="bibr" rid="B71">71</xref>,<xref ref-type="bibr" rid="B72">72</xref>]. Incremental net monetary benefit will be calculated for a threshold range from £0 to £160,000 using increments of £10,000.</p></sec></sec><sec><title>Format of tables for publishing the main trial results and within trial economic analysis</title><p>The format of tables for publishing the main trial results and economic analysis is shown in Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Table S2.</p></sec><sec><title>Qualitative study</title><p>Qualitative data analysis will start during the early stages of data collection and proceed iteratively so that emergent findings are incorporated into subsequent data collection, including the revision of data collection methods, such as interview topic guides. All pharmacy profile observations will be recorded in field notes and subsequently typed up. All interviews will be transcribed verbatim. The data will be then imported into qualitative analysis package NVivo; QSR International Pty Ltd. Version 9, 2010 (QSR International (UK) Limited, Warrington, Cheshire, WA2 7LT, United Kingdom) for the purpose of coding and thematic analysis. This will involve initial reading and re-reading of the transcribed data by multiple members of the research team to identify common codes and categories. These codes will be compared for their internal consistency and boundaries. A coding framework will be constructed iteratively and two members of the research team will systematically code data according to this framework. Coded extracts will be compared and differences of opinion discussed until agreement is reached. Actively searching for disconfirming data will be undertaken as well as regular detailed discussions amongst the qualitative researchers. To enhance the consistency of analysis, review meetings will be held with a third researcher (JW) who will oversee the process and negotiate consensus on the final thematic codes assigned to each response.</p><p>Consideration will then be given to how these issues group together in broader themes related to the research objectives. The principle of constant comparison will be used to test and refine the empirical conceptual consistency of codes and themes which were synthesized and narrated using a technique similar to that used by Ziebland and McPherson (2006) [<xref ref-type="bibr" rid="B73">73</xref>].</p></sec><sec><title>Trial organization</title><p>Professor Elliott and Dr Boyd will have overall responsibility for the day-to-day management of the trial and Professor Waring will have overall responsibility for the qualitative workstream. Professor Elliott will have responsibility for the economic analysis. Mr Mehta is the trial statistician.</p><p>A Project Management Group will be meeting monthly throughout the study to help ensure that all trial activities are organized according to the protocol and within the timescales set out in the original application for funding, will monitor and supervise the trial and comment on any proposed amendments to the protocol.</p><p>The NMS Evaluation Advisory Group (NEAG) is headed by Professor Nick Mays. The trial statistician will report to the NEAG, which will be responsible for reviewing the data from the trial. The NEAG has agreed to operate within the framework suggested in the <italic>MRC Guidelines for Good Clinical Practice in Clinical Trials</italic>[<xref ref-type="bibr" rid="B74">74</xref>].</p></sec><sec><title>Ethical aspects</title><p>The clinical trial will be conducted according to the Helsinki Declaration [<xref ref-type="bibr" rid="B70">70</xref>], the GCP Guidelines [<xref ref-type="bibr" rid="B75">75</xref>] and NHS Research Governance requirements. Patients agreeing to participate in the study have provided written informed consent in a form designed for such purpose. The patient may refuse to continue participating in the study at any time after providing his/her consent. The information generated by the study will be confidential and limited to the purposes stipulated in the protocol.</p><p>The study was given a favorable opinion on 2 May 2012 by the National Research Ethics Service (NRES) Black Country Research Ethics Committee (12/WM/0096).</p><p>All staff involved in data collection will have approval from the appropriate local NHS research and development offices.</p><p>The study was registered with the ClinicalTrials.gov trials database on 19 June 2012. Trial reference number <ext-link ext-link-type="uri" xlink:href="http://www.clinicaltrials.gov/NCT01635361">NCT01635361</ext-link> (<ext-link ext-link-type="uri" xlink:href="http://clinicaltrials.gov/ct2/show/NCT01635361">http://clinicaltrials.gov/ct2/show/NCT01635361</ext-link>).</p><p>The study was registered with the Current Controlled trials database on 5 July 2012. Trial reference number ISRCTN 23560818 (<ext-link ext-link-type="uri" xlink:href="http://www.controlled-trials.com/ISRCTN23560818/">http://www.controlled-trials.com/ISRCTN23560818/</ext-link>; DOI 10.1186/ISRCTN23560818).</p><p>This study is registered with the UK Clinical Research Network (UKCRN) study 12494 (<ext-link ext-link-type="uri" xlink:href="http://public.ukcrn.org.uk/Search/StudyDetail.aspx?StudyID=12494">http://public.ukcrn.org.uk/Search/StudyDetail.aspx?StudyID=12494</ext-link>).</p></sec><sec><title>Study timeline</title><p>Trial start: January 2012</p><p>Start of baseline data collection and interventions in pharmacies: August 2012</p><p>End of interventions in pharmacies: September 2013</p><p>End of six month follow-up data collection: March 2014 (one month after official funding ends)</p><p>Start of data analysis: September 2013</p><p>Planned study end date: February 2014</p><p>Duration: 26 months</p></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>As the NMS intervention is an advanced service being delivered by appropriately qualified pharmacists, we did not expect non-compliance with the intervention to be a large problem. However, some pharmacies were only offering the service to a limited number of patients, which has affected our patient recruitment rate and has required further site recruitment. We have also revised recruitment targets for individual pharmacies and developed a site suitability survey. On-going support provided to sites includes regular visits to sites have been made by members of the research team. Sites have also been kept up-to-date with recruitment progress through newsletters.</p><p>One strength of this study is the substantial qualitative workstream, supplemented by engagement and implementation activity. Many studies of pharmacy interventions have not included qualitative work. Qualitative work will enable us to learn about how this service is actually being implemented in practice and will enable us to explain the quantitative results we obtain. We will be carrying out stakeholder days where we will be inviting patients, service providers and commissioners to attend, to obtain views on the NMS in particular, and on managing medicines for long-term conditions in general.</p></sec><sec><title>Trial status</title><p>At the time of submission of this article, 58 pharmacies have been recruited into the study. Seventy pharmacists have been trained and 502 patients have been recruited (4 October 2013). At the time of submitting this protocol, analysis of quantitative data had not been undertaken.</p></sec><sec><title>Abbreviations</title><p>ACE: Angiotensin converting enzyme (inhibitor); CEACs: Cost effectiveness acceptability curves; CHD: Coronary heart disease; COPD: Chronic obstructive pulmonary disease; EMSY: East Midlands and South Yorkshire; EQ-5D-3 L: EuroQol 5 Dimension-3 level instrument; GL: Greater London; GCP: Good Clinical Practice; GP: General practitioner (or family practitioner); ICER: Incremental cost effectiveness ratios; INR: International normalized ratio; IT: Information technology; MMAS-8: Morisky’s medication adherence scale 8-item version; MRC: Medical Research Council; MUR: Medicine use review; NHS: The National Health Service in England; NMS: New medicines service; PCRN: Primary care research network; PCT: Primary care trust; PSSRU: Personal Social Services Research Unit; QALY: Quality-adjusted life-year; RCT: Randomized controlled trial.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>RAE, who has made substantial contributions to the conception and design of the study, is co-responsible with MB for the overall administration and direction of the project, the analysis and interpretation of data and will give the final approval of the version to be published. MB, JW, NB, AC, RM and AJA are also co-responsible for the overall design, administration and direction of the study. NS, AL and JD are responsible for the day-to-day management of the trial and qualitative studies LT for economic modeling. All authors participated in the design of the project. RM and RAE have had a major role in designing the statistical analysis for the trial. RE and LT have designed the economic analysis; JW, AL, JD led on the design of the qualitative analysis. AC is the patient and public involvement lead, providing PPI perspective on all aspects of the study design, conduct and analysis and provides contacts for further patient input. All authors read and approved the final manuscript.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional file 1</title><p><bold>File name: Elliott Additional file 1.</bold> File format: MS Word (.doc). Title of data: site suitability survey and outline of results tables. Description of data: two tables.</p></caption><media xlink:href="1745-6215-14-411-S1.doc"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
Effectiveness and cost effectiveness of guided online treatment for patients with major depressive disorder on a waiting list for psychotherapy: study protocol of a randomized controlled trial	<sec><title>Background</title><p>Depressive disorders are highly prevalent and result in negative consequences for both patients and society. It is therefore important that these disorders are treated adequately. However, due to increased demand for mental healthcare and subsequent increased costs, it would be desirable to reduce costs associated with major depressive disorder while maintaining or improving the quality of care within the healthcare system. Introducing evidence-based online self-help interventions in mental healthcare might be the way to maintain clinical effects while minimizing costs by reducing the number of face-to-face sessions. This study aims to evaluate the clinical and economical effects of a guided online self-help intervention when offered to patients with major depressive disorder on a waiting list for psychotherapy in specialized mental health centers (MHCs).</p></sec><sec><title>Methods</title><p>Patients at mental health centers identified with a <italic>Diagnostic and Statistical Manual of Mental Disorders</italic>, fourth edition (DSM-IV) diagnosis of major depression who are awaiting face-to-face treatment are studied in a randomized controlled trial. During this waiting list period, patients are randomized and either (1) receive an internet-based guided self-help treatment or (2) receive a self-help book. The 5-week internet-based guided self-help intervention and the self-help booklet are based on problem solving treatment. After the intervention, patients are allowed to start regular face-to-face treatment at MHCs. Costs and effects are measured at baseline, after the intervention at 6 to 8 weeks, 6 months and 12 months. The primary outcome measure is symptoms of depression. Secondary outcome measures are diagnosis of depression, number of face-to-face sessions, absence of work and healthcare uptake in general. Additional outcome measures are anxiety, insomnia, quality of life and mastery.</p></sec><sec><title>Discussion</title><p>This study evaluates the effectiveness and cost effectiveness of internet-based guided self-help in patients at specialized mental health centers. The aim is to demonstrate whether the introduction of internet-based self-help interventions in regular mental healthcare for depressive disorders can maintain clinical effects and reduce costs. Strengths and limitations of this study are discussed.</p></sec><sec><title>Trial registration</title><p>Netherlands Trial Register <ext-link ext-link-type="uri" xlink:href="http://www.trialregister.nl/trialreg/admin/rctview.asp?TC%20=%202824">NTR2824</ext-link></p></sec>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Kenter</surname><given-names>Robin Maria Francisca</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>robin.kenter@vu.nl</email></contrib><contrib contrib-type="author" id="A2"><name><surname>van Straten</surname><given-names>Annemieke</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>a.van.straten@vu.nl</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Hobbel</surname><given-names>Sabine Heleen</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>s.h.hobbel@vu.nl</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Smit</surname><given-names>Filip</given-names></name><xref ref-type="aff" rid="I3">3</xref><xref ref-type="aff" rid="I4">4</xref><email>fsmit@trimbos.nl</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Bosmans</surname><given-names>Judith</given-names></name><xref ref-type="aff" rid="I5">5</xref><xref ref-type="aff" rid="I6">6</xref><email>j.e.bosmans@vu.nl</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Beekman</surname><given-names>Aartjan</given-names></name><xref ref-type="aff" rid="I6">6</xref><xref ref-type="aff" rid="I7">7</xref><email>a.beekman@ggzingeest.nl</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Cuijpers</surname><given-names>Pim</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>p.cuijpers@vu.nl</email></contrib>	Trials	<sec><title>Background</title><p>Depressive disorders are highly prevalent [<xref ref-type="bibr" rid="B1">1</xref>-<xref ref-type="bibr" rid="B3">3</xref>], and are associated with high costs in the professional, social, personal and financial realms [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B4">4</xref>]. Those suffering from depression are more often absent from work, and have higher levels of healthcare consumption compared to healthy individuals [<xref ref-type="bibr" rid="B2">2</xref>]. It is desirable from a clinical point of view to reduce the burden of depression, as it is from a societal point of view to reduce the economic costs related to increased healthcare uptake, and reduced work productivity. In addition, given the increased demand for mental healthcare and limitation of recourses, it is important for mental health centers to optimize the efficient and effective use of resources.</p><p>In The Netherlands, a person with symptoms of depression who is seeking help is most likely to be seen first by a general practitioner (GP) [<xref ref-type="bibr" rid="B5">5</xref>], who acts as a gatekeeper for referral to specialized mental health services. After registration at a mental health service, patients usually have an assessment and are then assigned to a specific treatment. The time between registration and the first treatment session is normally at least 6 weeks. Long waiting lists caused by low workforce numbers are common. This time might be used to deploy internet-based self-help treatments [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref>], as previous studies have indicated that internet-based self-help therapies are clinically effective in diverse populations [<xref ref-type="bibr" rid="B8">8</xref>-<xref ref-type="bibr" rid="B11">11</xref>]. Internet-based treatments require less therapist time in comparison with standard face to face treatments and can therefore offer potential solutions as they are immediately accessible, less costly and put less strain on therapeutic resources. They might serve first step in a stepped delivery of care. This means that only those who do not respond adequately step up to a treatment of higher intensity, in this case the regular treatment in mental health centers. This stepped-care approach is suggested in several guidelines such as the Australian and National Institute for Health and Care Excellence (NICE) guidelines for depression, which recommend that patients receive the least burdensome treatment. Psychological treatments such as computerized cognitive behavioral therapy (cCBT) and individual guided self-help programs are recommended as low intensity treatments. After receiving such treatments, patients possibly need fewer or no face-to-face sessions.</p><p>With the current cuts in healthcare budgets, it would be desirable to reduce costs associated with major depressive disorder while maintaining or improving the quality of care. Introducing evidence-based internet-based self-help interventions in mental healthcare might be the way to speed up clinical recovery while minimizing costs by reducing the number of face-to-face sessions. However, internet-based guided self-help interventions are currently not offered as a first step to those waiting for specialized mental healthcare.</p><p>In the current study, we offer an internet-based guided self-help program to patients on a waitlist for psychological treatment. The self-help program is based on problem-solving therapy and uses self-examination therapy as a general framework [<xref ref-type="bibr" rid="B12">12</xref>]. This method has been found to be effective in several studies in the US [<xref ref-type="bibr" rid="B13">13</xref>], and for this study we used a guided self-help program that has been examined in two earlier trials, and has proved its effectiveness in reducing depressive symptoms in community samples [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref>].</p><sec><title>Aims and hypotheses</title><p>This study aims to establish the effectiveness of an internet-based guided self-help intervention for patients with major depressive disorder before face-to-face treatment in specialized mental healthcare in comparison to a control group who are on a waitlist for face-to-face treatment. Furthermore, we expect that offering internet-based treatment as a first step of care reduces costs as we predict the internet group to take up fewer regular treatment sessions.</p></sec></sec><sec sec-type="methods"><title>Methods</title><sec><title>Study design</title><p>The study is a randomized controlled trial with an economic evaluation. We are currently enrolling 248 patients over 2 large mental health centers (MHCs) in 10 different locations. We will recruit patients directly after registration for regular face-to-face mental health services who need to wait for at least 6 weeks before their first treatment session. They will be randomized to either an internet-based guided self-help intervention or the regular waitlist before the first treatment session. People on the waitlist will receive a self-help book without additional guidance in order to motivate them to participate in the randomize, controlled trial (RCT). Both groups are allowed to receive regular face-to-face (FTF) therapy at the MHCs after the waitlist period. Participants in both groups complete assessment at baseline, at intervention or waiting list (WL) completion (6 to 8 weeks), at a 6-month follow-up, and at 12 months. The protocol of this study has been approved by the Medical Ethics Committee of the VU University Medical Center (registration number 2011/223).</p></sec><sec><title>Inclusion and exclusion criteria</title><p>Eligible participants are adults, aged 18 or over, registering for regular treatment at one of the participating MHCs who meet the criteria for a <italic>Diagnostic and Statistical Manual of Mental Disorders</italic>, fourth edition (DSM-IV) diagnosis of major depression as measured with the Clinical International Diagnostic Interview (CIDI) by a trained research assistant, have access to the internet, and adequate proficiency in Dutch. Exclusion criteria are starting or changing type of dosage of antidepressant medication, the presence of a bipolar or psychotic disorder, and an increased risk of suicide. Comorbid disorders other than bipolar or psychotic disorders are allowed.</p></sec><sec><title>Recruitment</title><p>Participants will be recruited while registering at the participating MHCs. In routine care patients are briefly scanned by the MHCs (according to MHCs protocol to screen out patients who are in crisis and need immediate treatment), and for the purpose of this study, patients with mood problems who are eligible for an intake assessment are asked by the MHC to share their contact details with the researchers. Those who are willing to do so, will then be called by a member of the research team who further explains the aim of this study and performs an additional check of the inclusion and exclusion criteria. If eligible, patients receive a study brochure and an informed consent form. Patients will only be included in the study if they meet all the criteria, and sign the informed consent form. In order to confirm the diagnoses Major Depressive Disorder trained interviewers conduct a clinical interview (CIDI).</p></sec><sec><title>Sample size</title><p>The trial is powered to detect an effect size of d ≥0.40 [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B16">16</xref>] as statistically significant in a two-tailed test with α = 0.05 and power of (1 - β =) 0.80 with N = 99 per condition. To compensate for loss of follow-up of 20%, the trial requires starting with 99/0.8 = 124 participants per condition at baseline. The dropout rate of 20% is conservative, but from a power perspective a cautious approach. As this trial has two conditions, a total of N = 248 will be necessary for the complete trial.</p></sec><sec><title>Mental health centers characteristics</title><p>Two MHCs will be participating in this research. They each offer services at a number of different locations. In this trial, a total of ten locations will participate. These centers are chosen for pragmatic purposes, as they have a high number of patient enrolment, and have participated in prior research of the VU University. In general, patients are referred by their general practitioner to the mental health centers where the patients is screened and placed on a waiting list. Within 6 weeks an initial meeting with a therapist takes place in which the patient’s needs and preferences regarding to treatment are determined. At these participating MHCs it normally takes between 7 to 16 weeks for patients to have a first treatment session, depending on therapist workload, treatment modality and other factors. Treatment in both centers can consist of psychological therapies such as CBT, sometimes in combination with medication. The researchers are not involved in the face-to-face treatment, and neither in the decision-making process. However, prior to the start of this study all therapists at the participating MHCs will be informed about the goals of this study and the internet-based intervention, they will attend a Q&A session with the first author and receive the self-help book containing the intervention. Prior to the start of a patients’ face-to-face treatment at the MHC the therapists will be informed by the aforementioned note in electronic patient record and through email that their patients are participating in this study. One of the objectives of the internet-based intervention is that the subsequent face-to-face treatment can be adjusted to fewer sessions. The number and type of therapeutic sessions patients will receive is based on the needs of the patient, the judgment of the therapist, and protocol available at the MHC. The type and number of sessions will therefore vary per participant, and are outside the control of the researchers.</p></sec><sec><title>Internet intervention</title><p>The internet intervention that will be used is called 'Taking Control’ (original title: 'Alles Onder Controle’). This intervention uses the self-examination treatment model developed by Bowman and colleagues [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B13">13</xref>], which is based on problem-solving therapy and uses self-examination therapy (SET) as a general framework. We translated it into Dutch, elaborated on it, and added information and exercises. We built a website for this intervention and developed a system for email support. This intervention has been described in more detail in several other studies [<xref ref-type="bibr" rid="B12">12</xref>-<xref ref-type="bibr" rid="B14">14</xref>].</p><p>In brief, this intervention is short, structured and manualized. It consists of five weekly sessions. Each session contains a structured homework assignment to be completed by the participant. The first session requires participants to consider what is important in their lives. Next to this, participants will make a list of current problems and worries in their lives. This first session helps participants to divide their problems into the following three categories: not important (if the problem is not related to the list of important things), important and unsolvable (for example, permanent loss of health or a loved one), or important and solvable. In the second, third and fourth session participants will be offered various coping skills related to each of the categories of problems; the main focus is placed on adopting a structured six-step approach when encountering important, solvable problems. This structured approach is divided in the following steps: identifying the current problem; finding possible solutions; selecting one solution; create a plan to solve the problem with this solution; execute the plan and evaluate the plan. The last week of the intervention is reserved for both the reflection on long-term goals and the development of a structure to achieve these goals. Participants can only move on to the next session once the exercise in the current session has been submitted and when feedback on this session has been released by the research team.</p><p>The participants are supported by a coach, who gives feedback to the homework assignments of the participants in brief, weekly emails. The total amount of time spent on each patient is about 1.5 h (estimate based on our earlier trials). The writing of these emails takes about 15 or 20 minutes per week, and will be performed by a coach. The coaches will be trained by the psychologists who have developed the intervention and also wrote the protocol for giving feedback to ensure the consistency and integrity. An independent psychologist will verify whether the coaches have followed the protocol sufficiently by reading a random selection of the feedback emails.</p><p>The feedback has two purposes. Firstly, coaches will help participants to become familiar with the presented techniques. The second purpose consists of motivating the participant to continue with the intervention. Feedback will be received by participants within 3 working days after a session has been completed and submitted. When participants pose content related questions to their coaches via the website, email or phone, their coaches will provide additional guidance after receiving the question. After 5 weeks of guidance, patients can continue to use the internet-based treatment but will not receive any feedback on their assignments.</p></sec><sec><title>Control condition</title><p>To increase participation rates in the control group, this group receives an unguided self-help book posted to their home address. This control group will not receive any feedback from coaches, nor will it have the opportunity to pose questions. Earlier research shows that self-help without any form of guidance only results in a small effect on participants with increased levels of depressive symptoms [<xref ref-type="bibr" rid="B16">16</xref>].</p><p>Participants in both conditions will be referred for regular FTF treatment after registration at the MHC. FTF treatment might consist of additional waiting time; the duration of the waiting time is highly variable with a minimum of 7 weeks and a maximum of 16 weeks depending on the location of MHC. Variation fluctuates both between locations and over time within the centers, due to availability of therapists. In case the waiting time for any MHC falls below 8 weeks, new participants from this MHC will be temporarily excluded from participating in the research until the waiting time for new patients at this MHC exceeds 8 weeks again. This has been decided in order to not measure the effects of active treatment by the MHCs at the first post-intervention test at 8 weeks.</p><p>The researchers of this study do not influence the waiting time at the MHCs, nor will participation in this study influence the waiting time for the participants.</p></sec><sec><title>Randomization and blinding</title><p>The random allocation sequence will be generated by an independent researcher in the program 'Random Allocation Software’, stratified by location using blocks of six, eight and ten. After each inclusion another independent researcher will allocate the patient to either the intervention or the control condition. Due to the nature of the intervention the treatment group allocation cannot be concealed from the participants, nor from the research team and assessor of outcomes. Participants are assigned to one of two conditions within 2 working days from their baseline assessment Figure <xref ref-type="fig" rid="F1">1</xref>.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p>Study flow chart.</p></caption><graphic xlink:href="1745-6215-14-412-1"/></fig></sec><sec><title>Assessments</title><p>This study will utilize both clinical and economical outcome measures (see Table <xref ref-type="table" rid="T1">1</xref>). Assessments take place at baseline (before randomization), follow-up assessments are at post-intervention (6 to 8 weeks), and at 6 and 12 months after baseline.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Outcome measures</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Questionnaires</bold></th><th align="left"><bold>Aim</bold></th><th align="left"><bold>T0 Baseline (pre-test)</bold></th><th align="left"><bold>T1 Post-test (8 weeks)</bold></th><th align="left"><bold>T2 Follow-up I (6 months)</bold></th><th align="left"><bold>T3 Follow-up II (12 months)</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Primary outcome<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">CES-D<hr/></td><td align="left" valign="bottom">Symptoms of depression<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Secondary outcomes<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">CIDI (Section E)<hr/></td><td align="left" valign="bottom">Diagnosis of depression<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">HADS (Anxiety section)<hr/></td><td align="left" valign="bottom">Symptoms of anxiety<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">ISI<hr/></td><td align="left" valign="bottom">Level of insomnia<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">EQ-5D<hr/></td><td align="left" valign="bottom">Quality of life<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Resource use<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">F2F<hr/></td><td align="left" valign="bottom">Number of face-to-face appointments in MHC<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">TiC-P<hr/></td><td align="left" valign="bottom">General health service uptake and productivity losses<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Mediator<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Mastery scale<hr/></td><td align="left" valign="bottom">Mastery<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Moderators<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Demographic factors<hr/></td><td align="left" valign="bottom">User characteristics<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">NEO-FFI (sections N + C)<hr/></td><td align="left" valign="bottom">Neuroticism and conscientiousness traits<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">CAGE<hr/></td><td align="left" valign="bottom">Alcohol consumption<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">CEQ<hr/></td><td align="left" valign="bottom">Expectancy and treatment<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">AOCEQ<hr/></td><td align="left" valign="bottom">Expectations of the internet intervention<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">CSQ-8<hr/></td><td align="left" valign="bottom">Client satisfaction with treatment<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left">AOCSQ</td><td align="left">Satisfaction with type of intervention</td><td align="left"> </td><td align="left">X</td><td align="left"> </td><td align="left"> </td></tr></tbody></table><table-wrap-foot><p><italic>AOCEQ</italic> 'Alles Onder Controle’ Expectation Questionnaire, <italic>AOCSQ</italic> 'Alles Onder Controle’ Satisfaction Questionnaire, <italic>CAGE</italic> 'Cutting down, Annoyance by criticism, Guilty feeling, and Eye-openers’, <italic>CEQ</italic> Credibility/Expectancy Questionnaire, <italic>CES-D</italic> Center for Epidemiological Studies Depression scale, <italic>CIDI</italic> Composite International Diagnostic Interview, <italic>CSQ-8</italic> Client Satisfactory Questionnaire-8, <italic>EQ-5D</italic> EuroQol 5-Dimensions, <italic>F2F</italic> face-to-face, <italic>ISI</italic> Insomnia Severity Index, <italic>HADS</italic> Hospital Anxiety Depression Scale, <italic>MHC</italic> mental health center, <italic>NEO-FFI</italic> NEO-Five Factor Inventory, <italic>TiC-P</italic> Trimbos and iMTA Questionnaire on Costs Associated with Psychiatric Illness.</p></table-wrap-foot></table-wrap></sec><sec><title>Outcome measures</title><p>Our primary outcome will be symptoms of depression. Secondary outcomes are DSM-IV diagnosis of depression, number of face-to-face sessions in MHC, costs (in terms of all healthcare use and work absenteeism), anxiety, mastery, insomnia, quality of life, and satisfaction. Plausible moderators measured in this study will be alcohol use, personality, demographic factors and client expectations.</p><p>Because the Composite International Diagnostic Interview (CIDI) is part of the questionnaires at baseline and at post-intervention assessment (Table <xref ref-type="table" rid="T1">1</xref>), the baseline questionnaires and the 6 to 8 week assessment will be administered by phone. Follow-up assessments at 6 and 12 months will be administered online. All questionnaires will be in Dutch.</p></sec><sec><title>Primary outcome</title><sec><title>Symptoms of depression</title><p>Symptoms of depression will be registered by using the Center for Epidemiological Studies Depression scale (CES-D) [<xref ref-type="bibr" rid="B17">17</xref>]. This scale consists of 20 items, the total score ranges between 0 and 60; higher scores indicate higher levels of depression and a score of 16 and above indicates a clinical level of depression. This questionnaire has been tested in various populations and has been found valid and reliable [<xref ref-type="bibr" rid="B18">18</xref>]. A Dutch version of the CES-D has been validated for internet administration [<xref ref-type="bibr" rid="B19">19</xref>].</p></sec></sec><sec><title>Secondary outcomes</title><sec><title>Diagnosis of depression</title><p>The CIDI will be used to assess whether the DSM-IV diagnosis of depression has been met. The CIDI has been developed by the World Health Organization to assess DSM-IV Axis-I diagnoses [<xref ref-type="bibr" rid="B20">20</xref>]. For the purposes of this study, only section E will be administered, which allows for screening for depressive disorders. The CIDI will be conducted by phone by a trained interviewer.</p></sec><sec><title>Symptoms of anxiety</title><p>The subscale Anxiety of the Hospital Anxiety and Depression Scale (HADS) will be used to measure symptoms of anxiety. The depression scale will not be utilized in this research because depression will be measured by both CIDI and CES-D. The Anxiety subscale consists of 7 items, which scores range from 0 to 21; higher scores indicate higher levels of anxiety [<xref ref-type="bibr" rid="B21">21</xref>]. The HADS has shown to be reliable in Dutch populations [<xref ref-type="bibr" rid="B21">21</xref>].</p></sec><sec><title>Symptoms of insomnia</title><p>The perceived level of insomnia will be measured by the Insomnia Severity Index (ISI) [<xref ref-type="bibr" rid="B22">22</xref>]. This questionnaire measures both the concerns associated with the perceived level of insomnia, as well as symptoms and consequences of insomnia. Each item is rated on a 0 to 4 scale; a higher score indicates more severe insomnia. ISI has been found to be internally consistent and reliable [<xref ref-type="bibr" rid="B22">22</xref>].</p></sec><sec><title>Quality of life</title><p>The EuroQol 5-Dimensions questionnaire (EQ-5D), which consists of five items, will be used to measure quality of life. It registers the level of perceived problems (no, some or extreme) in five domains (mobility, self-care, usual activities, pain/discomfort and anxiety/depression). A total of 486 distinct health states can be scored which are located between 0, which indicates worst health possible, and 1, which indicates perfect health [<xref ref-type="bibr" rid="B23">23</xref>].</p></sec></sec><sec><title>Resource use</title><sec><title>Costs related to the intervention</title><p>Costs related to the intervention will be calculated based on the costs of running the intervention in both conditions. To calculate costs related to the internet intervention, the costs of running the website platform and the costs of the hours of coaching invested in the intervention will be taken into account. Additional costs in the control group are the publishing of the self-help books.</p></sec><sec><title>Costs related to mental healthcare</title><p>Costs related to mental healthcare will be measured as the reported number and type of sessions by the MHC. This information is sent to the first author 12 months after participants have registered at the MHCs. The direct costs of the sessions will be calculated based on the Dutch standard cost prices [<xref ref-type="bibr" rid="B24">24</xref>].</p></sec><sec><title>Costs from a societal perspective</title><p>Healthcare costs in general as well as productivity losses will be measured with the revised version of the Trimbos and iMTA Questionnaire on Costs Associated with Psychiatric Illness (TiC-P). Direct costs will be measured by investigating which contact with health professionals has occurred and which type of medication has been prescribed. Indirect costs will be measured by work absenteeism and reduced productivity. The baseline TiC-P measures the care consumption 4 weeks prior to the intervention. The follow-up assessments at 6 and 12 months use a 14 and 26 weeks recall period respectively. Previous research has shown that up until half a year later, patients can reliably recall their consumption of health services [<xref ref-type="bibr" rid="B25">25</xref>]. The TiC-P has been used previously in a population with depressive symptoms in The Netherlands [<xref ref-type="bibr" rid="B26">26</xref>].</p></sec></sec><sec><title>Mediator</title><sec><title>Mediator</title><p>The amount of perceived control in a person’s life will be measured by the Pearlin Mastery Scale [<xref ref-type="bibr" rid="B27">27</xref>]. The scale consists of seven distinct items that are rated on a four-point scale. Higher scores indicate more perceived control; scores range from 7 to 35. The scale has good reliability [<xref ref-type="bibr" rid="B27">27</xref>].</p></sec></sec><sec><title>Moderators</title><sec><title>Demographics</title><p>Demographic variables such as age, gender, parental nationality, family composition, family income and educational level will be screened for in a general questionnaire that measures user characteristics.</p></sec><sec><title>Alcohol use</title><p>The use of alcohol will be monitored using the four questions that make up the acronym CAGE: 'Cutting down, Annoyance by criticism, Guilty feeling, and Eye-openers’ [<xref ref-type="bibr" rid="B28">28</xref>]. The CAGE is a widely used concise screening tool for problematic alcohol consumption [<xref ref-type="bibr" rid="B29">29</xref>].</p></sec><sec><title>Personality</title><p>In order to measure the constructs neuroticism and conscientiousness, the NEO-Five Factor Inventory (NEO-FFI) will be administered [<xref ref-type="bibr" rid="B30">30</xref>]. Previous research has indicated that neuroticism often coincides with depression (for example, [<xref ref-type="bibr" rid="B31">31</xref>]). Those scoring higher on conscientiousness are expected to adhere better to the homework exercises and consequently benefit more from internet-based therapies. Only these two domains will be tested for to not exhaust participants more than necessary. A total of 12 questions on each of the domains will be answered on a 5-point Likert scale.</p></sec><sec><title>General expectancy</title><p>The Credibility/Expectancy Questionnaire of Devilly and Borkovec (CEQ) will be used to measure the expected change and credibility of proposed treatment [<xref ref-type="bibr" rid="B32">32</xref>]. It consists of six questions; four questions measure the 'thinking’ aspects about the treatment, two questions measure the 'feeling’ aspects of the questions. One question in both the 'feeling’ and 'thinking’ domain is rated from 0% to 100%; the remaining four questions are rated on a Likert-type scale from 1 to 9. Both factors have been found to have a high internal consistency and to have high test-retest reliability [<xref ref-type="bibr" rid="B32">32</xref>].</p></sec><sec><title>Expectancy with internet intervention</title><p>The 'Alles Onder Controle’ Expectancy Questionnaire (AOCEQ) has been designed for this research to ask what participants’ expectations are about the course. An open question investigates in which ways participants expect to gain benefits from this course. Three five-point Likert scale questions measure to which degree participants appreciate starting with the intervention immediately, that personalized feedback will be given in the internet group and to which degree they expect the course to help them feel less miserable.</p></sec><sec><title>General satisfaction with treatment</title><p>The satisfaction with the internet intervention will be measured by the Client Satisfactory Questionnaire-8 (CSQ-8), which consists of eight questions, each question is scored on a Likert-type scale from 1 to 4 [<xref ref-type="bibr" rid="B33">33</xref>]. The questionnaire addresses several elements that contribute to overall service satisfaction and is reported in a single dimension of overall satisfaction. A high internal consistency has been reported [<xref ref-type="bibr" rid="B33">33</xref>].</p></sec><sec><title>Satisfaction with the internet intervention</title><p>The 'Alles Onder Controle’ Satisfaction Questionnaire (AOCSQ) has been designed specifically to investigate to what degree participants are satisfied with this internet-based intervention. It includes questions on the number of sessions completed and, if applicable, the reasons for not finishing the course. The questionnaire further researches satisfaction with the separate elements of the intervention, such as the quality of the feedback, the clarity of the website and the appropriateness of the examples. Lastly, it explores whether participants were satisfied with the provided alternative to waiting, to which degree internet interventions are preferred over book interventions and to which degree feedback is preferred over non-feedback.</p></sec></sec><sec><title>Statistical analysis</title><p>The analyses will be conducted in agreement with the intention to treat (ITT) principle, as per the CONSORT statement [<xref ref-type="bibr" rid="B34">34</xref>]. Therefore missing endpoints will be imputed using state of the art imputation methods, as a reliable method for handling missing values [<xref ref-type="bibr" rid="B35">35</xref>]. Imputation allows for analyzing all participants in the condition to which they have been randomized, which contributes to guarantee the integrity of the randomization and restores loss of power due to dropout.</p><p>To answer the research questions we will first look at post-test differences between the two groups. We will use analysis of variance (ANOVA), with the baseline values and waiting time as covariates. Subgroup analyses will be performed for different characteristics. The difference in scores between the intervention group and the control group will also be expressed in effect sizes. We use Cohen’s <italic>d</italic> which is calculated by dividing the difference in mean scores of the two groups by their pooled standard deviation (Xexp-Xctrl/SDpooled). Effect sizes under 0.2 are considered to be small, those of 0.5 are moderate and effect sizes of 0.8 are considered to be large [<xref ref-type="bibr" rid="B36">36</xref>]. Furthermore, we will compare the rate of DSM-IV diagnosis of depression in both groups with logistic regression analysis. The clinical effects will also be calculated using reliable change [<xref ref-type="bibr" rid="B37">37</xref>]. The long-term outcomes will be analyzed with longitudinal analyses.</p></sec><sec><title>Economic outcomes</title><p>The economic evaluation will be conducted from a societal perspective, therefore it will include not only the intervention costs and costs stemming from healthcare uptake (direct medical costs), but also the patients' out of pocket costs (direct non-medical costs) and costs stemming from productivity losses due to absenteeism and work cutback days (indirect non-medical costs). Costs will be based on the Dutch standard cost prices [<xref ref-type="bibr" rid="B24">24</xref>] and productivity losses will be valued using the friction costs method, as per the Dutch guideline for economic evaluation. Quality-adjusted life years (QALYs) will be calculated on the basis of the EQ-5D. Having calculated the costs and QALYs allows for a cost-utility analysis, which can be used to compare this intervention’s gains against those of other interventions for depressive disorders. A cost-effectiveness analysis can be carried out by dividing the difference in costs by the difference in effect, as is customary in the field of mental health. Bootstrapping will be used to ascertain the amount of uncertainty surrounding the incremental cost effectiveness ratio (ICER) estimates and graphically depicted on the ICER plane and in the acceptability curve.</p></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>This study will examine the effectiveness of offering an internet-based guided self-help intervention to patients before to start of face-to-face treatment in comparison with patients who have to wait for face-to-face treatment. We will also compare the uptake of regular treatment in both groups. An economic evaluation will determine whether a guided internet intervention followed by face-to-face treatment is economically more sensible compared to waiting for face-to-face treatment. A number of strengths and challenges have been identified by the researchers in this study.</p><sec><title>Strengths</title><p>Existing evidence shows that internet-based treatments are effective in treating depressive disorders in general populations [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B38">38</xref>]. This trial however, will shed new light on the question whether patients with major depressive disorder waiting for specialized care in MHCs can benefit from this type of internet-based guided self-help intervention. This population has, to the best of our knowledge, not previously been studied in this manner, while the relevance of conducting a study on people with major depressive disorder who receive an internet-based intervention as a first step of treatment is high for a number of reasons. As demonstrated previously, depression is a widely prevalent, invasive disorder, which affects the patients, their direct environments, and the society as a whole in multiple ways. Any measures which contribute to more people receiving better care for their depression in specialized mental healthcare against a lower costs will be welcomed by all stakeholders. Stepped care is suggested for treating depression in multiple guidelines, for example the NICE guidelines in the UK (2009) Although there is some supporting evidence, there are few RCTs to demonstrate the (cost) effectiveness of this program in specialized mental healthcare. Furthermore, considering the increase of online interventions to the standard treatment in more and more MHCs, this study will contribute valuable information on the effects of adding online treatment as a first step towards treating major depressive disorder.</p><p>An additional strength of this intervention is the online delivery. The online intervention allows participants not only to access the intervention any convenient hour, with the rise of portable internet there are hardly any limitations related to the location of a participant. In addition, offering guided online treatment during the time otherwise lost to waiting may result in benefits for the patient as well as for the MHCs in terms of patient satisfaction and clinical effectiveness.</p><p>A final strength of this study is the possibility of applicability. If the study indicates that it is economically and clinically beneficial to deliver internet interventions to those with a major depressive disorder as part of their treatment, the intervention could easily be integrated as a standard component in the treatment of depression, as only short training is necessary to become a successful coach for this intervention. This intervention could be applied widely in case the intervention is beneficial for treating people with major depressive disorder.</p></sec><sec><title>Limitations</title><p>One of the possible challenges of this study concerns the general attrition from and non-adherence to internet interventions. Internet treatments require a degree of motivation of the participant. This study includes depressed patients who, by definition, have impaired motivation and lack of energy that may make it more difficult for patients adhere to the intervention. To prevent dropouts and maximize the uptake rates the participants will receive emails and phone calls stating the importance adhering to the intervention.</p><p>Another challenge for the trial could be the prevailing attitude of psychologists at several MHCs who fear redundancy due to internet interventions. However, when MHCs would start treating patients online as a first step towards better health, more patients could be reached as online treatment seems to be less time consuming for health professionals compared to face-to-face treatment.</p><p>A further challenge is the extent the therapists take into account that patients might already have acquired skills and knowledge due to the internet intervention, so that there might be no need to start patients a fixed number of face-to-face sessions as prescribed by the standard treatment protocol. We aim that only those patients who require additional treatment due to a more complex psychological problem will be seen face-to-face. Therefore, if internet interventions are proven to be more clinically effective, it is feasible that psychologists at MHCs will be able to treat more patients in a better manner because of the implementation of internet interventions.</p><p>In summary, existing internet-based guided self-help treatments focus mainly on depression in the general population. This trial, however, is to the best of our knowledge the first effectiveness study of an internet-based guided self-help intervention for major depressive disorder in specialized mental healthcare that also focuses on reduction of face-to-face sessions and costs in general. The findings of this study will contribute to the body of knowledge on the additional value of internet-based treatments for depression. And if the low intensity internet intervention shows to be (cost) effective, it might serve as a first step towards the treatment of major depressive disorder.</p></sec></sec><sec><title>Trial status</title><p>The status of the trial is ongoing recruitment.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>PC, AvS and JB obtained funding for this study. PC, AvS, JB, and FS contributed to the design of the study. PC and AvS created the intervention. RK coordinates the recruitment of patients and data collection. AB, PC and AvS are responsible for the overall supervision. RK and SHH wrote the manuscript. All authors read, contributed towards and approved of the final manuscript.</p></sec>
Comparison between three types of stented pericardial aortic valves (Trivalve trial): study protocol for a randomized controlled trial	<sec><title>Background</title><p>Aortic valve stenosis is one of the most common heart diseases in older patients. Nowadays, surgical aortic valve replacement is the ‘gold standard’ treatment for this pathology and the most implanted prostheses are biological ones. The three most implanted bovine bioprostheses are the Trifecta valve (St. Jude Medical, Minneapolis, MN, USA), the Mitroflow valve (Sorin Group, Saluggia, Italy), and the Carpentier-Edwards Magna Ease valve (Edwards Lifesciences, Irvine, CA, USA). We propose a randomized trial to objectively assess the hemodynamic performances of these bioprostheses.</p></sec><sec><title>Methods and design</title><p>First, we will measure the aortic annulus diameter using CT-scan, echocardiography and by direct sizing in the operating room after native aortic valve resection. The accuracy of information, in terms of size and spatial dimensions of each bioprosthesis provided by manufacturers, will be checked. Their hemodynamic performances will be assessed postoperatively at the seventh day and the sixth month after surgery.</p></sec><sec><title>Discussion</title><p>This prospective controlled randomized trial aims to verify and compare the hemodynamic performances and the sizing of these three bioprostheses. The data obtained may help surgeons to choose the best suitable bioprosthesis according to each patient’s morphological characteristics.</p></sec><sec><title>Trial registration</title><p>ClinicalTrials.gov Identifier: <ext-link ext-link-type="uri" xlink:href="http://clinicaltrials.gov/ct2/show/NCT01522352?term=azarnoush&rank=4">NCT01522352</ext-link></p></sec>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Azarnoush</surname><given-names>Kasra</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>kazarnoush@chu-clermontferrand.fr</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Pereira</surname><given-names>Bruno</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>bpereira@chu-clermontferrand.fr</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Dualé</surname><given-names>Christian</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>cduale@chu-clermontferrand.fr</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Dorigo</surname><given-names>Enrica</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>enrica.dorigo@gmail.com</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Farhat</surname><given-names>Mehdi</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>farhatmehdi@yahoo.fr</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Innorta</surname><given-names>Andrea</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>ainnorta@chu-clermontferrand.fr</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Dauphin</surname><given-names>Nicolas</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>ndauphin@chu-clermontferrand.fr</email></contrib><contrib contrib-type="author" id="A8"><name><surname>Geoffroy</surname><given-names>Etienne</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>egeoffroy@chu-clermontferrand.fr</email></contrib><contrib contrib-type="author" id="A9"><name><surname>Chabrot</surname><given-names>Pascal</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>pchabrot@chu-clermontferrand.fr</email></contrib><contrib contrib-type="author" id="A10"><name><surname>Camilleri</surname><given-names>Lionel</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>lcamilleri@chu-clermontferrand.fr</email></contrib>	Trials	<sec><title>Background</title><p>A critical aspect of aortic valve replacement is to achieve an optimal matching between the patient’s morphology and the implanted valve prosthesis. Specifically, the implanted prosthesis should not impair left ventricle ejection and this is even more crucial in cases where there is a small aortic annulus.</p><p>Pericardial bioprostheses have good hemodynamic performance because of their central opening and the flexibility of their leaflets. We already know that the durability of these pericardial bioprostheses is about 10 to 15 years [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>].</p><p>At present, porcine bioprostheses are less well-performing than bovine pericardial ones [<xref ref-type="bibr" rid="B3">3</xref>] and among these, we focused on three bioprostheses offering high hemodynamic performance, especially for small aortic valves.</p><p>Since March 2010, a new pericardial aortic valve bioprosthesis (Trifecta valve, St. Jude Medical, Minneapolis, MN, USA) has been approved by the Food and Drugs Administration (FDA approval: St Jude Medical Trifecta Valve - P100029) and has recently received the CE mark. However, its hemodynamic characteristics still need to be compared with other bioprostheses already available on the market. Another bioprosthesis chosen for this trial is the Mitroflow valve (Sorin Group, Saluggia, Italy) [<xref ref-type="bibr" rid="B4">4</xref>]: it received the CE mark in July 2011 and it is characterized by an innovative phospholipid reduction treatment (PRT) expressively conceived to reduce the calcification process and, as a consequence, to improve its durability.</p><p>The third bioprosthesis is the Carpentier-Edwards Magna Ease valve (Edwards Lifesciences, Irvine, CA, USA) which has been designed by developing the renowned and highly performing Carpentier-Edwards PERIMOUNT valves. It allows easier implantation and its pericardial tissue is additionally treated to prevent calcification. This bioprosthesis received the CE mark in 2007 and FDA approval in 2009.</p><sec><title>Objectives of the Trivalve study</title><p>Each aortic valve has its own hemodynamic characteristics related to its geometry and each patient has their own morphology (weight, size, anatomy of aortic valve), as well as different physiological and pathophysiological conditions (ejection fraction, size and degree of calcification of the aortic annulus, degree of left ventricular hypertrophy and so on). Consequently, the choice of valve prosthesis and the surgical implantation technique are the only two directly adjustable variables; nonetheless, current literature does not provide clear differences among available bioprostheses.</p><p>The main objective of this study is to measure the hemodynamic performance of the three aortic bioprostheses: the Trifecta valve (St. Jude Medical, Minneapolis, MN, USA), the Mitroflow valve (Sorin Group, Saluggia, Italy), and the Magna Ease valve (Edwards Lifesciences, Irvine, CA, USA) Table <xref ref-type="table" rid="T1">1</xref>.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Bioprosthesis characteristics</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Bioprosthesis type</bold></th><th align="left"><bold>Manufacturer</bold></th><th align="left"><bold>Valve diameters (mm)</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Trifecta<hr/></td><td align="left" valign="bottom">St. Jude Medical, Minneapolis, USA<hr/></td><td align="left" valign="bottom">19 to 29<hr/></td></tr><tr><td align="left" valign="bottom">Mitroflow<hr/></td><td align="left" valign="bottom">Sorin Group, Saluggia, Italy<hr/></td><td align="left" valign="bottom">19 to 29<hr/></td></tr><tr><td align="left">Carpentier-Edwards Magna Ease</td><td align="left">Edwards Lifesciences, Irvine, USA</td><td align="left">19 to 29</td></tr></tbody></table></table-wrap><p>Secondary end points will focus on 1) comparison between the effective aortic orifice area measured by computerized tomography (CT)-scan and echocardiography and the intraoperative measurement performed by a flat-head candle; 2) comparison between the diameter of the aortic orifice measured by a flat-head candle and the size of the implanted bioprosthesis provided by the manufacturer; 3) testing the accuracy of information provided by the manufacturers about bioprosthesis diameters.</p></sec></sec><sec><title>Methods and design</title><p>The Trivalve trial is a single-center, prospective, randomized trial. It will evaluate the short-term (six month) hemodynamic performance of three pericardial bioprostheses: the Trifecta valve (St. Jude Medical, Minneapolis, MN, USA), the Mitroflow valve (Sorin Group, Saluggia, Italy), and the Carpentier-Edwards Magna Ease valve (Edwards Lifesciences, Irvine, CA, USA), Figure <xref ref-type="fig" rid="F1">1</xref>.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p>Study’s flowchart.</p></caption><graphic xlink:href="1745-6215-14-413-1"/></fig><p>ClinicalTrials.gov Identifier: NCT01522352.</p><sec><title>Patient’s enrollment and randomization</title><p>All patients scheduled for surgical aortic valve replacement by bioprosthesis will be screened according to inclusion and exclusion criteria (Table <xref ref-type="table" rid="T2">2</xref>).</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Inclusion and exclusion criteria</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"> </th><th align="left"> </th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom"><bold>Inclusion criteria</bold><hr/></td><td align="left" valign="bottom">Isolated aortic valve replacement or associated with myocardial revascularization and/or tricuspid valve repair<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Age (> 18 years and < 85 years)<hr/></td></tr><tr><td align="left" valign="bottom"><bold>Exclusion criteria</bold><hr/></td><td align="left" valign="bottom">Emergency surgery<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Surgery other than full sternotomy<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Heart transplantation<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Any procedure involving the aorta (such as Bentall procedure, surgery for dissection, and so on)<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Redo surgery<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Active infective endocarditis<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Associated mitral valve surgery<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Heart failure (ejection fraction < 40%) or preoperative cardiogenic shock<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Systolic pulmonary arterial pressure > 60 mmHg<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Patient’s protocol refusal<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Pregnancy<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Mentally handicapped patients, pre-existing psychiatric disease or addiction<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Advanced respiratory failure (forced expiratory volume in 1 second or vital capacity below 50% of the predicted)<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Severe renal failure<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">History of allergy or intolerance to iodinated contrast infusion<hr/></td></tr><tr><td align="left"> </td><td align="left">Patients living more than 100 km away from the investigation center</td></tr></tbody></table></table-wrap><p>Patients who have given their signed informed consent to participate in this clinical trial will undergo, preoperatively, a CT-scan and a transthoracic echocardiogram to measure the aortic annulus. Included patients will be randomly allocated to receive one of the three bioprostheses, in a 1:1:1 ratio. When a patient is considered eligible and informed consent has been obtained, randomization will be performed automatically (using STATA software (StataCorp, College Station, TX, USA) before surgery by an independent biostatistician. No stratification will be done. The selected bioprosthesis will be implanted.</p></sec><sec><title>Preoperative measurements</title><p>Preoperative CT-scan measured data, echocardiography and surgical measurements are shown in Table <xref ref-type="table" rid="T3">3</xref>.</p><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>CT-scan, echocardiographic and surgical measurements</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"> </th><th align="left"><bold>Preoperative data</bold></th><th align="left"><bold>Operative data</bold></th><th align="left"><bold>Day 7 data</bold></th><th align="left"><bold>Month 6 data</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">CT-scan<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Native aortic annulus (mm)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Ascending aorta diameter (mm)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Echocardiography<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">LVTS (mm)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">LVTD (mm)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">LVPWT (mm)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">IVST (mm)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">LVSF (%)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">LVEF (%)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Pulmonary arterial pressure<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Cardiac output (L/min<sup>-1</sup>)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Cardiac index (L/min<sup>-1</sup>/m<sup>-2</sup>)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Mean transvalvular gradient (mmHg)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Maximal transvalvular gradient (mmHg)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Aortic orifice area (m<sup>2</sup>)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Aortic regurgitation degree (0–4)<hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Paravalvular leak<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom">X<hr/></td></tr><tr><td align="left" valign="bottom">Surgery<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Internal aortic annulus diameter (mm)<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">Estimated valve diameter (mm)<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">X<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left">Implanted valve diameter (mm)</td><td align="left"> </td><td align="left">X</td><td align="left"> </td><td align="left"> </td></tr></tbody></table><table-wrap-foot><p>IVST: inter ventricular septum thickness, LVEF: left ventricular ejection fraction, LVPWT: left ventricle posterior wall thickness, LVSF: left ventricular shortening fraction, LVTD: left ventricle tele diastolic diameter, LVTS: left ventricle tele systolic diameter.</p></table-wrap-foot></table-wrap></sec><sec><title>Surgery</title><p>During surgery, the aortic valve of the patient will be completely removed and the aortic annulus measured using a flat-head candle. This universal candle has been specifically designed to give a single objective value of the internal diameter of the aortic annulus. The Hegar dilators will not be used as its round shape and arched aerodynamics overestimate the size of the annulus by applying opening force to its passage (Figure <xref ref-type="fig" rid="F2">2</xref>). All measurements, early postoperative complications and reoperations for bleeding will be recorded in the operative report and in the case-report form.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p>Importance of the use of flat-head candles to measure the aortic ring (bottom of the picture) compared to Hegar dilators.</p></caption><graphic xlink:href="1745-6215-14-413-2"/></fig></sec><sec><title>Postoperative endpoints</title><p>All preoperative echocardiography collected measures will be reassessed at day seven and month six after surgery in addition to maximal and mean transvalvular gradients (mmHg). ICU, total hospital stay and any other postoperative complications will be recorded in the postoperative report and in the case-report form.</p><p>The primary endpoint is the mean transvalvular gradient (mmHg) six months after surgery. All secondary endpoints are indicated in Table <xref ref-type="table" rid="T4">4</xref>.</p><table-wrap position="float" id="T4"><label>Table 4</label><caption><p>Primary and secondary endpoints</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"> </th><th align="left"> </th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom"><bold>Primary endpoint</bold><hr/></td><td align="left" valign="bottom">Mean transvalvular gradient (mmHg), six months after surgery<hr/></td></tr><tr><td rowspan="3" align="left" valign="bottom"><bold>Secondary endpoints</bold><hr/></td><td align="left" valign="bottom">Effective aortic orifice diameter measured by CT-scan and echocardiography compared with the surgical data (mm)<hr/></td></tr><tr><td align="left" valign="bottom">Mean transvalvular gradient (mmHg), at day seven after surgery<hr/></td></tr><tr><td align="left" valign="bottom">Aortic bioprosthesis orifice area (m<sup>2</sup>), six months after surgery<hr/></td></tr><tr><td align="left"> </td><td align="left">Diameter of the aortic orifice measured by a flat-head candle compared with the size of the implanted bioprosthesis (mm)</td></tr></tbody></table></table-wrap></sec><sec><title>Statistical considerations</title><sec><title>Sample size estimation</title><p>The estimation of the number of patients required was considered by using previous data provided by the manufacturers on patients who had cardiovascular surgery which showed the mean postoperative gradients (mmHg) for the three studied types of valve (data not published). A minimum difference (δ) of 4 mmHg could be expected between the three types of valve for the most relevant diameters (21 and 23 mm). As the information concerning statistical variability was not provided in this document, the standard deviation (σ) was estimated on the basis of data from 103 patients observed in our center (84 with the Edwards Ease prosthesis, 19 with a Mitroflow diameter of 21 or 23 mm): σ = 5.8. Thus, for a type 1 error α = 0.05 (two-sided), a 90 %-power, δ = 4 and σ = 5.8, 44 subjects per group are needed. Taking into account multiple comparisons between the three randomized groups, 55 patients per group will be included (165 patients in total).</p></sec></sec><sec><title>Statistical analyses</title><p>All analyses will be performed using STATA v11 (StataCorp, College Station, TX, USA). A two-tailed <italic>P</italic>-value of 0.05 will be considered statistically significant. All analyses will be performed on an intention-to-treat (ITT) basis. The number of included patients and the rate of inclusions will be presented over time for each group. The patients will be described and compared between groups at baseline according to the following variables: compliance with eligibility criteria, epidemiological features, clinical features (including echocardiographic) and biological characteristics. The comparison concerning the postoperative means of the transvalvular gradients (measured by echocardiography at six months post surgery) between the three groups will be evaluated using ANOVA followed by the Tukey-Kramer <italic>post hoc</italic> test, or the Kruskal-Wallis nonparametric test if conditions of ANOVA are not met (homoscedasticity studied by Bartlett’s test and normality verify by Shapiro-Wilk) followed by Dunn’s test as appropriate. Comparisons between the groups will be realized systematically 1) without adjustment and 2) when appropriate, after adjustment (by multivariate linear regression model) on factors whose distribution could be unbalanced between the arms despite randomization. Quantitative secondary endpoints (for example hemodynamic data, CT-scan, in-hospital stay) will be analyzed as described above. Categorical parameters (that is, proportion of reoperations) will be compared between the groups using the chi-squared test or Fisher’s exact test, when necessary. To assess the relationships between the quantitative parameters (comparison between aortic orifice measurements by echocardiography, CT-scans, intraoperative measurement using the flat- candle versus the size of implanted valve prosthesis given by the manufacturer), the correlation coefficients (Pearson or Spearman), the Lin concordance coefficient and the intra-class coefficient (ICC) will be calculated. Later on, an ANCOVA could be proposed to consider group effect. The intra-group comparisons related to the quantitative criteria (hemodynamic data by echocardiography on preoperative period and at six months) will be made using paired the ANOVA or Wilcoxon test. Finally, to avoid bias induced by the presence of missing data, particularly with regards to the mean postoperative transvalvular gradient at six months (lost at follow-up or deaths), the primary analysis (ITT with imputation data determined according to quantity and type of missing data) will be completed on a second time by a per-protocol analysis.</p></sec><sec><title>Expected adverse events</title><p>These three prosthetic valves are made of three layers of fixed bovine pericardium assembled on a support (stent). They are then fixed in a glutaraldehyde solution and conditioned in a sterile manner. A correctly sized and implanted valve leads to very few complications. They have an average lifespan of > 10 years when implanted in patients aged > 65 years [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>]. The expected adverse events of these bioprostheses are those of usual heart valve replacement surgery on cardiopulmonary bypass and mortality can be predicted by Euroscore 2 [<xref ref-type="bibr" rid="B5">5</xref>], which is systematically calculated for all our patients. Postoperative adverse events will be evaluated according to the Clavien-Dindo classification for surgical complications [<xref ref-type="bibr" rid="B6">6</xref>].</p></sec><sec><title>Funding</title><p>Edwards Lifesciences (Irvine, CA, USA), St. Jude Medical (Minneapolis, MN, USA), and Sorin Group (Saluggia, Italy) gave a contribution of €5,000 each and the Hospital Clinical Research Program (PHRC) of the French Ministry of Health contributed an amount of €15,000 for the realization of this study.</p></sec></sec><sec><title>The status of this trial</title><p>This trial has been actively recruiting patients since March 2012. The French Committee on Human Research (CPP Sud-Est VI) consented to this trial on 17 January 2012. Patients give their informed consent before being enrolled in this study. Agreement from the French Competent Authority (ANSM) was obtained on 23 June 23 2011. The completion date for this study is estimated as December 2014. The ClinicalTrials.gov identifier is NCT01522352.</p></sec><sec sec-type="discussion"><title>Discussion</title><p>Several studies have emphasized the importance of valve prosthesis hemodynamic performance [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B8">8</xref>], but none has taken into account more than one or two bioprostheses at a time [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. Thus, we strongly believe that a randomized trial, with no direct conflicts of interest with industry, is mandatory to compare the three bioprostheses most commonly implanted in France and all around the world.</p><p>Our secondary objective is to compare the reliability of preoperative CT-scan and echocardiography used to assess the size of the aortic annulus in comparison to the surgical measurement. Both techniques (CT-scans and echocardiography) are already successfully used to predict the correct size of the aortic annulus before trans-catheter valve implantation procedures [<xref ref-type="bibr" rid="B11">11</xref>]. Another secondary objective, once we have assessed the surgical diameter of the aortic annulus intraoperatively, is to verify the reliability of the prosthesis size provided by the manufacturer. This last point is interesting as it aims to clarify an issue often debated by surgeons who complain to manufacturers that they over- or underestimate valve sizes. Precise information on the size of the implanted bioprosthesis compared to the real dimensions of the aortic annulus will guide cardiac surgeons to choose between these three bioprostheses according to the patient’s morphological characteristics.</p></sec><sec><title>Abbreviations</title><p>ANSM: Agence nationale de sécurité du médicament; CE: Conformité Européenne; CPP: Comité de protection des personnes; CT: Computerized Tomography; FDA: Food and drugs administration; ICC: Intra-class coefficient; ITT: Intention-to-treat; mmHg: Millimeter of mercury; PRT: Phospholipid reduction treatment.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests other than mentioned in this manuscript. Edwards Lifesciences (Irvine, CA, USA), St. Jude Medical (Minneapolis, MN, USA), and Sorin Group (Saluggia, Italy) gave a contribution of €5000 each and the Hospital Clinical Research Program (PHRC) of the French Ministry of Health contributed an amount of €15,000 for the realization of this study.</p></sec><sec><title>Authors’ contributions</title><p>KA: conception, design, surgical management and writing of the manuscript. BP: conception, design and statistical management. CD: conception, design and statistical management. ED: critical revisions of the manuscript. MF: conception, design and surgical management. AI: conception, design and surgical management. ND: conception, design and echocardiography. EG: conception, design and echocardiography. PC: conception, design and CT-scan. LC: conception, design, surgical management, critical revisions and final approval. All authors read and approved the final manuscript.</p></sec>
Risk of preterm delivery with increasing depth of excision for cervical intraepithelial neoplasia in England: nested case-control study	<p><bold>Objective</bold> To determine the association between depth of excision of cervical intraepithelial neoplasia and risk of preterm birth.</p><p><bold>Design</bold> Case-control study nested in record linkage cohort study.</p><p><bold>Setting</bold> 12 hospitals in England.</p><p><bold>Participants</bold> From a cohort of 11 471 women with at least one histological sample taken at colposcopy and a live singleton birth (before or after colposcopy), 1313 women with a preterm birth (20-36 weeks) were identified and frequency matched on maternal age at delivery, parity, and study site to 1313 women with term births (38-42 weeks).</p><p><bold>Main outcome measures</bold> Risk of preterm birth and very/extreme preterm birth by depth of excisional treatment of the cervical transformation zone.</p><p><bold>Results</bold> After exclusions, 768 preterm births (cases) and 830 term births after colposcopy remained. The risk of preterm birth was no greater in women with a previous small (<10 mm) excision (absolute risk 7.5%, 95% confidence interval 6.0% to 8.9%) than in women with a diagnostic punch biopsy (7.2%, 5.9% to 8.5%). Women with a medium (10-14 mm) (absolute risk 9.6%; relative risk 1.28, 0.98 to 1.68), large (15-19 mm) (15.3%; 2.04, 1.41 to 2.96), or very large (≥20 mm) excision (18.0%; 2.40, 1.53 to 3.75) had a higher risk of preterm delivery than those with small excision. The same pattern was seen in 161 women with very/extremely preterm births (20-31 weeks) and with increasing volume excised. Most births were conceived more than three years after colposcopy, and the risk of preterm delivery did not seem to depend on time from excision to conception.</p><p><bold>Conclusions</bold> The risk of preterm birth is at most minimally affected by a small excision. Larger excisions, particularly over 15 mm or 2.66 cm<sup>3</sup>, are associated with a doubling of the risk of both preterm and very preterm births. The risk does not decrease with increasing time from excision to conception. Efforts should be made to excise the entire lesion while preserving as much healthy cervical tissue as possible. Close obstetric monitoring is warranted for women who have large excisions of the cervical transformation zone.</p>	<contrib contrib-type="author"><name><surname>Castanon</surname><given-names>Alejandra</given-names></name><role>epidemiologist</role><xref ref-type="aff" rid="aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Landy</surname><given-names>Rebecca</given-names></name><role>statistician</role><xref ref-type="aff" rid="aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Brocklehurst</surname><given-names>Peter</given-names></name><role>professor of women’s health</role><xref ref-type="aff" rid="aff2">2</xref></contrib><contrib contrib-type="author"><name><surname>Evans</surname><given-names>Heather</given-names></name><role>consultant in obstetrics and gynaecology</role><xref ref-type="aff" rid="aff3">3</xref></contrib><contrib contrib-type="author"><name><surname>Peebles</surname><given-names>Donald</given-names></name><role>professor of maternal and foetal medicine</role><xref ref-type="aff" rid="aff2">2</xref></contrib><contrib contrib-type="author"><name><surname>Singh</surname><given-names>Naveena</given-names></name><role>consultant histopathologist</role><xref ref-type="aff" rid="aff4">4</xref></contrib><contrib contrib-type="author"><name><surname>Walker</surname><given-names>Patrick</given-names></name><role>consultant in obstetrics and gynaecology</role><xref ref-type="aff" rid="aff3">3</xref></contrib><contrib contrib-type="author"><name><surname>Patnick</surname><given-names>Julietta</given-names></name><role>director </role><xref ref-type="aff" rid="aff5">5</xref></contrib><contrib contrib-type="author" corresp="yes"><name><surname>Sasieni</surname><given-names>Peter</given-names></name><role>professor of biostatistics and cancer epidemiology</role><xref ref-type="aff" rid="aff1">1</xref></contrib><on-behalf-of>for the PaCT Study Group</on-behalf-of><aff id="aff1"><label>1</label>Centre for Cancer Prevention, Wolfson Institute of Preventive Medicine, Bart’s and the London School of Medicine, Queen Mary University of London, London EC1M 6BQ, UK</aff><aff id="aff2"><label>2</label>Institute for Women’s Health, University College London, London, UK</aff><aff id="aff3"><label>3</label>Department of Gynaecology, Royal Free Hampstead NHS Trust, London, UK</aff><aff id="aff4"><label>4</label>Division of Cellular Pathology, Barts Health, London, UK</aff><aff id="aff5"><label>5</label>NHS Cancer Screening Programmes, Public Health England, Sheffield, UK</aff>	The BMJ	<sec sec-type="intro"><title>Introduction</title><p>Observational studies and two meta-analyses have found that women treated for cervical intraepithelial neoplasia (CIN) are at increased risk of preterm delivery.<xref rid="ref1" ref-type="bibr">1</xref> <xref rid="ref2" ref-type="bibr">2</xref> <xref rid="ref3" ref-type="bibr">3</xref> The literature suggests an increased risk with greater depth of excision,<xref rid="ref4" ref-type="bibr">4</xref> <xref rid="ref5" ref-type="bibr">5</xref> but confidence intervals are wide and whether a safe depth exists below which the risk is not increased is unclear.</p><p>As part of the NHS cervical screening programme in England, women with abnormal cytology are referred to colposcopy for further investigation. At the first appointment, management can vary from colposcopy only (for low grade cytology and normal colposcopy) to colposcopy plus a diagnostic punch biopsy (for low to moderate cytology with a colposcopic abnormality), and a few women may be treated (“see and treat” for high grade cytology with a defined colposcopic lesion). Women found to have high grade CIN on punch biopsy are treated most commonly by large loop excision of the transformation zone. Rigorous quality assurance in the cervical screening programme could explain why performance data for colposcopy and treatment in the NHS are often higher than those found in international studies.<xref rid="ref6" ref-type="bibr">6</xref> <xref rid="ref7" ref-type="bibr">7</xref></p><p>Results from the earlier phase of the study reported here found that compared with the general population, women attending colposcopy had an additional risk of preterm birth of 2.1 per 100 births (8.8% compared with 6.7%).<xref rid="ref8" ref-type="bibr">8</xref> The risk was greater in women who were treated than in those with only a diagnostic punch biopsy, both when the treatment preceded the birth (relative risk 1.19, 95% confidence interval 1.01 to 1.41) and when the birth preceded colposcopy (1.31, 0.97 to 1.76). Therefore, the increased risk in women attending colposcopy may be a consequence of confounding and not caused by treatment. In this phase of the study, we aimed to explore the association between preterm birth and the depth and volume of tissue removed in treatment for cervical disease, the number of excisions carried out before the birth, and the time from treatment to conception.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Participants</title><p>We identified women with cervical histology between April 1988 and December 2011 from clinical records in 12 NHS hospitals. They were linked using their NHS number (a unique identifier) and date of birth by Hospital Episode Statistics to hospital obstetric records between April 1998 and March 2011 for the whole of England. Hospital Episode Statistics is a data warehouse containing details of all admissions to NHS hospitals in England.<xref rid="ref9" ref-type="bibr">9</xref> From Hospital Episode Statistics records, we obtained information on month and year of each delivery, birth weight, whether the birth was a normal vaginal or operative delivery, the mode of onset of labour (spontaneous versus induced), parity, overall index of multiple deprivation at the time of delivery, and any other inpatient diagnoses or operations recorded for the mother.</p><p>As obtaining colposcopy and pathology data was the most resource intensive aspect of our pilot study, and many of the women who had attended colposcopy had no births, we designed this study in two phases: the cohort study (phase I)<xref rid="ref6" ref-type="bibr">6</xref> and the nested case-control study (phase II) reported here. In addition to the births reported in phase I, we identified (using the same methods) 90 from a new site (Worcestershire) and 30 extra births from patients at Hammersmith Hospital (fig 1<xref ref-type="fig" rid="fig1"/>).</p><fig id="fig1" position="float"><caption><p><bold>Fig 1</bold> Inclusions in and exclusions from study</p></caption><graphic xlink:href="casa019275.f1_default"/></fig><p>We divided preterm births (gestational age of 20-36 completed weeks) into moderate (32-36 weeks) and very/extreme preterm (20-31 weeks).<xref rid="ref10" ref-type="bibr">10</xref> Term births had a gestational age of 38-42 completed weeks. We excluded births at 37 weeks’ gestational age to allow a clear divide between term and preterm births. From the cohort, we identified the earliest occurring singleton preterm birth (with any parity) in each woman and frequency matched these to singleton term births in women with no preterm births. Matching was on maternal age at delivery, parity, study site, and whether the birth occurred before or after the first colposcopy to ensure similar characteristics among women with term and preterm births. Only one birth per woman was included in the case-control study.</p><p>Hospitals entered colposcopy details into a study database and submitted anonymised pathology reports to Barts Health NHS Trust. Two trained operators entered pathology reports into the study database to ensure that measurements were entered in a standardised way, facilitating the identification of the length, width, and depth of specimens. People searching for and coding colposcopy information were blind to the case-control status of the women.</p><p>Extensive data cleaning and checking led the exclusion of one site and of 63 births for which colposcopy records were known to be incomplete (see supplementary methods and fig 1<xref ref-type="fig" rid="fig1"/>). We also excluded women for whom the only pathology sample reported was non-cervical (n=18) and one woman who was recorded as being sterilised while pregnant.</p><p>We excluded women with a diagnosis of cervical cancer at any time (n=34). We also excluded 173 women whose pregnancy was high risk (ICD10 (international classification of diseases, 10th revision) diagnostic codes in supplementary methods and supplementary table A). These diagnoses included diabetes mellitus, hypertension, placenta praevia with haemorrhage, supervision of high risk pregnancy, mental disorders, and diseases of the nervous system complicating pregnancy, childbirth, and the puerperium. We defined excisional treatment as large loop excision of the transformation zone, laser excision, knife cone biopsy, or cone excision not otherwise specified. This paper is restricted to births after colposcopy (188 cases and 209 controls had births before colposcopy).</p></sec><sec><title>Statistical methods</title><p>The main exposure of interest was depth of excision before birth, defined as the distance from the distal or external margin to the proximal or internal margin of the excised specimen.<xref rid="ref11" ref-type="bibr">11</xref> In all participating laboratories, the standard process was to report the depth as the last of three measurements, whereas the reporting of the other two measurements was arbitrary. When the excision was piecemeal, we used the largest fragment depth. For women with more than one excisional treatment, we summed the depths. To assess whether the risk after multiple excisions was greater than that associated with the total depth of tissue excised, we included an indicator for multiple treatments before birth. We did the same analysis to assess the effect of piecemeal excisions. We grouped depth (in mm) as 0-9, 10-14, 15-19, ≥20, and unknown depth in accordance with the pre-specified statistical analysis plan. We chose these because they are of practical use and comparable to other cut-offs reported in the literature.</p><p>We calculated the volume of the samples assuming treatments were hemi-ellipsoid in shape, using the formula “volume=(1/2)×(4/3)×π×(length/2)×(width/2)×depth.”<xref rid="ref12" ref-type="bibr">12</xref> When only the depth and either length or width were recorded, we assumed the length or width to be the diameter of a circular base. When more than one fragment existed or a woman had multiple treatments, we summed the volumes. Following the statistical analysis plan, we categorised volume by using the 60th and 80th centiles in controls as cut-offs, corresponding to 1.77 cm<sup>3</sup> and 2.66 cm<sup>3</sup>. We developed a risk classification combining depth and volume. We defined time from treatment to conception as time from the last excisional treatment to the date of birth, minus the gestational age at birth.</p><p>We calculated relative risks and 95% confidence intervals by using inverse probability of sampling weighted relative risk regression (a generalised linear model with binomial family and log link function)<xref rid="ref13" ref-type="bibr">13</xref> adjusted for maternal age at delivery (<25, 25-34, >34), parity (none, one, two, three or more), index of multiple deprivation (in national fifths), and study site. The weights reflect the proportions of term and preterm births from the cohort (phase I) included in the case-control study (see supplementary methods). We estimated absolute risks by the average predicted probability of a preterm birth (averaging over a standardised distribution of confounding variables). We tested trends among women who had excisions, excluding those with unknown depth.</p><p>In addition to the analysis including all births, to assess sensitivity to the inclusion criteria we re-analysed the data restricting the analyses to preterm births with a spontaneous onset of labour; separating preterm births after excisional treatments that were known to be large loop excisions of the transformation zone from those known to be other excisional procedures; separating preterm births after a piecemeal excision from those with a single piece excised; excluding births with improbable birth weight (>3.5 kg if preterm or <2.5 kg if term); excluding women with a diagnosis of CIN2/CIN3 who did not have excisional treatment (as the clinical recommendation is to treat, we may be missing excisions provided in a different clinic); and excluding births after multiple excisional treatments. We used Stata 12 for all analyses.</p></sec></sec><sec sec-type="results"><title>Results</title><p>From a cohort of 11 471 women and 15 718 births, we identified 1313 women with a preterm birth and matched them to 1313 women with term births (fig 1<xref ref-type="fig" rid="fig1"/>). We obtained full colposcopy details for 87% (2284/2626) of the selected women. After exclusions, 1598 post-colposcopy births (768 preterm and 830 term) were included. Seventy per cent of women had an excisional treatment— most (1005/1101 (91%) of the women with a single excision before birth) were large loop excision of the transformation zone—and measurements were obtained for 88% (1193/1360) of excisions. Among the 1114 women with an excision, 301 (27%) had piecemeal samples and 99 (8.9%) had more than one excisional treatment before the birth.</p><p>Of the preterm births, 607 (79.0%) had a gestational age of 32-36 weeks and 161 (21.0%) were very/extreme preterm. Women with a very preterm birth were more likely to have undergone excision than were those with term births (78% <italic>v</italic> 67%; P=0.020) (table 1<xref ref-type="table" rid="tbl1"/>).</p><table-wrap id="tbl1" position="float"><label>Table 1</label><caption><p> Main characteristics of women included in study (n=1598). Values are numbers (percentages) unless stated otherwise</p></caption><table frame="hsides" rules="groups"><thead><tr><th colspan="1" rowspan="1" align="left" valign="bottom">Characteristics</th><th colspan="1" rowspan="1" align="center" valign="bottom">Term birth (38-43 weeks) (n=830)</th><th colspan="1" rowspan="1" align="center" valign="bottom">Moderate preterm birth (32-36 weeks) (n=607)</th><th colspan="1" rowspan="1" align="center" valign="bottom">Very/extreme preterm birth (20-31 weeks) (n=161)</th><th colspan="1" rowspan="1" align="center" valign="bottom">Absolute risk of preterm birth (20-36 weeks) (%)</th></tr></thead><tbody><tr><td colspan="5" rowspan="1" align="left" valign="top" content-type="TableSubHead"><bold>Maternal age at delivery (years)</bold></td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top"><20</td><td colspan="1" rowspan="1" align="center" valign="top">2 (0.2)</td><td colspan="1" rowspan="1" align="center" valign="top">1 (0.2)</td><td colspan="1" rowspan="1" align="center" valign="top">1 (0.6)</td><td colspan="1" rowspan="1" align="center" valign="top">9.5</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">20 to 24</td><td colspan="1" rowspan="1" align="center" valign="top">47 (5.7)</td><td colspan="1" rowspan="1" align="center" valign="top">48 (7.9)</td><td colspan="1" rowspan="1" align="center" valign="top">15 (9)</td><td colspan="1" rowspan="1" align="center" valign="top">12.3</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">25 to 29</td><td colspan="1" rowspan="1" align="center" valign="top">204 (24.6)</td><td colspan="1" rowspan="1" align="center" valign="top">161 (26.5)</td><td colspan="1" rowspan="1" align="center" valign="top">45 (28)</td><td colspan="1" rowspan="1" align="center" valign="top">9.6</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">30 to 34</td><td colspan="1" rowspan="1" align="center" valign="top">356 (42.9)</td><td colspan="1" rowspan="1" align="center" valign="top">244 (40.2)</td><td colspan="1" rowspan="1" align="center" valign="top">59 (37)</td><td colspan="1" rowspan="1" align="center" valign="top">8.2</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">35 to 39</td><td colspan="1" rowspan="1" align="center" valign="top">196 (23.6)</td><td colspan="1" rowspan="1" align="center" valign="top">137 (22.6)</td><td colspan="1" rowspan="1" align="center" valign="top">36 (22)</td><td colspan="1" rowspan="1" align="center" valign="top">8.5</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">≥40</td><td colspan="1" rowspan="1" align="center" valign="top">25 (3.0)</td><td colspan="1" rowspan="1" align="center" valign="top">16 (2.6)</td><td colspan="1" rowspan="1" align="center" valign="top">5 (3)</td><td colspan="1" rowspan="1" align="center" valign="top">8.1</td></tr><tr><td colspan="5" rowspan="1" align="left" valign="top" content-type="TableSubHead"><bold>Parity before index birth</bold></td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">None</td><td colspan="1" rowspan="1" align="center" valign="top">423 (51.0)</td><td colspan="1" rowspan="1" align="center" valign="top">314 (51.7)</td><td colspan="1" rowspan="1" align="center" valign="top">87 (54)</td><td colspan="1" rowspan="1" align="center" valign="top">9.0</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">1</td><td colspan="1" rowspan="1" align="center" valign="top">205 (24.7)</td><td colspan="1" rowspan="1" align="center" valign="top">154 (25.4)</td><td colspan="1" rowspan="1" align="center" valign="top">36 (22)</td><td colspan="1" rowspan="1" align="center" valign="top">8.9</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">2</td><td colspan="1" rowspan="1" align="center" valign="top">100 (12.0)</td><td colspan="1" rowspan="1" align="center" valign="top">67 (11.0)</td><td colspan="1" rowspan="1" align="center" valign="top">25 (16)</td><td colspan="1" rowspan="1" align="center" valign="top">8.8</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">≥3</td><td colspan="1" rowspan="1" align="center" valign="top">102 (12.3)</td><td colspan="1" rowspan="1" align="center" valign="top">72 (11.9)</td><td colspan="1" rowspan="1" align="center" valign="top">13 (8)</td><td colspan="1" rowspan="1" align="center" valign="top">8.0</td></tr><tr><td colspan="5" rowspan="1" align="left" valign="top" content-type="TableSubHead"><bold>Worst diagnosis at colposcopy</bold></td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">Normal/benign</td><td colspan="1" rowspan="1" align="center" valign="top">109 (13.1)</td><td colspan="1" rowspan="1" align="center" valign="top">68 (11.2)</td><td colspan="1" rowspan="1" align="center" valign="top">17 (11)</td><td colspan="1" rowspan="1" align="center" valign="top">7.6</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">Low grade disease</td><td colspan="1" rowspan="1" align="center" valign="top">233 (28.1)</td><td colspan="1" rowspan="1" align="center" valign="top">136 (22.4)</td><td colspan="1" rowspan="1" align="center" valign="top">34 (21)</td><td colspan="1" rowspan="1" align="center" valign="top">7.1</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">High grade disease†</td><td colspan="1" rowspan="1" align="center" valign="top">480 (57.8)</td><td colspan="1" rowspan="1" align="center" valign="top">394 (64.9)</td><td colspan="1" rowspan="1" align="center" valign="top">110 (68)</td><td colspan="1" rowspan="1" align="center" valign="top">9.9</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">Inadequate/unknown</td><td colspan="1" rowspan="1" align="center" valign="top">8 (1.0)</td><td colspan="1" rowspan="1" align="center" valign="top">9 (1.5)</td><td colspan="1" rowspan="1" align="center" valign="top">0 (0)</td><td colspan="1" rowspan="1" align="center" valign="top">10.6</td></tr><tr><td colspan="5" rowspan="1" align="left" valign="top" content-type="TableSubHead"><bold>Most invasive procedure at colposcopy</bold></td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">Punch biopsy</td><td colspan="1" rowspan="1" align="center" valign="top">274 (33.0)</td><td colspan="1" rowspan="1" align="center" valign="top">175 (28.8)</td><td colspan="1" rowspan="1" align="center" valign="top">35 (22)</td><td colspan="1" rowspan="1" align="center" valign="top">7.4</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">LLETZ/loop/cone‡</td><td colspan="1" rowspan="1" align="center" valign="top">556 (67.0)</td><td colspan="1" rowspan="1" align="center" valign="top">432 (71.2)</td><td colspan="1" rowspan="1" align="center" valign="top">126 (78)</td><td colspan="1" rowspan="1" align="center" valign="top">9.5</td></tr></tbody></table><table-wrap-foot><p>Includes human papilloma virus, cervical intraepithelial neoplasia (CIN) grade 1, and low grade glandular CIN.</p><p>†Includes CIN grade 2 and 3, high grade glandular CIN, and ungraded CIN or glandular CIN.</p><p>‡Large loop excision of the transformation zone (LLETZ), loop excision, or cone excision.</p></table-wrap-foot></table-wrap><p>The relative risk of a preterm birth was 38% greater in women who had an excision than in those who had a punch biopsy (relative risk 1.38, 95% confidence interval 1.10 to 1.72). The relative risk of a preterm birth among women who had multiple excisional treatments before the birth (1.95, 1.28 to 2.95) was greater than the risk among those who had a single excision (1.34, 1.06 to 1.68). However, after adjustment for (total) depth of excision, the relative risk associated with multiple treatments was 1.14 (0.75 to 1.76). This suggests that the excess risk in women with multiple excisions is primarily due to the large amount of tissue removed (that is, the total depth). Therefore, we assigned women with multiple excisional treatments before birth to the appropriate depth category, rather than including them in a separate category. Similarly, the risk of preterm birth in women with piecemeal excisions was non-significantly higher than that predicted by the depth of the largest fragment (relative risk 1.15, 0.89 to 1.49), so they are not considered separately.</p><p>Among women with excisional treatment, the risk of a preterm birth increased with the depth of excision (table 2<xref ref-type="table" rid="tbl2"/>) (P for trend<0.001). The risk of a preterm birth was significantly increased for depths of 15-19 mm (relative risk 2.04, 1.41 to 2.96) and 20 mm or more (2.40, 1.53 to 3.75) compared with women with small excisions (<10 mm). The relative risks among women with medium (1.28, 0.98 to 1.68) or unknown (1.24, 0.86 to 1.79) excisional depth were intermediate. The absolute risk of a preterm birth was 7.5% for small excisions, 9.6% for 10-14 mm excisions, 15.3% for 15-19 mm, and 18.0% for excisions 20 mm or more deep. The risk of preterm birth in women with a small excision was similar to that among women with a punch biopsy (7.5% <italic>v</italic> 7.2%; relative risk 1.04, 0.79 to 1.37). This result was robust in a variety of sub-analyses (fig 2<xref ref-type="fig" rid="fig2"/>).</p><table-wrap id="tbl2" position="float"><label>Table 2</label><caption><p> Relative and absolute risk of preterm birth by depth of excisional treatment in women attending colposcopy before birth</p></caption><table frame="hsides" rules="groups"><thead><tr><th colspan="1" rowspan="1" align="left" valign="bottom">Procedure at colposcopy</th><th colspan="1" rowspan="1" align="center" valign="bottom">No (%) cases</th><th colspan="1" rowspan="1" align="center" valign="bottom">No (%) controls</th><th colspan="1" rowspan="1" align="center" valign="bottom">Relative risk* (95% CI)</th><th colspan="1" rowspan="1" align="center" valign="bottom">Absolute risk (95% CI)</th></tr></thead><tbody><tr><td colspan="5" rowspan="1" align="left" valign="top" content-type="TableSubHead"><bold>All preterm births (20-36 weeks)</bold></td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">Punch biopsy before birth</td><td colspan="1" rowspan="1" align="center" valign="top">210 (27.3)</td><td colspan="1" rowspan="1" align="center" valign="top">274 (33.0)</td><td colspan="1" rowspan="1" align="center" valign="top">0.96 (0.73 to 1.27)</td><td colspan="1" rowspan="1" align="center" valign="top">7.2 (5.9 to 8.5)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">Treatment before birth:</td><td colspan="1" rowspan="1" align="center" valign="top"/><td colspan="1" rowspan="1" align="center" valign="top"/><td colspan="1" rowspan="1" align="center" valign="top"/><td colspan="1" rowspan="1" align="center" valign="top"/></tr><tr><td colspan="1" rowspan="1" align="left" valign="top"> Small excision (1-9 mm deep )</td><td colspan="1" rowspan="1" align="center" valign="top">173 (22.5)</td><td colspan="1" rowspan="1" align="center" valign="top">223 (26.9)</td><td colspan="1" rowspan="1" align="center" valign="top">1 (reference)</td><td colspan="1" rowspan="1" align="center" valign="top">7.5 (6.0 to 8.9)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top"> Medium excision (10-14 mm deep)</td><td colspan="1" rowspan="1" align="center" valign="top">182 (23.7)</td><td colspan="1" rowspan="1" align="center" valign="top">186 (22.4)</td><td colspan="1" rowspan="1" align="center" valign="top">1.28 (0.98 to 1.68)</td><td colspan="1" rowspan="1" align="center" valign="top">9.6 (7.7 to 11.5)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top"> Large excision (15-19 mm deep)</td><td colspan="1" rowspan="1" align="center" valign="top">80 (10.4)</td><td colspan="1" rowspan="1" align="center" valign="top">48 (5.8)</td><td colspan="1" rowspan="1" align="center" valign="top">2.04 (1.41 to 2.96)</td><td colspan="1" rowspan="1" align="center" valign="top">15.3 (10.5 to 20.1)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top"> Very large excision (≥20 mm deep)</td><td colspan="1" rowspan="1" align="center" valign="top">54 (7.0)</td><td colspan="1" rowspan="1" align="center" valign="top">28 (3.4)</td><td colspan="1" rowspan="1" align="center" valign="top">2.40 (1.53 to 3.75)</td><td colspan="1" rowspan="1" align="center" valign="top">18.0 (10.7 to 25.1)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top"> Unknown treatment depth</td><td colspan="1" rowspan="1" align="center" valign="top">69 (9.0)</td><td colspan="1" rowspan="1" align="center" valign="top">71 (8.6)</td><td colspan="1" rowspan="1" align="center" valign="top">1.24 (0.86 to 1.79)</td><td colspan="1" rowspan="1" align="center" valign="top">9.3 (6.4 to 12.2)</td></tr><tr><td colspan="5" rowspan="1" align="left" valign="top" content-type="TableSubHead"><bold>Spontaneous preterm births (20-36 weeks)</bold></td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">Punch biopsy before birth</td><td colspan="1" rowspan="1" align="center" valign="top">121 (23.7)</td><td colspan="1" rowspan="1" align="center" valign="top">274 (33.0)</td><td colspan="1" rowspan="1" align="center" valign="top">0.84 (0.61 to 1.17)</td><td colspan="1" rowspan="1" align="center" valign="top">4.3 (3.4 to 5.2)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">Treatment before birth:</td><td colspan="1" rowspan="1" align="center" valign="top"/><td colspan="1" rowspan="1" align="center" valign="top"/><td colspan="1" rowspan="1" align="center" valign="top"/><td colspan="1" rowspan="1" align="center" valign="top"/></tr><tr><td colspan="1" rowspan="1" align="left" valign="top"> Small excision (1-9 mm deep )</td><td colspan="1" rowspan="1" align="center" valign="top">117 (23.0)</td><td colspan="1" rowspan="1" align="center" valign="top">223 (26.9)</td><td colspan="1" rowspan="1" align="center" valign="top">1 (reference)</td><td colspan="1" rowspan="1" align="center" valign="top">5.1 (4.0 to 6.2)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top"> Medium excision (10-14 mm deep)</td><td colspan="1" rowspan="1" align="center" valign="top">137 (26.9)</td><td colspan="1" rowspan="1" align="center" valign="top">186 (22.4)</td><td colspan="1" rowspan="1" align="center" valign="top">1.48 (1.09 to 2.01)</td><td colspan="1" rowspan="1" align="center" valign="top">7.5 (5.9 to 9.2)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top"> Large excision (15-19 mm deep)</td><td colspan="1" rowspan="1" align="center" valign="top">57 (11.2)</td><td colspan="1" rowspan="1" align="center" valign="top">48 (5.8)</td><td colspan="1" rowspan="1" align="center" valign="top">2.33 (1.52 to 3.58)</td><td colspan="1" rowspan="1" align="center" valign="top">11.9 (7.6 to 16.1)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top"> Very large excision (≥20 mm deep)</td><td colspan="1" rowspan="1" align="center" valign="top">30 (5.9)</td><td colspan="1" rowspan="1" align="center" valign="top">28 (3.4)</td><td colspan="1" rowspan="1" align="center" valign="top">2.27 (1.32 to 3.91)</td><td colspan="1" rowspan="1" align="center" valign="top">11.6 (5.9 to 17.2)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top"> Unknown treatment depth</td><td colspan="1" rowspan="1" align="center" valign="top">47 (9.2)</td><td colspan="1" rowspan="1" align="center" valign="top">71 (8.6)</td><td colspan="1" rowspan="1" align="center" valign="top">1.23 (0.81 to 1.88)</td><td colspan="1" rowspan="1" align="center" valign="top">6.3 (4.0 to 8.5)</td></tr><tr><td colspan="5" rowspan="1" align="left" valign="top" content-type="TableSubHead"><bold>All very/extreme preterm births (20-31 weeks)</bold></td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">Punch biopsy before birth</td><td colspan="1" rowspan="1" align="center" valign="top">35 (22)</td><td colspan="1" rowspan="1" align="center" valign="top">274 (33.0)</td><td colspan="1" rowspan="1" align="center" valign="top">0.63 (0.38 to1.06)</td><td colspan="1" rowspan="1" align="center" valign="top">1.3 (0.8 to 1.7)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">Treatment before birth:</td><td colspan="1" rowspan="1" align="center" valign="top"/><td colspan="1" rowspan="1" align="center" valign="top"/><td colspan="1" rowspan="1" align="center" valign="top"/><td colspan="1" rowspan="1" align="center" valign="top"/></tr><tr><td colspan="1" rowspan="1" align="left" valign="top"> Small excision (1-9 mm deep )</td><td colspan="1" rowspan="1" align="center" valign="top">44 (27)</td><td colspan="1" rowspan="1" align="center" valign="top">223 (26.9)</td><td colspan="1" rowspan="1" align="center" valign="top">1 (reference)</td><td colspan="1" rowspan="1" align="center" valign="top">2.0 (1.3 to 2.7)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top"> Medium excision (10-14 mm deep)</td><td colspan="1" rowspan="1" align="center" valign="top">38 (24)</td><td colspan="1" rowspan="1" align="center" valign="top">186 (22.4)</td><td colspan="1" rowspan="1" align="center" valign="top">1.12 (0.70 to 1.82)</td><td colspan="1" rowspan="1" align="center" valign="top">2.3 (1.4 to 3.1)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top"> Large excision (15-19 mm deep)</td><td colspan="1" rowspan="1" align="center" valign="top">17 (11)</td><td colspan="1" rowspan="1" align="center" valign="top">48 (5.8)</td><td colspan="1" rowspan="1" align="center" valign="top">1.80 (0.91 to 3.54)</td><td colspan="1" rowspan="1" align="center" valign="top">3.6 (1.5 to 5.7)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top"> Very large excision (≥20 mm deep)</td><td colspan="1" rowspan="1" align="center" valign="top">17 (11)</td><td colspan="1" rowspan="1" align="center" valign="top">28 (3.4)</td><td colspan="1" rowspan="1" align="center" valign="top">3.17 (1.56 to 6.47)</td><td colspan="1" rowspan="1" align="center" valign="top">6.4 (2.4 to 10.3)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top"> Unknown treatment depth</td><td colspan="1" rowspan="1" align="center" valign="top">10 (6)</td><td colspan="1" rowspan="1" align="center" valign="top">71 (8.6)</td><td colspan="1" rowspan="1" align="center" valign="top">0.72 (0.34 to 1.55)</td><td colspan="1" rowspan="1" align="center" valign="top">1.4 (0.4 to 2.4)</td></tr></tbody></table><table-wrap-foot><p>Adjusted for parity, index of multiple deprivation, maternal age at delivery, and study site.</p></table-wrap-foot></table-wrap><fig id="fig2" position="float"><caption><p><bold>Fig 2</bold> Sensitivity analyses and sub-analyses. Solid squares: relative risk of preterm birth for women with small excisions (<10 mm) relative to punch biopsy only. Open squares: relative risk of preterm birth with increasing depth of excision, excluding women with unknown depth. LLETZ=large loop excision of the transformation zone</p></caption><graphic xlink:href="casa019275.f2_default"/></fig><p>Results restricting the sample to women with spontaneous onset of labour recorded in Hospital Episode Statistics (n=1075) showed a similar association (P for trend <0.001). The risk of a spontaneous preterm birth was significantly increased for depths of 10-14 mm (relative risk 1.48, 1.09 to 2.01), 15-19 mm (2.33, 1.52 to 3.58), and 20 mm or more (2.27, 1.32 to 3.91) compared with women with small excisions (table 2<xref ref-type="table" rid="tbl2"/>).</p><p>We found no difference in risk of preterm birth by type of excisional treatment once results were adjusted for depth of excision (relative risk for large loop excision of the transformation zone compared with other excisional treatment 0.94, 0.56 to 1.59). We observed a twofold increase in the risk of preterm delivery for excisions that were 15 mm or greater, for both women treated by large loop excision of the transformation zone and those receiving other excisional treatments (supplementary table B).</p><p>We obtained similar results with a somewhat stronger association for all very/extreme preterm births (table 2<xref ref-type="table" rid="tbl2"/>). Compared with small excisions, those that were 20 mm or more deep carried a threefold relative risk (3.17, 1.56 to 6.47) of a very preterm birth, corresponding to an absolute risk of 6.4%.</p><p>Results were robust to a series of sensitivity and sub-analyses. Excluding 59 births with an improbable birth weight or 92 women who had a diagnosis of CIN2/CIN3 but only ever had punch biopsies (fig 2<xref ref-type="fig" rid="fig2"/>) did not appreciably alter the results. Considering births after samples that were piecemeal and those that were excised in one piece separately made the relative risk for large/very large excisions slightly smaller but still statistically significant (fig 2<xref ref-type="fig" rid="fig2"/>; supplementary table C). Trends with increasing depth of excision remained significant in all sub-groups and sensitivity analyses (fig 2<xref ref-type="fig" rid="fig2"/>).</p><p>The strength of association was similar when we considered volume of tissue excised (table 3<xref ref-type="table" rid="tbl3"/>). In particular, the relative risk comparing small volume with punch biopsy was 1.03 (0.79 to 1.33). We compared total volume of tissue removed with depth of excision to assess whether extra information could be gained by combining these measures (supplementary table D). We observed a trend of increasing risk of preterm delivery with increasing volume in women with medium depth (10-14 mm) of excision (P=0.004) and a non-significant trend in those with <10 mm depth of excision (P=0.583). We therefore combined volume and depth to classify risk (table 4<xref ref-type="table" rid="tbl4"/>). Compared with small depth and volume, excisions deeper than 14 mm or with a total volume greater than 2.65 cm<sup>3</sup> carried double the risk of preterm birth (relative risk 1.97, 1.43 to 2.72).</p><table-wrap id="tbl3" position="float"><label>Table 3</label><caption><p> Adjusted relative and absolute risk of preterm birth by volume of tissue excised</p></caption><table frame="hsides" rules="groups"><thead><tr><th colspan="1" rowspan="1" align="left" valign="bottom">Volume of excisional treatment</th><th colspan="1" rowspan="1" align="center" valign="bottom">No (%) cases</th><th colspan="1" rowspan="1" align="center" valign="bottom">No (%) controls</th><th colspan="1" rowspan="1" align="center" valign="bottom">Relative risk (95% CI)</th><th colspan="1" rowspan="1" align="center" valign="bottom">Absolute risk (%)</th></tr></thead><tbody><tr><td colspan="1" rowspan="1" align="left" valign="top">Punch biopsy </td><td colspan="1" rowspan="1" align="center" valign="top">210 (27.3)</td><td colspan="1" rowspan="1" align="center" valign="top">274 (33.0)</td><td colspan="1" rowspan="1" align="center" valign="top">0.97 (0.75 to 1.27)</td><td colspan="1" rowspan="1" align="center" valign="top">7.3</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">Small volume (≤1.77 cm<sup>3</sup>)</td><td colspan="1" rowspan="1" align="center" valign="top">229 (29.8)</td><td colspan="1" rowspan="1" align="center" valign="top">297 (35.8)</td><td colspan="1" rowspan="1" align="center" valign="top">1 (reference)</td><td colspan="1" rowspan="1" align="center" valign="top">7.4</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">Medium volume (1.78-2.65 cm<sup>3</sup>)</td><td colspan="1" rowspan="1" align="center" valign="top">95 (12.4)</td><td colspan="1" rowspan="1" align="center" valign="top">90 (10.8)</td><td colspan="1" rowspan="1" align="center" valign="top">1.40 (1.02 to 1.93)</td><td colspan="1" rowspan="1" align="center" valign="top">10.4</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">Large volume (≥2.66 cm<sup>3</sup>)</td><td colspan="1" rowspan="1" align="center" valign="top">165 (21.5)</td><td colspan="1" rowspan="1" align="center" valign="top">98 (11.8)</td><td colspan="1" rowspan="1" align="center" valign="top">2.07 (1.56 to 2.76)</td><td colspan="1" rowspan="1" align="center" valign="top">15.4</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">Unknown volume</td><td colspan="1" rowspan="1" align="center" valign="top">69 (9.0)</td><td colspan="1" rowspan="1" align="center" valign="top">71 (8.6)</td><td colspan="1" rowspan="1" align="center" valign="top">1.26 (0.88 to 1.79)</td><td colspan="1" rowspan="1" align="center" valign="top">9.4</td></tr></tbody></table><table-wrap-foot><p>Adjusted for parity, index of multiple deprivation, maternal age at delivery, and study site.</p></table-wrap-foot></table-wrap><table-wrap id="tbl4" position="float"><label>Table 4</label><caption><p> Relative risk of preterm birth by risk group in women with treatment before birth</p></caption><table frame="hsides" rules="groups"><thead><tr><th colspan="1" rowspan="1" align="left" valign="bottom">Risk group (largest category for volume or depth)</th><th colspan="1" rowspan="1" align="center" valign="bottom">No (%) cases</th><th colspan="1" rowspan="1" align="center" valign="bottom">No (%) controls</th><th colspan="1" rowspan="1" align="center" valign="bottom">Relative risk (95% CI)</th></tr></thead><tbody><tr><td colspan="1" rowspan="1" align="left" valign="top">Small depth AND volume (<10 mm and ≤1.77 cm<sup>3</sup>)</td><td colspan="1" rowspan="1" align="center" valign="top">142 (25.4)</td><td colspan="1" rowspan="1" align="center" valign="top">192 (34.5)</td><td colspan="1" rowspan="1" align="center" valign="top">1 (reference)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">Medium depth OR volume (10-14 mm or 1.78-2.65 cm<sup>3</sup>)</td><td colspan="1" rowspan="1" align="center" valign="top">144 (25.8)</td><td colspan="1" rowspan="1" align="center" valign="top">170 (30.6)</td><td colspan="1" rowspan="1" align="center" valign="top">1.15 (0.85 to 1.55)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">Large depth OR volume (15-20 mm or ≥2.66 cm<sup>3</sup>)</td><td colspan="1" rowspan="1" align="center" valign="top">149 (26.7)</td><td colspan="1" rowspan="1" align="center" valign="top">95 (17.1)</td><td colspan="1" rowspan="1" align="center" valign="top">1.97 (1.43 to 2.72)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">Very large depth (≥20 mm)</td><td colspan="1" rowspan="1" align="center" valign="top">54 (9.7)</td><td colspan="1" rowspan="1" align="center" valign="top">28 (5.0)</td><td colspan="1" rowspan="1" align="center" valign="top">2.44 (1.52 to 3.93)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">Unknown depth AND volume </td><td colspan="1" rowspan="1" align="center" valign="top">69 (12.4)</td><td colspan="1" rowspan="1" align="center" valign="top">71 (12.8)</td><td colspan="1" rowspan="1" align="center" valign="top">1.30 (0.89 to 1.91)</td></tr></tbody></table><table-wrap-foot><p>Adjusted for parity, index of multiple deprivation, maternal age at delivery, and study site.</p></table-wrap-foot></table-wrap><p>Among women with large excisions, the risk of preterm birth was not less in those who conceived three years or more after treatment (table 5<xref ref-type="table" rid="tbl5"/>). We found no significant excess risk of preterm birth for women with small and medium excisions (table 5<xref ref-type="table" rid="tbl5"/>). However, insufficient data were available on women who conceived within six months of small or medium excisions for us to conclude that they are not at increased risk of preterm delivery (relative risk 1.37, 0.74 to 2.53).</p><table-wrap id="tbl5" position="float"><label>Table 5</label><caption><p> Relative and absolute risk of preterm birth by time from last excisional treatment to conception</p></caption><table frame="hsides" rules="groups"><thead><tr><th colspan="1" rowspan="1" align="left" valign="bottom">Time from last excisional treatment to conception (years)</th><th colspan="1" rowspan="1" align="center" valign="bottom">No (%) cases</th><th colspan="1" rowspan="1" align="center" valign="bottom">No (%) controls</th><th colspan="1" rowspan="1" align="center" valign="bottom">Relative risk (95% CI)</th><th colspan="1" rowspan="1" align="center" valign="bottom">Absolute risk (%)</th></tr></thead><tbody><tr><td colspan="5" rowspan="1" align="left" valign="top" content-type="TableSubHead"><bold>Women with large or very large excision (depth or volume)</bold></td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">0 to 0.5</td><td colspan="1" rowspan="1" align="center" valign="top">13 (7)</td><td colspan="1" rowspan="1" align="center" valign="top">10 (8)</td><td colspan="1" rowspan="1" align="center" valign="top">0.88 (0.35 to 2.24)</td><td colspan="1" rowspan="1" align="center" valign="top">13.5 (1.7 to 25.3)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">0.5 to 1</td><td colspan="1" rowspan="1" align="center" valign="top">12 (6)</td><td colspan="1" rowspan="1" align="center" valign="top">13 (11)</td><td colspan="1" rowspan="1" align="center" valign="top">0.62 (0.26 to 1.47)</td><td colspan="1" rowspan="1" align="center" valign="top">9.4 (1.9 to 17.0)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">1 to 3</td><td colspan="1" rowspan="1" align="center" valign="top">52 (26)</td><td colspan="1" rowspan="1" align="center" valign="top">27 (22)</td><td colspan="1" rowspan="1" align="center" valign="top">1.04 (0.58 to 1.84)</td><td colspan="1" rowspan="1" align="center" valign="top">15.8 (8.6 to 23.1)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">≥3</td><td colspan="1" rowspan="1" align="center" valign="top">123 (62)</td><td colspan="1" rowspan="1" align="center" valign="top">72 (59)</td><td colspan="1" rowspan="1" align="center" valign="top">1 (reference)</td><td colspan="1" rowspan="1" align="center" valign="top">15.3 (11.0 to 19.5)</td></tr><tr><td colspan="5" rowspan="1" align="left" valign="top" content-type="TableSubHead"><bold>Women with small, medium, or unknown excision (depth or volume)</bold></td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">0 to 0.5</td><td colspan="1" rowspan="1" align="center" valign="top">27 (8)</td><td colspan="1" rowspan="1" align="center" valign="top">25 (6)</td><td colspan="1" rowspan="1" align="center" valign="top">1.37 (0.74 to 2.53)</td><td colspan="1" rowspan="1" align="center" valign="top">10.2 (4.3 to 16.2)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">0.5 to 1</td><td colspan="1" rowspan="1" align="center" valign="top">28 (8)</td><td colspan="1" rowspan="1" align="center" valign="top">38 (9)</td><td colspan="1" rowspan="1" align="center" valign="top">0.84 (0.50 to 1.43)</td><td colspan="1" rowspan="1" align="center" valign="top">6.3 (3.2 to 9.5)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">1 to 3</td><td colspan="1" rowspan="1" align="center" valign="top">100 (28)</td><td colspan="1" rowspan="1" align="center" valign="top">108 (25)</td><td colspan="1" rowspan="1" align="center" valign="top">1.17 (0.84 to 1.63)</td><td colspan="1" rowspan="1" align="center" valign="top">8.8 (6.4 to 11.1)</td></tr><tr><td colspan="1" rowspan="1" align="left" valign="top">≥3</td><td colspan="1" rowspan="1" align="center" valign="top">196 (56)</td><td colspan="1" rowspan="1" align="center" valign="top">258 (60)</td><td colspan="1" rowspan="1" align="center" valign="top">1 (reference)</td><td colspan="1" rowspan="1" align="center" valign="top">7.5 (6.2 to 8.9)</td></tr></tbody></table><table-wrap-foot><p>*Adjusted for parity, index of multiple deprivation, maternal age at delivery, and study site.</p></table-wrap-foot></table-wrap></sec><sec sec-type="discussion"><title>Discussion</title><p>Although the overall risk associated with excision is minimal, the excess risk seen in women with deeper excisions is substantial. The risk of preterm birth after colposcopy among women who had only a diagnostic punch biopsy and those who had a small (<10 mm) depth of tissue removed was similar (absolute risk 1 in 13). This risk doubled to 1 in 6 when either the depth of the excision or the volume of tissue excised was large. In absolute terms, the risk was 15.3% in women with large excisions and 7.2% in those with a punch biopsy before birth compared with 6.7% in the general population in England.<xref rid="ref10" ref-type="bibr">10</xref> We found a stronger association for very preterm births: 6.4% of births in women with very large excisions were very preterm compared with 2% in those with small excisions and 1.4% in the general population.<xref rid="ref14" ref-type="bibr">14</xref> Multiple treatments did not themselves increase the risk of preterm birth; the increased risk in women with multiple excisions seems to be due to the large total depth excised. The increased risk was still apparent three years after treatment and did not seem to change over time.</p><sec><title>Strengths and limitations of study</title><p>Owing to the nature of data collection in phase I, some colposcopy dates and procedures were inevitably misclassified. The detailed colposcopy records collected for phase II make such misclassification unlikely. We had very limited information on potential confounding factors such as smoking and ethnicity. Although ethnicity may be related to the risk of having cervical disease, it is unlikely to influence the depth of excision. We have near complete (82%) smoking information (at the time of colposcopy) for just one of the 12 sites in the study. Analysis of these data suggests no association between smoking and depth of excision, but the numbers are small (143 ever smokers and 86 never smokers) and generalisability is limited. Smoking is unlikely to be able to explain the dose-response relation observed between depth of excision and risk of preterm birth, but we cannot exclude it as a confounder.</p><p>Although detailed information on colposcopy was collected by each site and cross referenced with Hospital Episode Statistics records, we had no information on treatments from before 1995 and those at other NHS trusts. Some women in the punch biopsy group may also have received ablative treatment, as only 24 ablative treatments (almost all of which came from one site) were recorded; however, ablation is now rare in England. Before the introduction of large loop excision of the transformation zone in the early 1990s, the preferred treatment was laser ablation (59% of treatments).<xref rid="ref15" ref-type="bibr">15</xref> By 1993, after the introduction of national colposcopy guidelines, 61% of treatments were large loop excision of the transformation zone and only 18% were laser ablation.<xref rid="ref15" ref-type="bibr">15</xref> In 2011-12 just 0.7% of women had ablation at first colposcopy compared with 44% having a punch biopsy and 15% having an excision.<xref rid="ref16" ref-type="bibr">16</xref> Furthermore, the literature suggests no increased risk of preterm delivery associated with ablative treatment,<xref rid="ref2" ref-type="bibr">2</xref> so inclusion in the punch biopsy group would not have substantially changed the results.</p><p>Our results rely on birth data submitted to Hospital Episode Statistics, and the quality of these data has been questioned. A quarter of births identified were from hospitals that did not participate in our study, showing that we succeeded in identifying such births. However, some (17%) births in Hospital Episode Statistics do not have gestational age recorded; this could affect the absolute preterm rate. Some high risk pregnancies are likely to have remained in the study, but results were little changed either by inclusion of those we know about or by restricting analysis to births after spontaneous onset of labour. However, Hospital Episode Statistics data are routinely recorded for administrative purposes and are not used as a medical record for management of patients. Misclassification of whether a particular birth was spontaneous will tend to reduce the relative risk but should not affect the excess risk.</p><p>The dimensions of the excision were missing for 7% of participants and were difficult to determine in those with piecemeal excision. However, we saw no indication that using the largest fragment depth when the excision was piecemeal significantly underestimated the effect. A similar problem exists for women treated more than once. Summing the depths of each treatment excludes the possibility of regeneration. It has been suggested that the greater amount of tissue excised the lower the volume regenerated,<xref rid="ref17" ref-type="bibr">17</xref> but the literature also suggests that women treated more than once are at particularly high risk.<xref rid="ref7" ref-type="bibr">7</xref> <xref rid="ref18" ref-type="bibr">18</xref> <xref rid="ref19" ref-type="bibr">19</xref></p></sec><sec><title>Comparison with other studies</title><p>This study has more preterm deliveries with information on size of previous excision (n=489) than all previous studies combined. A meta-analysis looking at depth included 107 preterm births from three studies,<xref rid="ref5" ref-type="bibr">5</xref> <xref rid="ref18" ref-type="bibr">18</xref> <xref rid="ref20" ref-type="bibr">20</xref> but it did not explicitly consider the trend with increasing depth.<xref rid="ref2" ref-type="bibr">2</xref> Compared with women who had not attended colposcopy, the relative risk in those with a depth greater than 10 mm was 2.61 (1.28 to 5.34); it was 1.45 (0.55 to 3.86) in those with a depth less than 11 mm. Since this meta-analysis, four further studies (with 318 preterm deliveries) have looked at the depth of excision.<xref rid="ref4" ref-type="bibr">4</xref> <xref rid="ref19" ref-type="bibr">19</xref> <xref rid="ref21" ref-type="bibr">21</xref> <xref rid="ref22" ref-type="bibr">22</xref> The largest found an increased risk of preterm birth associated with treatment of 20 mm or more compared with less than 10 mm (odds ratio 1.79, 1.23 to 2.60).<xref rid="ref4" ref-type="bibr">4</xref> In keeping with the results from our study, almost all published studies concur that large excisions (depth greater than 15 mm) increase the risk of preterm birth. However, ours is the first study to show clinical equivalence in terms of subsequent risk of preterm delivery between small excision (<10 mm) and punch biopsy. Although we did not use an a priori definition of equivalence in this study, the upper limit of the 95% confidence interval is 1.37 (for small depth or 1.33 for small volume), compared with 3.86 in the meta-analysis, and none of the subsequent studies included an untreated comparator.</p><p>NHS colposcopy is carried out by clinicians certified by the British Society for Colposcopy and Cervical Pathology (BSCCP). Most will have trained under the joint BSCCP/Royal College of Obstetrics and Gynaecology programme. This includes a defined number of supervised colposcopies, a module in cytopathology and histopathology, and an exit examination. Three yearly recertification including an audit of activity and confirmation of attendance at a national educational conference is required. Each unit is visited every three years, and the activity and outcomes are examined against defined standards. Colposcopists must see a minimum of 50 women each year. These measures ensure homogenously high standards. This is reflected in three quarters of treated women in the study having a depth of excision less than 15 mm.</p><p>We observed no diminution of risk with increasing time from treatment to conception. Literature on this topic is sparse, but one recent paper supports this result.<xref rid="ref21" ref-type="bibr">21</xref> Although the relative risk of a preterm birth in treated women overall depends on the quality of colposcopy, relative risks associated with increasing size of excision, multiple excisions, and time to conception are generalisable to other colposcopic settings worldwide.</p></sec><sec><title>Conclusion</title><p>The risk of preterm birth in women who undergo loop excision of less than 10 mm in depth and 1.77 cm<sup>3</sup> in volume is similar to that in women who have a diagnostic punch biopsy only. Larger excisions, particularly over 15 mm or 2.66 cm<sup>3</sup>, are associated with a doubling of the risk of both preterm and very preterm births. Efforts should be made to excise the entire lesion while preserving as much healthy cervical tissue as possible. Multiple treatments in themselves did not increase the risk of preterm birth; the increased risk was largely due to the large total depth in women who have had multiple excisions. Colposcopic treatment guidelines should be updated accordingly.</p><boxed-text position="float" content-type="style4"><sec><title>What is already known on this topic</title><list list-type="simple"><list-item><p>Most studies of preterm delivery after large loop excision of the transformation zone found that treatment was associated with increased risk</p></list-item><list-item><p>A meta-analysis (including three studies) found an increased risk of preterm birth when the depth of the excised tissue exceeded 10 mm; a large study from Denmark found an increased risk of preterm birth for those with treatments of ≥20 mm</p></list-item><list-item><p>The precise role of increasing depth of excision and whether a safe depth exists below which there is no increased risk are unclear</p></list-item></list></sec><sec><title>What this study adds</title><list list-type="simple"><list-item><p>One in six births in women who had previously had a large (≥15 mm or ≥2.66 cm<sup>3</sup>) excisional procedure at colposcopy were preterm</p></list-item><list-item><p>The risk of a preterm birth among women with small procedures at colposcopy was similar to the risk among those not treated at colposcopy before birth</p></list-item><list-item><p>The increased risk in women with multiple excisions seems to be due to the large total depth excised rather than multiple treatments in themselves</p></list-item></list></sec></boxed-text></sec></sec>
Modelling cholera in periodic environments	<p>We propose a deterministic compartmental model for cholera dynamics in periodic environments. The model incorporates seasonal variation into a general formulation for the incidence (or, force of infection) and the pathogen concentration. The basic reproduction number of the periodic model is derived, based on which a careful analysis is conducted on the epidemic and endemic dynamics of cholera. Several specific examples are presented to demonstrate this general model, and numerical simulation results are used to validate the analytical prediction.</p>	<contrib contrib-type="author"><name><surname>Posny</surname><given-names>Drew</given-names></name><xref ref-type="aff" rid="AF1"> <sup>a</sup> </xref></contrib><contrib contrib-type="author"><name><surname>Wang</surname><given-names>Jin</given-names></name><xref ref-type="aff" rid="AF1"> <sup>a</sup> </xref><xref ref-type="corresp" rid="cor1"> <sup>*</sup> </xref></contrib><aff id="AF1"><label><sup>a</sup></label><institution><named-content content-type="department">Department of Mathematics and Statistics</named-content>, <named-content content-type="institution-name">Old Dominion University</named-content></institution>, <named-content content-type="city">Norfolk</named-content>, <named-content content-type="state">VA</named-content><named-content content-type="postal-code">23529</named-content>, <country>USA</country></aff>	Journal of Biological Dynamics	<sec sec-type="intro" id="S001"><label>1. </label><title>Introduction</title><p>Limited access to safe water and sanitation resources is common in developing countries, leaving them vulnerable to cholera outbreaks. Cholera is an intestinal infection caused by ingesting food or water contaminated with the bacterium <italic>Vibrio cholerae</italic>. If left untreated, an infected individual may become severely dehydrated and die within several days. In addition to prompt rehydration and medical treatment, proper sanitation facilities are needed to prevent infected individuals from shedding the bacteria back into the environment further fuelling the pathogen concentration and the persistence of the disease. Besides the transmission route based on environment–human interaction, the human-to-human direct transmission is also found important in shaping a cholera epidemic. A recent cholera outbreak in Zimbabwe, a land-locked country in Africa, during 2008–2009 underscores such a direct transmission pathway [<xref rid="CIT0012" ref-type="bibr">12</xref>].</p><p>Numerous mathematical models have been published to analyse cholera outbreaks in an effort to better understand the complex disease transmission and determine adequate prevention and effective control strategies (see, for example, [<xref rid="CIT0006" ref-type="bibr">6</xref>,<xref rid="CIT0007" ref-type="bibr">7</xref>,<xref rid="CIT0009" ref-type="bibr">9</xref>,<xref rid="CIT0011" ref-type="bibr">11</xref>,<xref rid="CIT0012" ref-type="bibr">12</xref>,<xref rid="CIT0016" ref-type="bibr">16</xref>,<xref rid="CIT0017" ref-type="bibr">17</xref>,<xref rid="CIT0019" ref-type="bibr">19</xref>]). In particular, Wang and Liao [<xref rid="CIT0019" ref-type="bibr">19</xref>] recently proposed a deterministic cholera model that incorporates general incidence and pathogen functions and that can unify many of the existing cholera models. These studies have certainly produced many useful results and have improved our understanding of cholera dynamics. One limitation of these models, however, is that most of them assumed that the model parameters are constant in time, meaning that the disease contact rate, recovery rate, pathogen growth rate, etc., all take fixed values independent of time. An exception, we note, is the work in [<xref rid="CIT0007" ref-type="bibr">7</xref>] where, in addition to the main discussion on the autonomous cholera model, the author also conducted simple numerical tests to three scenarios with periodic coefficients. From the mathematical point of view, the constant parameter assumption has the advantage of simplifying the models and analysis, and facilitating the use of some well-known theory in autonomous dynamical systems.</p><p>On the other hand, environmental concerns, such as floods, droughts, temperatures and other climatic factors, are seasonal and could significantly affect cholera dynamics. For example, it has been observed that cholera is a seasonal disease in many endemic places and infection peaks often occur annually in the rainy or monsoon season [<xref rid="CIT0010" ref-type="bibr">10</xref>,<xref rid="CIT0018" ref-type="bibr">18</xref>]. Such filed observations underline the limitation of most (if not all) current mathematical cholera models and imply that mathematical insights into cholera seasonality has largely lagged behind. It is thus important for mathematical cholera studies to incorporate these seasonal factors to gain deeper quantitative understanding of the short- and long-term evolution of cholera dynamics, and to better predict and prevent future cholera outbreaks.</p><p>The objective of this paper is to propose a general cholera model in a periodic environment by extending the model proposed in [<xref rid="CIT0019" ref-type="bibr">19</xref>] to include seasonal variations in the environment and the disease transmission pathways. In particular, the incidence (or, force of infection) and the rate of change for the pathogen concentration are subject to periodicity. Using the framework introduced in [<xref rid="CIT0020" ref-type="bibr">20</xref>], we will analyse the basic reproduction number, <italic>R</italic> <sub>0</sub>, for this cholera model and establish that <italic>R</italic> <sub>0</sub> is a sharp threshold for cholera dynamics in periodic environments: when <italic>R</italic> <sub>0</sub><1, the disease-free equilibrium (DFE) is globally asymptotically stable, and the disease completely dies out; when <italic>R</italic> <sub>0</sub>>1, the system admits a positive periodic solution, and the disease is uniformly persistent. We mention that extinction and persistence results for some periodic epidemic systems are also discussed in [<xref rid="CIT0004" ref-type="bibr">4</xref>,<xref rid="CIT0005" ref-type="bibr">5</xref>,<xref rid="CIT0015" ref-type="bibr">15</xref>,<xref rid="CIT0022" ref-type="bibr">22</xref>].</p><p>The remainder of the paper is organized as follows. In Section 2, we introduce the periodic cholera model and state the necessary assumptions. In Section 3, the basic reproduction number is derived, followed by a global stability analysis of the disease-free equilibrium in Section 4. The existence and uniform persistence of an endemic periodic solution are analysed in Section 5. We then briefly study several specific cholera models in Section 6. Finally, conclusions are drawn in Section 7.</p></sec><sec id="S002"><label>2. </label><title>Mathematical model</title><p>Building on the cholera model in [<xref rid="CIT0019" ref-type="bibr">19</xref>], we construct the following non-autonomous dynamical system to describe cholera dynamics in a periodic environment: <disp-formula-group id="M0001"><disp-formula><graphic xlink:href="tjbd-8-001-e001.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0002"><disp-formula><graphic xlink:href="tjbd-8-001-e002.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0003"><disp-formula><graphic xlink:href="tjbd-8-001-e003.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0004"><disp-formula><graphic xlink:href="tjbd-8-001-e004.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <italic>S, I, R</italic> and <italic>B</italic> denote the susceptible population, infected population, recovered population and the concentration of vibrios in the contaminated water, respectively. The total population <italic>N</italic>=<italic>S</italic>+<italic>I</italic>+<italic>R</italic> is assumed to be a constant for all <italic>t</italic>≥0. The parameter <italic>b</italic> represents the natural human birth/death rate, and γ represents the rate of recovery from cholera. In this general model, the incidence function <italic>f</italic>(<italic>t, I, B</italic>) which determines the rate of new infection and the function <italic>h</italic>(<italic>t, I, B</italic>) which describes the rate of change for the pathogen in the environment are both differentiable and periodic in time with a common period ω. That is, <disp-formula id="UM0001"><graphic xlink:href="tjbd-8-001-u001.jpg" position="float" orientation="portrait"/></disp-formula> To make biological sense, we assume that the functions <italic>f</italic> and <italic>h</italic> satisfy the following conditions for all <italic>t</italic>≥0: <list list-type="simple"><list-item><p>(A1) <inline-formula><inline-graphic xlink:href="tjbd-8-001-m001.jpg"/></inline-formula>.</p></list-item><list-item><p> <inline-formula><inline-graphic xlink:href="tjbd-8-001-m002.jpg"/></inline-formula>.</p></list-item><list-item><p> <inline-formula><inline-graphic xlink:href="tjbd-8-001-m003.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m004.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m005.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m006.jpg"/></inline-formula>.</p></list-item><list-item><p> <italic>f</italic>(<italic>t, I, B</italic>) and <italic>h</italic>(<italic>t, I, B</italic>) are both concave for any <italic>t</italic>≥0; i.e. the matrices <disp-formula id="UM0002"><graphic xlink:href="tjbd-8-001-u002.jpg" position="float" orientation="portrait"/></disp-formula> are negative semidefinite everywhere.</p></list-item></list> </p><p>The assumption (A1) ensures that the model has a unique, constant disease-free equilibrium (DFE) <disp-formula-group id="M0005"><disp-formula><graphic xlink:href="tjbd-8-001-e005.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> The assumption (A2) ensures a non-negative force of infection. The first two inequalities in (A3) state that the rate of new infection increases with both the infected population size and the pathogen concentration, and the third inequality states that increased human infection and, consequently, higher level of human contribution to the environmental vibrios, lead to higher growth rate for the pathogen. The last inequality in (A3) is based on experimental observation that the vibrios cannot sustain themselves in the environment in the absence of human contribution [<xref rid="CIT0013" ref-type="bibr">13</xref>]; in other words, without the contribution from infected human population, the rate of change of the pathogen concentration would be negatively related to itself. The condition (A4) is based on saturation effect, a common assumption in epidemic models [<xref rid="CIT0019" ref-type="bibr">19</xref>].</p><p>In addition, we assume that</p><list list-type="simple"><list-item><p>(A5) <inline-formula><inline-graphic xlink:href="tjbd-8-001-m007.jpg"/></inline-formula> </p></list-item></list><p>The first condition in (A5) implies that infection can start by the indirect transmission route alone; in other words, a positive bacterial concentration can lead to a positive incidence even if <italic>I</italic>=0 initially. The second condition in (A5) states that infected people will contribute to the growth of the vibrios in the environment (e.g. by shedding) even if <italic>B</italic>=0 initially.</p><p>Furthermore, we introduce an additional regulation on the profiles of the incidence and pathogen functions for small <italic>I</italic> and <italic>B</italic>. We assume that</p><list list-type="simple"><list-item><p>(A6) There exists <inline-formula><inline-graphic xlink:href="tjbd-8-001-m008.jpg"/></inline-formula> such that when <inline-formula><inline-graphic xlink:href="tjbd-8-001-m009.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m010.jpg"/></inline-formula>, <disp-formula id="UM0003"><graphic xlink:href="tjbd-8-001-u003.jpg" position="float" orientation="portrait"/></disp-formula> and <disp-formula id="UM0004"><graphic xlink:href="tjbd-8-001-u004.jpg" position="float" orientation="portrait"/></disp-formula> </p></list-item></list><p>Here we make some comments on the assumption (A6). Based on the concavity of <italic>f</italic> (assumption A4), the surface of <italic>f</italic> is below its tangent plane everywhere. Meanwhile, since the matrix <italic>D</italic> <sup>2</sup> <italic>f</italic> is negative semidefinite, we have <disp-formula id="UM0005"><graphic xlink:href="tjbd-8-001-u005.jpg" position="float" orientation="portrait"/></disp-formula> Thus, assumption (A6) essentially states that at least in a small neighbourhood of <italic>I</italic>=<italic>B</italic>=0, the surface of <italic>f</italic> lies below its tangent plane and above a concave tangent paraboloid. Similar reasoning holds for <italic>h</italic>.</p><p>Finally, we mention that many well-known cholera models, such as those in [<xref rid="CIT0007" ref-type="bibr">7</xref>,<xref rid="CIT0009" ref-type="bibr">9</xref>,<xref rid="CIT0012" ref-type="bibr">12</xref>,<xref rid="CIT0017" ref-type="bibr">17</xref>], all satisfy the above assumptions (A1)–(A6), though these models are based on autonomous dynamical systems. For example, the model in [<xref rid="CIT0012" ref-type="bibr">12</xref>] has <inline-formula><inline-graphic xlink:href="tjbd-8-001-m011.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-001-m012.jpg"/></inline-formula>. It is straightforward to verify that (A1)–(A6) hold; in particular, expanding <italic>f</italic>(<italic>I, B</italic>) at (0, 0) to second order yields <inline-formula><inline-graphic xlink:href="tjbd-8-001-m013.jpg"/></inline-formula>, and it can be readily seen that <italic>f</italic>(<italic>I, B</italic>) satisfies (A6) as <inline-formula><inline-graphic xlink:href="tjbd-8-001-m014.jpg"/></inline-formula> for all <italic>B</italic>>0. Similar verification can be done for the model in [<xref rid="CIT0007" ref-type="bibr">7</xref>], where <inline-formula><inline-graphic xlink:href="tjbd-8-001-m015.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m016.jpg"/></inline-formula>, and the model in [<xref rid="CIT0017" ref-type="bibr">17</xref>], where <inline-formula><inline-graphic xlink:href="tjbd-8-001-m017.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m018.jpg"/></inline-formula>. We will discuss in detail these models with periodic parameters in Section 6.</p></sec><sec id="S003"><label>3. </label><title>Basic reproduction number</title><p>A fundamental concept in epidemiology is the basic reproduction number, which measures the average number of secondary infections that occur when one infective is introduced into a completely susceptible host population. Following the standard next-generation matrix theory [<xref rid="CIT0008" ref-type="bibr">8</xref>], we consider the subsystem of model (1)–(4) that is directly related to the infection: <disp-formula id="UM0006"><graphic xlink:href="tjbd-8-001-u006.jpg" position="float" orientation="portrait"/></disp-formula> where <inline-formula><inline-graphic xlink:href="tjbd-8-001-m019.jpg"/></inline-formula> denotes the input rate of new infections and <inline-formula><inline-graphic xlink:href="tjbd-8-001-m020.jpg"/></inline-formula> denotes the rate of transfer of individuals into or out of each population set. The next-generation matrix is defined as <italic>F</italic>(<italic>t</italic>)<italic>V</italic> <sup>−1</sup>(<italic>t</italic>), where <italic>F</italic>(<italic>t</italic>) and <italic>V</italic>(<italic>t</italic>) are the Jacobian matrices given by <disp-formula id="UM0007"><graphic xlink:href="tjbd-8-001-u007.jpg" position="float" orientation="portrait"/></disp-formula> and where <inline-formula><inline-graphic xlink:href="tjbd-8-001-m021.jpg"/></inline-formula> is the disease-free equilibrium of the model defined in Equation (5).</p><p>For a compartmental epidemiological model based on an <italic>autonomous</italic> system, the basic reproduction number is determined by the spectral radius of the next-generation matrix (which is independent of time) [<xref rid="CIT0008" ref-type="bibr">8</xref>]. The definition of the basic reproduction number of a general non-autonomous model system, however, is still an open question. Bacaër and Guernaoui introduced <italic>R</italic> <sub>0</sub> for periodic epidemic models (including ODE and PDE systems) as the spectral radius of an integral operator [<xref rid="CIT0002" ref-type="bibr">2</xref>]; related work for some periodic ODE systems was also discussed in [<xref rid="CIT0001" ref-type="bibr">1</xref>]. In addition, Wang and Zhao [<xref rid="CIT0020" ref-type="bibr">20</xref>] extended the framework in [<xref rid="CIT0008" ref-type="bibr">8</xref>] to include epidemiological models in periodic environments. They introduced the next infection operator <italic>L</italic> by <disp-formula-group id="M0006"><disp-formula><graphic xlink:href="tjbd-8-001-e006.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <italic>Y</italic>(<italic>t, s</italic>), <italic>t</italic>≥<italic>s</italic>, is the evolution operator of the linear ω-periodic system <inline-formula><inline-graphic xlink:href="tjbd-8-001-m022.jpg"/></inline-formula> and φ(<italic>t</italic>), the initial distribution of infectious individuals, is ω-periodic and nonnegative. The basic reproduction number is then defined as the spectral radius of the next infection operator, <disp-formula-group id="M0007"><disp-formula><graphic xlink:href="tjbd-8-001-e007.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> For our cholera model (1)–(4), the evolution operator can be easily determined by solving the system of differential equations <inline-formula><inline-graphic xlink:href="tjbd-8-001-m023.jpg"/></inline-formula> with the initial condition <inline-formula><inline-graphic xlink:href="tjbd-8-001-m024.jpg"/></inline-formula>; thus, <disp-formula-group id="M0008"><disp-formula><graphic xlink:href="tjbd-8-001-e008.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula-group id="M0009"><disp-formula><graphic xlink:href="tjbd-8-001-e009.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> The basic reproduction number defined in Equation (7) can be numerically evaluated by using the methods presented in [<xref rid="CIT0001" ref-type="bibr">1</xref>,<xref rid="CIT0014" ref-type="bibr">14</xref>,<xref rid="CIT0020" ref-type="bibr">20</xref>]. From [<xref rid="CIT0020" ref-type="bibr">20</xref>], we immediately obtain the following result regarding the local stability of the DFE:</p><statement><label>Theorem 1 </label><p>Let <italic>R</italic> <sub>0</sub> be defined as (7). Then the disease-free equilibrium of system (1)–(4) is locally asymptotically stable if <italic>R</italic> <sub>0</sub><1, and unstable if <italic>R</italic> <sub>0</sub>>1.</p></statement></sec><sec id="S004"><label>4. </label><title>Disease extinction</title><p>We proceed to investigate the global stability of the DFE for our cholera model, which will also provide a condition for the extinction of the disease. Consider the matrix function <italic>F</italic>(<italic>t</italic>)−<italic>V</italic>(<italic>t</italic>): <disp-formula-group id="M0010"><disp-formula><graphic xlink:href="tjbd-8-001-e010.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> It can be easily verified that the above matrix function is continuous, cooperative, irreducible and ω-periodic. Let <inline-formula><inline-graphic xlink:href="tjbd-8-001-m025.jpg"/></inline-formula> be the fundamental solution matrix of the linear ordinary differential system: <disp-formula-group id="M0011"><disp-formula><graphic xlink:href="tjbd-8-001-e011.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> and <inline-formula><inline-graphic xlink:href="tjbd-8-001-m026.jpg"/></inline-formula> be the spectral radius of <inline-formula><inline-graphic xlink:href="tjbd-8-001-m027.jpg"/></inline-formula>.</p><p>From Lemma 2.1 in [<xref rid="CIT0022" ref-type="bibr">22</xref>], we immediately obtain the following result:</p><statement><label>lemma 2 </label><p>Let <inline-formula><inline-graphic xlink:href="tjbd-8-001-m028.jpg"/></inline-formula>. Then there exists a positive ω-periodic function <italic>v</italic>(<italic>t</italic>) such that <inline-formula><inline-graphic xlink:href="tjbd-8-001-m029.jpg"/></inline-formula> is a solution to Equation (11).</p></statement><p>Now let us consider Equations (2) and (4) from our cholera model. It can be easily obtained, using assumption (<italic>A</italic>4), that <disp-formula id="UM0008"><graphic xlink:href="tjbd-8-001-u008.jpg" position="float" orientation="portrait"/></disp-formula> and <disp-formula id="UM0009"><graphic xlink:href="tjbd-8-001-u009.jpg" position="float" orientation="portrait"/></disp-formula> That is, <disp-formula-group id="M0012"><disp-formula><graphic xlink:href="tjbd-8-001-e012.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Meanwhile, based on Lemma 2, there exists <italic>v</italic>(<italic>t</italic>) such that <disp-formula-group id="M0013"><disp-formula><graphic xlink:href="tjbd-8-001-e013.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> is a solution to Equation (11), with <inline-formula><inline-graphic xlink:href="tjbd-8-001-m030.jpg"/></inline-formula>. It follows from Equations (11) and (12) that <disp-formula-group id="M0014"><disp-formula><graphic xlink:href="tjbd-8-001-e014.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> when <italic>t</italic> is large. From [20, Theorem 2.2], it is known that <italic>R</italic> <sub>0</sub><1 if and only if <inline-formula><inline-graphic xlink:href="tjbd-8-001-m031.jpg"/></inline-formula>. Therefore, μ<0. Then, given (13) and (14), it is clear that <disp-formula-group id="M0015"><disp-formula><graphic xlink:href="tjbd-8-001-e015.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>Next, we consider Equation (3) from our model. For any ε>0, there exists <italic>T</italic>>0 such that whenever <italic>t</italic>><italic>T</italic>, we have <disp-formula id="UM0010"><graphic xlink:href="tjbd-8-001-u010.jpg" position="float" orientation="portrait"/></disp-formula> Thus, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m032.jpg"/></inline-formula> for <italic>t</italic>><italic>T</italic>. Since ε>0 is arbitrary, it is clear that <disp-formula-group id="M0016"><disp-formula><graphic xlink:href="tjbd-8-001-e016.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Finally, since the total population <italic>N</italic>=<italic>S</italic>+<italic>I</italic>+<italic>R</italic> is a constant, we have that <disp-formula-group id="M0017"><disp-formula><graphic xlink:href="tjbd-8-001-e017.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Hence, we have established the following result:</p><statement><label>Theorem 3 </label><p>If <italic>R</italic> <sub>0</sub><1, then the disease-free equilibrium of model (1)–(4) is globally asymptotically stable, and <inline-formula><inline-graphic xlink:href="tjbd-8-001-m033.jpg"/></inline-formula> for any solution <italic>x</italic>(<italic>t</italic>) of system (1)–(4).</p></statement><p>Theorem 3 shows that the disease will completely die out as long as <italic>R</italic> <sub>0</sub><1. This further implies that reducing and keeping <italic>R</italic> <sub>0</sub> below the unity would be sufficient to eradicate cholera infection even in a periodic environment. Similar result was established for the autonomous system in [<xref rid="CIT0019" ref-type="bibr">19</xref>]; i.e. the cholera model with time-independent <italic>f</italic> and <italic>h</italic>.</p></sec><sec id="S005"><label>5. </label><title>Disease persistence</title><p>Now we consider the dynamics of the periodic model (1)–(4) when <italic>R</italic> <sub>0</sub>>1. For ease of discussion, let us omit Equation (3) from the system, since the total population <italic>N</italic> is fixed such that <italic>R</italic>=<italic>N</italic>−<italic>S</italic>−<italic>I</italic>. Define <disp-formula id="UM0011"><graphic xlink:href="tjbd-8-001-u011.jpg" position="float" orientation="portrait"/></disp-formula> Let <inline-formula><inline-graphic xlink:href="tjbd-8-001-m034.jpg"/></inline-formula> be the Poincaré map associated with models (1)–(4) such that <inline-formula><inline-graphic xlink:href="tjbd-8-001-m035.jpg"/></inline-formula> for all <italic>x</italic> <sub>0</sub>∈<italic>X</italic>, where <italic>u</italic>(<italic>t, x</italic> <sub>0</sub>) denotes the unique solution of the system with <italic>u</italic>(0, <italic>x</italic> <sub>0</sub>)=<italic>x</italic> <sub>0</sub>.</p><statement><label>definition 4 </label><p>The solutions of system (1)–(4) are said to be uniformly persistent if there exists some η>0 such that <disp-formula id="UM0012"><graphic xlink:href="tjbd-8-001-u012.jpg" position="float" orientation="portrait"/></disp-formula> whenever <italic>S</italic>(0)>0, <italic>I</italic>(0)>0, and <italic>B</italic>(0)>0.</p></statement><p>A more general definition of uniform persistence can be found in [<xref rid="CIT0024" ref-type="bibr">24</xref>]. We now state the following theorem, the proof of which is inspired by the work of Zhang and Zhao [<xref rid="CIT0022" ref-type="bibr">22</xref>].</p><statement><label>Theorem 5 </label><p>Let <italic>R</italic> <sub>0</sub>>1 and let (A1)–(A6) hold. Then the solutions of system (1)–(4) are uniformly persistent, and the system admits at least one positive ω-periodic solution.</p></statement><p> <italic>Proof</italic> Set <disp-formula id="UM0013"><graphic xlink:href="tjbd-8-001-u013.jpg" position="float" orientation="portrait"/></disp-formula> We first show that <disp-formula-group id="M0018"><disp-formula><graphic xlink:href="tjbd-8-001-e018.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Clearly, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m036.jpg"/></inline-formula>. Consider any initial values <inline-formula><inline-graphic xlink:href="tjbd-8-001-m037.jpg"/></inline-formula>. If <italic>I</italic>(0)=0 and <italic>B</italic>(0)>0, then <italic>I</italic>′(0)>0 by assumption (<italic>A</italic>5). Similarly, if <italic>B</italic>(0)=0 and <italic>I</italic>(0)>0, then <italic>B</italic>′(0)>0. Thus, it follows that <inline-formula><inline-graphic xlink:href="tjbd-8-001-m038.jpg"/></inline-formula> for 0<<italic>t</italic>≪1. This implies that <inline-formula><inline-graphic xlink:href="tjbd-8-001-m039.jpg"/></inline-formula>, and hence, we have (18).</p><p>Now, let us consider the fixed point <italic>M</italic> <sub>0</sub>=(<italic>N</italic>, 0, 0) and define <inline-formula><inline-graphic xlink:href="tjbd-8-001-m040.jpg"/></inline-formula>. We show that <disp-formula-group id="M0019"><disp-formula><graphic xlink:href="tjbd-8-001-e019.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Based on the continuity of solutions with respect to the initial conditions, for any ε>0 and <inline-formula><inline-graphic xlink:href="tjbd-8-001-m041.jpg"/></inline-formula>, there exists δ>0 small enough such that for all <inline-formula><inline-graphic xlink:href="tjbd-8-001-m042.jpg"/></inline-formula> with <inline-formula><inline-graphic xlink:href="tjbd-8-001-m043.jpg"/></inline-formula>, we have <disp-formula-group id="M0020"><disp-formula><graphic xlink:href="tjbd-8-001-e020.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> We claim that <disp-formula-group id="M0021"><disp-formula><graphic xlink:href="tjbd-8-001-e021.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Suppose by contradiction; that is, we suppose <inline-formula><inline-graphic xlink:href="tjbd-8-001-m044.jpg"/></inline-formula> for some <inline-formula><inline-graphic xlink:href="tjbd-8-001-m045.jpg"/></inline-formula>. Without loss of generality, we assume that <inline-formula><inline-graphic xlink:href="tjbd-8-001-m046.jpg"/></inline-formula>. Thus, <disp-formula-group id="M0022"><disp-formula><graphic xlink:href="tjbd-8-001-e022.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Moreover, for any <italic>t</italic>≥0, we can write <inline-formula><inline-graphic xlink:href="tjbd-8-001-m047.jpg"/></inline-formula> with <inline-formula><inline-graphic xlink:href="tjbd-8-001-m048.jpg"/></inline-formula> and <italic>n</italic> being the greatest integer less than or equal to <italic>t</italic>/ω. Then we obtain <disp-formula-group id="M0023"><disp-formula><graphic xlink:href="tjbd-8-001-e023.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> for any <italic>t</italic>≥0. Let <inline-formula><inline-graphic xlink:href="tjbd-8-001-m049.jpg"/></inline-formula>. It follows that <inline-formula><inline-graphic xlink:href="tjbd-8-001-m050.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m051.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-001-m052.jpg"/></inline-formula>. Note again that <inline-formula><inline-graphic xlink:href="tjbd-8-001-m053.jpg"/></inline-formula>. Then, based on assumptions (A1) and (A6), we have <disp-formula id="UM0014"><graphic xlink:href="tjbd-8-001-u014.jpg" position="float" orientation="portrait"/></disp-formula> and <disp-formula id="UM0015"><graphic xlink:href="tjbd-8-001-u015.jpg" position="float" orientation="portrait"/></disp-formula> Hence, we obtain <disp-formula-group id="M0024"><disp-formula><graphic xlink:href="tjbd-8-001-e024.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <italic>F</italic>−<italic>V</italic> is given by (10) and <disp-formula-group id="M0025"><disp-formula><graphic xlink:href="tjbd-8-001-e025.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Again based on [20, Theorem 2.2], <italic>R</italic> <sub>0</sub>>1 if and only if <inline-formula><inline-graphic xlink:href="tjbd-8-001-m054.jpg"/></inline-formula>. Thus, for ε>0 small enough, we have <inline-formula><inline-graphic xlink:href="tjbd-8-001-m055.jpg"/></inline-formula>. Using Lemma 2 and the comparison principle, we immediately obtain: <disp-formula-group id="M0026"><disp-formula><graphic xlink:href="tjbd-8-001-e026.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> which is a contradiction. Hence, <italic>M</italic> <sub>0</sub> is acyclic in <italic>M</italic> <sub>∂</sub>, and <italic>P</italic> is uniformly persistent with respect to <inline-formula><inline-graphic xlink:href="tjbd-8-001-m056.jpg"/></inline-formula>, which implies the uniform persistence of the solutions to the original system [<xref rid="CIT0023" ref-type="bibr">23</xref>]. Consequently, the Poincaré map <italic>P</italic> has a fixed point <inline-formula><inline-graphic xlink:href="tjbd-8-001-m057.jpg"/></inline-formula>, and it can be easily seen that <inline-formula><inline-graphic xlink:href="tjbd-8-001-m058.jpg"/></inline-formula>. Thus, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m059.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-001-m060.jpg"/></inline-formula> is a positive ω-periodic solution of the system.</p></sec><sec id="S006"><label>6. </label><title>Examples</title><p>In this section, we briefly discuss three different, and specific, cholera models in periodic environments. The models presented below are extended from recent work of Codeço [<xref rid="CIT0007" ref-type="bibr">7</xref>], Mukandavire <italic>et al</italic>. [<xref rid="CIT0012" ref-type="bibr">12</xref>], and Tien and Earn [<xref rid="CIT0017" ref-type="bibr">17</xref>], respectively. We focus on simulating seasonal variations by incorporating periodic environment-to-human transmission rates and periodic rates of human contribution to the population of <italic>V. cholerae</italic> in the aquatic environment. We study the epidemic and endemic cholera dynamics of a hypothetical community with <italic>N</italic>=10, 000 as the (normalized) total population, and compute the basic reproduction number <italic>R</italic> <sub>0</sub> for each model.</p><p>For comparison, we will also calculate the time-averaged reproduction number, denoted by [<italic>R</italic> <sub>0</sub>], for these cholera models. For any continuous periodic function <italic>g</italic>(<italic>t</italic>) with period ω, we may define its average as <disp-formula id="UM0016"><graphic xlink:href="tjbd-8-001-u016.jpg" position="float" orientation="portrait"/></disp-formula> Keeping with this notation, we define the time-averaged matrices of <italic>F</italic>(<italic>t</italic>) and <italic>V</italic>(<italic>t</italic>) for the general cholera model (1)–(4) as the following, respectively, <disp-formula id="UM0017"><graphic xlink:href="tjbd-8-001-u017.jpg" position="float" orientation="portrait"/></disp-formula> The time-averaged reproduction number of systems (1)–(4) is defined as the spectral radius of the time-averaged next-generation matrix [<italic>F</italic>][<italic>V</italic>]<sup>−1</sup>, and is given by <disp-formula-group id="M0027"><disp-formula><graphic xlink:href="tjbd-8-001-e027.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Based on Equation (27), the time-averaged reproduction number [<italic>R</italic> <sub>0</sub>] can be easily calculated for each of the three specific cholera models. It has been noted, however, that [<italic>R</italic> <sub>0</sub>] may overestimate or underestimate the infection risk for a non-autonomous epidemiological system [<xref rid="CIT0003" ref-type="bibr">3</xref>,<xref rid="CIT0020" ref-type="bibr">20</xref>]. Analytical estimates of the difference between <italic>R</italic> <sub>0</sub> and [<italic>R</italic> <sub>0</sub>] for some periodic systems are also presented in [<xref rid="CIT0003" ref-type="bibr">3</xref>]. Thus, it is of interest to compare the values of <italic>R</italic> <sub>0</sub> and [<italic>R</italic> <sub>0</sub>] for the three cholera models under consideration.</p><p>Meanwhile, we conduct numerical simulation for each model with initial conditions <italic>B</italic>(0)=<italic>R</italic>(0)=0, <italic>S</italic>(0)=<italic>N</italic>−<italic>I</italic>(0), <italic>I</italic>(0)=1; that is, one infected individual enters an entirely susceptible community. For easy comparison, we use the same parameter setting for all the three models, and these parameter values are based on the cholera data published on the recent Zimbabwe cholera outbreak [<xref rid="CIT0012" ref-type="bibr">12</xref>,<xref rid="CIT0021" ref-type="bibr">21</xref>]. We present typical infection curves for both scenarios, <italic>R</italic> <sub>0</sub><1 and <italic>R</italic> <sub>0</sub>>1, demonstrating disease extinction and disease persistence. Finally, in presenting each of these models, we keep the same notation for variables and parameters from the original autonomous model. We will clarify the different notation among the three extended models when necessary.</p><sec id="S006-S2001"><label>6.1 </label><title>The model of Codeço with periodic parameters</title><p>The original model in [<xref rid="CIT0007" ref-type="bibr">7</xref>] is now modified as <disp-formula-group id="M0028"><disp-formula><graphic xlink:href="tjbd-8-001-e028.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0029"><disp-formula><graphic xlink:href="tjbd-8-001-e029.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0030"><disp-formula><graphic xlink:href="tjbd-8-001-e030.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> which includes seasonal oscillations of the rate of exposure to contaminated water, <italic>a</italic>(<italic>t</italic>), and the rate of human contribution to the population of the pathogen, <italic>e</italic>(<italic>t</italic>), that are both periodic functions of time with a common period, ω=365 days, or 1 year: <disp-formula-group id="M0031"><disp-formula><graphic xlink:href="tjbd-8-001-e031.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Here <inline-formula><inline-graphic xlink:href="tjbd-8-001-m061.jpg"/></inline-formula> (or <inline-formula><inline-graphic xlink:href="tjbd-8-001-m062.jpg"/></inline-formula>) is the baseline value, or the time average, of <italic>a</italic>(<italic>t</italic>) (or <italic>e</italic>(<italic>t</italic>)), and <italic>ã</italic> (or <italic>e˜</italic>) denotes the (relative) amplitude of the seasonal oscillation in <italic>a</italic>(<italic>t</italic>) (or <italic>e</italic>(<italic>t</italic>)). To ensure both rates to be positive, we require that 0<<italic>ã</italic><1, 0<<italic>e˜</italic><1. In this model, <italic>H</italic> is the total population, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m063.jpg"/></inline-formula> is the probability a susceptible person becomes infected with cholera, β=<italic>mb</italic>−<italic>nb</italic> represents the net death rate of vibrios, and only the environment-to-human transmission pathway is considered. The incidence is <inline-formula><inline-graphic xlink:href="tjbd-8-001-m064.jpg"/></inline-formula> and the pathogen function is <inline-formula><inline-graphic xlink:href="tjbd-8-001-m065.jpg"/></inline-formula>. It is easily verified that the assumptions (A1)–(A6) hold for systems (28)–(30).</p><p>The disease-free equilibrium is given by <inline-formula><inline-graphic xlink:href="tjbd-8-001-m066.jpg"/></inline-formula>. From the next-generation matrices <disp-formula id="UM0018"><graphic xlink:href="tjbd-8-001-u018.jpg" position="float" orientation="portrait"/></disp-formula> it follows that basic reproduction number of the time-averaged autonomous system, based on (27), is given by <disp-formula-group id="M0032"><disp-formula><graphic xlink:href="tjbd-8-001-e032.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> The evolution operator <italic>Y</italic>(<italic>t, s</italic>), defined in Equation (8), for this model is given by <disp-formula id="UM0019"><graphic xlink:href="tjbd-8-001-u019.jpg" position="float" orientation="portrait"/></disp-formula> where <disp-formula id="UM0020"><graphic xlink:href="tjbd-8-001-u020.jpg" position="float" orientation="portrait"/></disp-formula> We then numerically evaluate the next infection operator (see Equation (6)) by <disp-formula-group id="M0033"><disp-formula><graphic xlink:href="tjbd-8-001-e033.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula-group id="M0034"><disp-formula><graphic xlink:href="tjbd-8-001-e034.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> for some positive integer <italic>M</italic>. Thus, for models (28)–(30), <disp-formula id="UM0021"><graphic xlink:href="tjbd-8-001-u021.jpg" position="float" orientation="portrait"/></disp-formula> To compute the basic reproduction number <italic>R</italic> <sub>0</sub>, we reduce the operator eigenvalue problem to a matrix eigenvalue problem in the form of <italic>A x</italic>=λ <italic>x</italic>, where matrix <italic>A</italic> can be constructed by arranging the entries of the function <italic>G</italic>. The basic reproduction number <italic>R</italic> <sub>0</sub> can then be approximated by numerically calculating the spectral radius of the matrix <italic>A</italic> [<xref rid="CIT0014" ref-type="bibr">14</xref>]. Other methods for computing <italic>R</italic> <sub>0</sub> also exist; for example, <italic>R</italic> <sub>0</sub> can be numerically calculated by solving the equation <italic>f</italic>(<italic>R</italic>)=1, where <italic>f</italic>(<italic>R</italic>) is the dominant Floquet multiplier of <inline-formula><inline-graphic xlink:href="tjbd-8-001-m067.jpg"/></inline-formula> [<xref rid="CIT0001" ref-type="bibr">1</xref>].</p><p>We have conducted numerical simulation to this model, and computed the reproductive numbers <italic>R</italic> <sub>0</sub> and [<italic>R</italic> <sub>0</sub>], for various values of <italic>a</italic>(<italic>t</italic>) and <italic>e</italic>(<italic>t</italic>). For illustration, we focus on the variation of <italic>a</italic>(<italic>t</italic>) here. In <xref rid="F0001" ref-type="fig">Figure 1</xref>(a) and 1(b), we vary <inline-formula><inline-graphic xlink:href="tjbd-8-001-m068.jpg"/></inline-formula> and <italic>ã</italic>, respectively, while keeping the values of other parameters fixed (see [<xref rid="CIT0012" ref-type="bibr">12</xref>]): <italic>H</italic>=10, 000, <italic>K</italic>=10<sup>6</sup>, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m069.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m070.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m071.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m072.jpg"/></inline-formula>, and <italic>e˜</italic>=0.5. In <xref rid="F0001" ref-type="fig">Figure 1</xref>(a), we see that <italic>R</italic> <sub>0</sub>=1 when <inline-formula><inline-graphic xlink:href="tjbd-8-001-m073.jpg"/></inline-formula>, and [<italic>R</italic> <sub>0</sub>]=1 when <inline-formula><inline-graphic xlink:href="tjbd-8-001-m074.jpg"/></inline-formula>. The value of <italic>ã</italic> is set as 0.5. It is clear that the time-averaged reproduction number underestimates the infection risk. Meanwhile, in <xref rid="F0001" ref-type="fig">Figure 1</xref>(b), we see that <italic>R</italic> <sub>0</sub>=1 when <inline-formula><inline-graphic xlink:href="tjbd-8-001-m075.jpg"/></inline-formula>, whereas [<italic>R</italic> <sub>0</sub>]=0.90 for all <italic>ã</italic>, again showing the inaccuracy of using [<italic>R</italic> <sub>0</sub>] for infection prediction. The value of <inline-formula><inline-graphic xlink:href="tjbd-8-001-m076.jpg"/></inline-formula> is set as 0.06 in this case. In addition, <xref rid="F0004" ref-type="fig">Figure 4</xref>(a) shows a typical infection curve of this model when <italic>R</italic> <sub>0</sub><1, where we observe that the disease quickly dies out and the disease-free equilibrium is asymptotically stable. In contrast, <xref rid="F0005" ref-type="fig">Figure 5</xref>(a) is a typical infection curve of this model for <italic>R</italic> <sub>0</sub>>1, where the disease persists and there is a positive ω-periodic solution. <fig id="F0004" orientation="portrait" position="float"><label>Fig. 4. </label><caption><p>A typical infection curve for each model when <italic>R</italic> <sub>0</sub><1, with initial condition <italic>I</italic>(0)=1. The solution quickly converges to the disease-free equilibrium with <italic>I</italic> <sub>0</sub>=0. (a) Model 6.1, (b) Model 6.2, (c) Model 6.3, (d) Model 6.3 in long term.</p></caption><graphic xlink:href="tjbd-8-001-g004"/></fig> <fig id="F0005" orientation="portrait" position="float"><label>Fig. 5. </label><caption><p>A typical infection curve for each model when <italic>R</italic> <sub>0</sub>>1, with initial condition <italic>I</italic>(0)=1. A periodic solution with ω=365 days forms after a long transient in each case. (a) Model 6.1, (b) Model 6.2, (c) Model 6.3, (d) Model 6.2 zoom-in.</p></caption><graphic xlink:href="tjbd-8-001-g005"/></fig> </p><fig id="F0001" orientation="portrait" position="float"><label>Fig. 1. </label><caption><p>Plots of the periodic threshold of <italic>R</italic> <sub>0</sub> for various <italic>ā</italic> and <italic>ã</italic>, respectively, in model 6.1. (a) <italic>R</italic> <sub>0</sub>=1 when <italic>ā</italic> ≈ 0.0625, and [<italic>R</italic> <sub>0</sub>]=1 when <italic>ā</italic> ≈ 0.0667; (b) <italic>R</italic> <sub>0</sub>=1 when <italic>ã</italic> ≈ 0.8407, and [<italic>R</italic> <sub>0</sub>]=0.90 for all <italic>ã</italic>.</p></caption><graphic xlink:href="tjbd-8-001-g001"/></fig></sec><sec id="S006-S2002"><label>6.2 </label><title>The model of Mukandavire <italic>et al</italic>. with periodic parameters</title><p>We extend the original model in [<xref rid="CIT0012" ref-type="bibr">12</xref>] to a periodic environment based on the following differential equations: <disp-formula-group id="M0035"><disp-formula><graphic xlink:href="tjbd-8-001-e035.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0036"><disp-formula><graphic xlink:href="tjbd-8-001-e036.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0037"><disp-formula><graphic xlink:href="tjbd-8-001-e037.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0038"><disp-formula><graphic xlink:href="tjbd-8-001-e038.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> The two periodic parameters are defined as <disp-formula-group id="M0039"><disp-formula><graphic xlink:href="tjbd-8-001-e039.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where β<sub><italic>e</italic></sub>(<italic>t</italic>) is the environment-to-human transmission rate and ξ(<italic>t</italic>) is the rate of contribution to <italic>V. cholerae</italic> in the aquatic environment. Though in different notation, β<sub><italic>e</italic></sub>(<italic>t</italic>) and ξ(<italic>t</italic>) have the same meaning as <italic>a</italic>(<italic>t</italic>) and <italic>e</italic>(<italic>t</italic>) in Equation (31). The incidence is <inline-formula><inline-graphic xlink:href="tjbd-8-001-m077.jpg"/></inline-formula> and the rate of change for the bacterial concentration is <inline-formula><inline-graphic xlink:href="tjbd-8-001-m078.jpg"/></inline-formula>. Both environment-to-human and human-to-human transmission pathways are included in this model; in particular, the environment-to-human transmission factor is based on a saturating form, which is the same as that in model (28)–(30), and the human-to-human transmission mode takes a bilinear form. It is clear that assumptions (A1)–(A6) hold for systems (35)–(38) as long as <inline-formula><inline-graphic xlink:href="tjbd-8-001-m079.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-001-m080.jpg"/></inline-formula>.</p><p>The disease-free equilibrium is given by <inline-formula><inline-graphic xlink:href="tjbd-8-001-m081.jpg"/></inline-formula>. From the next generation matrices <disp-formula id="UM0022"><graphic xlink:href="tjbd-8-001-u022.jpg" position="float" orientation="portrait"/></disp-formula> it follows that the time-averaged basic reproduction is <disp-formula-group id="M0040"><disp-formula><graphic xlink:href="tjbd-8-001-e040.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> The evolution operator <italic>Y</italic>(<italic>t, s</italic>) is given by <disp-formula id="UM0023"><graphic xlink:href="tjbd-8-001-u023.jpg" position="float" orientation="portrait"/></disp-formula> where <disp-formula id="UM0024"><graphic xlink:href="tjbd-8-001-u024.jpg" position="float" orientation="portrait"/></disp-formula> Thus, for this model, <disp-formula id="UM0025"><graphic xlink:href="tjbd-8-001-u025.jpg" position="float" orientation="portrait"/></disp-formula> for some positive integer <italic>M</italic>. Using the function <italic>G</italic>(<italic>t, s</italic>), the basic reproduction number <italic>R</italic> <sub>0</sub> can be numerically approximated by calculating the spectral radius of the corresponding matrix <italic>A</italic>.</p><p>In <xref rid="F0002" ref-type="fig">Figure 2</xref>(a) and 2(b), we vary <inline-formula><inline-graphic xlink:href="tjbd-8-001-m082.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-001-m083.jpg"/></inline-formula>, respectively, while keeping other parameters fixed: <italic>N</italic>=10, 000, κ=10<sup>6</sup>, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m084.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m085.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m086.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m087.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m088.jpg"/></inline-formula>, and <inline-formula><inline-graphic xlink:href="tjbd-8-001-m089.jpg"/></inline-formula>. In <xref rid="F0002" ref-type="fig">Figure 2</xref>(a), we again observe that the curve of [<italic>R</italic> <sub>0</sub>] is below that of <italic>R</italic> <sub>0</sub>, and we note that <italic>R</italic> <sub>0</sub>=1 when <inline-formula><inline-graphic xlink:href="tjbd-8-001-m090.jpg"/></inline-formula>. In <xref rid="F0002" ref-type="fig">Figure 2</xref>(b), we see that <italic>R</italic> <sub>0</sub>=1 when <inline-formula><inline-graphic xlink:href="tjbd-8-001-m091.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-001-m092.jpg"/></inline-formula> for all <inline-formula><inline-graphic xlink:href="tjbd-8-001-m093.jpg"/></inline-formula>. Note that <inline-formula><inline-graphic xlink:href="tjbd-8-001-m094.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-001-m095.jpg"/></inline-formula> correspond to <inline-formula><inline-graphic xlink:href="tjbd-8-001-m096.jpg"/></inline-formula> and <italic>ã</italic>, respectively, in Equation (31). Comparing the result in <xref rid="F0002" ref-type="fig">Figure 2</xref>(a) to that in <xref rid="F0001" ref-type="fig">Figure 1</xref>(a), we see that a lower value of the magnitude of the indirect transmission rate (<inline-formula><inline-graphic xlink:href="tjbd-8-001-m097.jpg"/></inline-formula> versus <inline-formula><inline-graphic xlink:href="tjbd-8-001-m098.jpg"/></inline-formula>) is needed to reach the threshold value <italic>R</italic> <sub>0</sub>=1 for the current model, due to the incorporation of the direct transmission mode. Similarly, we observe that the values of [<italic>R</italic> <sub>0</sub>] in <xref rid="F0002" ref-type="fig">Figure 2</xref>(a) and 2(b) are lower than those in <xref rid="F0001" ref-type="fig">Figure 1</xref>(a) and 1(b) for the same value of the parameter. In addition, <xref rid="F0004" ref-type="fig">Figure 4</xref>(b) is an infection curve when <italic>R</italic> <sub>0</sub><1, and <xref rid="F0005" ref-type="fig">Figure 5</xref>(b) is an infection curve when <italic>R</italic> <sub>0</sub>>1, for the current model. We observe similar patterns as in <xref rid="F0004" ref-type="fig">Figures 4</xref>(a) and 5(a).</p><fig id="F0002" orientation="portrait" position="float"><label>Fig. 2. </label><caption><p>Plots of the periodic threshold of <italic>R</italic> <sub>0</sub> for various β<sub><italic>ē</italic></sub> and β<sub><italic>e˜</italic></sub>, respectively, in model 6.2. (a) <italic>R</italic> <sub>0</sub>=1 when β<sub><italic>ē</italic></sub> ≈ 0.0321 and [<italic>R</italic> <sub>0</sub>]=1 when β<sub><italic>ē</italic></sub> ≈ 0.0334; (b) <italic>R</italic> <sub>0</sub>=1 when β<sub><italic>e˜</italic></sub> ≈ 0.5688 and [<italic>R</italic> <sub>0</sub>]=0.9797 for all β<sub><italic>e˜</italic></sub>.</p></caption><graphic xlink:href="tjbd-8-001-g002"/></fig></sec><sec id="S006-S2003"><label>6.3 </label><title>The model of Tien and Earn with periodic parameters</title><p>The original model in [<xref rid="CIT0017" ref-type="bibr">17</xref>], where the pathogen concentration is denoted by <italic>W</italic> instead of <italic>B</italic>, is extended to a periodic environment in the form of <disp-formula-group id="M0041"><disp-formula><graphic xlink:href="tjbd-8-001-e041.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0042"><disp-formula><graphic xlink:href="tjbd-8-001-e042.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0043"><disp-formula><graphic xlink:href="tjbd-8-001-e043.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0044"><disp-formula><graphic xlink:href="tjbd-8-001-e044.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula-group id="M0045"><disp-formula><graphic xlink:href="tjbd-8-001-e045.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> denote the water-to-person transmission rate and the shedding rate from infected individuals into the water, respectively. Here the time-periodic parameters <italic>b</italic> <sub><italic>W</italic></sub>(<italic>t</italic>) and α(<italic>t</italic>) play the same role as <italic>a</italic>(<italic>t</italic>) and <italic>e</italic>(<italic>t</italic>) in model (28)–(30), or β<sub><italic>e</italic></sub>(<italic>t</italic>) and ξ(<italic>t</italic>) in model (35)–(38). The incidence in the current model is <inline-formula><inline-graphic xlink:href="tjbd-8-001-m099.jpg"/></inline-formula> and the pathogen function is <inline-formula><inline-graphic xlink:href="tjbd-8-001-m100.jpg"/></inline-formula>. The dual-transmission pathways are included in this model by using bi-linear forms, however, no saturation effect was considered. It is straightforward to verify that assumptions (A1)–(A6) hold for systems (41)–(44) given that <inline-formula><inline-graphic xlink:href="tjbd-8-001-m101.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m102.jpg"/></inline-formula>.</p><p>Clearly, the DFE is given by <inline-formula><inline-graphic xlink:href="tjbd-8-001-m103.jpg"/></inline-formula>. From the next-generation matrices <disp-formula id="UM0026"><graphic xlink:href="tjbd-8-001-u026.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>it follows that <disp-formula-group id="M0046"><disp-formula><graphic xlink:href="tjbd-8-001-e046.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> The evolution operator <italic>Y</italic>(<italic>t, s</italic>) is given by <disp-formula id="UM0027"><graphic xlink:href="tjbd-8-001-u027.jpg" position="float" orientation="portrait"/></disp-formula> where <disp-formula id="UM0028"><graphic xlink:href="tjbd-8-001-u028.jpg" position="float" orientation="portrait"/></disp-formula> Thus, for models (41)–(44), <disp-formula id="UM0029"><graphic xlink:href="tjbd-8-001-u029.jpg" position="float" orientation="portrait"/></disp-formula> We have conducted similar numerical simulations as before and calculated the two reproduction numbers. In presenting the results of <italic>R</italic> <sub>0</sub>, we could, in principle, vary <inline-formula><inline-graphic xlink:href="tjbd-8-001-m104.jpg"/></inline-formula> while keeping other parameters fixed. However, due to the bilinear form of the indirect transmission mode employed in the current model (and due to the very high value of <italic>W</italic>), the meaningful values of <inline-formula><inline-graphic xlink:href="tjbd-8-001-m105.jpg"/></inline-formula> are several magnitudes smaller than those of <inline-formula><inline-graphic xlink:href="tjbd-8-001-m106.jpg"/></inline-formula> in Equation (31), or <inline-formula><inline-graphic xlink:href="tjbd-8-001-m107.jpg"/></inline-formula> in Equation (39), making it impossible to compare the result with the other two models. Thus, we have chosen to present only the result of <italic>R</italic> <sub>0</sub> (and [<italic>R</italic> <sub>0</sub>]) versus <italic>b</italic> <sub><italic>W˜</italic></sub> in <xref rid="F0003" ref-type="fig">Figure 3</xref>. Values of the other parameters are: <italic>N</italic>=10, 000, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m108.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m109.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m110.jpg"/></inline-formula>, <italic>b</italic> <sub><italic>I</italic></sub>=0.00001, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m111.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-001-m112.jpg"/></inline-formula>, and <inline-formula><inline-graphic xlink:href="tjbd-8-001-m113.jpg"/></inline-formula>. We see that <italic>R</italic> <sub>0</sub>=1 when <inline-formula><inline-graphic xlink:href="tjbd-8-001-m114.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-001-m115.jpg"/></inline-formula> for all <italic>b</italic> <sub><italic>W˜</italic></sub>. The result shows similar pattern to that in <xref rid="F0002" ref-type="fig">Figure 2</xref>(b) as both models include dual transmission pathways. <xref rid="F0004" ref-type="fig">Figure 4</xref>(c) displays an infection curve when <italic>R</italic> <sub>0</sub><1 for the current model, and <xref rid="F0005" ref-type="fig">Figure 5</xref>(c) shows an infection curve when <italic>R</italic> <sub>0</sub>>1. <fig id="F0003" orientation="portrait" position="float"><label>Fig. 3. </label><caption><p>Plot of the periodic threshold of <italic>R</italic> <sub>0</sub> for various <italic>b</italic> <sub><italic>W˜</italic></sub> in model 6.3. <italic>R</italic> <sub>0</sub>=1 when <italic>b</italic> <sub><italic>W˜</italic></sub> ≈ 0.3706 and [<italic>R</italic> <sub>0</sub>]=0.9872 for all <italic>b</italic> <sub><italic>W˜</italic></sub>.</p></caption><graphic xlink:href="tjbd-8-001-g003"/></fig> </p><p>Finally, from <xref rid="F0004" ref-type="fig">Figure 4</xref>(a)– 4(c), as expected, we see that when <italic>R</italic> <sub>0</sub><1, the infected population <italic>I</italic> quickly decreases to zero and stays there forever (for example, see <xref rid="F0004" ref-type="fig">Figure 4</xref>(d) for the long-term behaviour of the model 6.3), showing that the disease dies out in each model. Indeed, similar patterns were observed for various initial conditions (not shown here), an evidence that the disease-free equilibrium is globally asymptotically stable for each model. <xref rid="F0005" ref-type="fig">Figure 5</xref>(a)–5(c) illustrates typical infection curves for the three models when <italic>R</italic> <sub>0</sub>>1. In this case, for each model, the disease persists and after a long, transient period, the infection approaches a positive ω-periodic solution. <xref rid="F0005" ref-type="fig">Figure 5</xref>(d) shows a zoomed-in picture for the model 6.2 where the periodic solution is highlighted and a period of ω=365 days (or 1 year) can be observed.</p></sec></sec><sec sec-type="conclusions" id="S007"><label>7. </label><title>Conclusions</title><p>We have presented a general non-autonomous cholera model in a periodic environment. Seasonally variational factors have been incorporated into the incidence function <italic>f</italic> and the pathogen function <italic>h</italic>. Using the next infection operator introduced in [<xref rid="CIT0020" ref-type="bibr">20</xref>], we have derived and computed the basic reproduction number <italic>R</italic> <sub>0</sub> of our periodic cholera model, and have conducted a careful analysis on the epidemic and endemic dynamics. Our results have established <italic>R</italic> <sub>0</sub> as a sharp threshold for cholera dynamics in periodic environments; i.e. disease completely dies out if <italic>R</italic> <sub>0</sub><1 and uniformly persists if <italic>R</italic> <sub>0</sub>>1. The general analysis is demonstrated through three specific cholera models, and numerical simulation results are consistent with analytical predictions.</p><p>The complication of cholera modelling lies in that, on top of the multiple transmission pathways that involve both environment-to-human (or, indirect) and human-to-human (or, direct) routes, disease dynamics are also subject to strong seasonal variation. Thus, many different factors, ranging from ecological, environmental, societal, and climatic, need to be considered in constructing a more accurate mathematical model. We have incorporated periodicity into the general incidence and pathogen functions in our model, in order to represent these various seasonal oscillations in a generic manner. Although in the three specific examples presented in Section 6 we have focused on two periodic parameters (i.e. the rates of human–environment contact and human contribution to environmental vibrios) for the purposes of demonstration and easy comparison, one can easily incorporate periodicity into other model parameters, depending on the context of the modelling. In addition, similar analysis can be conducted to other cholera models (e.g. [<xref rid="CIT0009" ref-type="bibr">9</xref>]), and the framework can be extended to model other water-borne infectious diseases, such as dysentery, typhoid fever, and campylobacteriosis.</p><p>This work was partially supported by the National Science Foundation under Grant Numbers 0813691 and 1245769. The authors are grateful to the two anonymous referees for their helpful comments to improve this paper.</p></sec>
Apoptosis in virus infection dynamics models	<p>In this paper, on the basis of the simplified two-dimensional virus infection dynamics model, we propose two extended models that aim at incorporating the influence of activation-induced apoptosis which directly affects the population of uninfected cells. The theoretical analysis shows that increasing apoptosis plays a positive role in control of virus infection. However, after being included the third population of cytotoxic T lymphocytes immune response in HIV-infected patients, it shows that depending on intensity of the apoptosis of healthy cells, the apoptosis can either promote or comfort the long-term evolution of HIV infection. Further, the discrete-time delay of apoptosis is incorporated into the pervious model. Stability switching occurs as the time delay in apoptosis increases. Numerical simulations are performed to illustrate the theoretical results and display the different impacts of a delay in apoptosis.</p>	<contrib contrib-type="author"><name><surname>Fan</surname><given-names>Ruili</given-names></name><xref ref-type="aff" rid="AF1"> <sup>a</sup> </xref><xref rid="AN1" ref-type="author-notes"/></contrib><contrib contrib-type="author"><name><surname>Dong</surname><given-names>Yueping</given-names></name><xref ref-type="aff" rid="AF2"> <sup>b</sup> </xref><xref rid="AN2" ref-type="author-notes"/></contrib><contrib contrib-type="author"><name><surname>Huang</surname><given-names>Gang</given-names></name><xref ref-type="aff" rid="AF1"> <sup>a</sup> </xref><xref ref-type="corresp" rid="cor1"> <sup>*</sup> </xref></contrib><contrib contrib-type="author"><name><surname>Takeuchi</surname><given-names>Yasuhiro</given-names></name><xref ref-type="aff" rid="AF3"> <sup>c</sup> </xref><xref rid="AN3" ref-type="author-notes"/></contrib><aff id="AF1"><label><sup>a</sup></label><institution><named-content content-type="department">School of Mathematics and Physics</named-content>, <named-content content-type="institution-name">China University of Geosciences</named-content></institution>, <named-content content-type="city">Wuhan</named-content><named-content content-type="postal-code">430074</named-content>, <country>China</country></aff><aff id="AF2"><label><sup>b</sup></label><institution><named-content content-type="department">Graduate School of Science and Technology</named-content>, <named-content content-type="institution-name">Shizuoka University</named-content></institution>, <named-content content-type="city">Hamamatsu</named-content><named-content content-type="postal-code">432-8561</named-content>, <country>Japan</country></aff><aff id="AF3"><label><sup>c</sup></label><institution><named-content content-type="department">Department of Physics and Mathematics</named-content>, <named-content content-type="institution-name">Aoyama Gakuin University</named-content></institution>, <named-content content-type="city">Sagamihara</named-content><named-content content-type="postal-code">252-5258</named-content>, <country>Japan</country></aff>	Journal of Biological Dynamics	<sec sec-type="intro" id="S001"><label>1. </label><title>Introduction</title><p>There are two different ways by which a cell can die: necrosis and apoptosis. Necrosis is a form of traumatic cell death that results from acute cellular injury, such as poison, a bodily injury, an infection or getting cut off from the blood supply. When cells die from necrosis, it is rather a messy affair. The death causes inflammation that can lead to further injury within the body. Apoptosis, or programmed cell death, on the other hand, is a naturally occurring process of cell suicide in response to a variety of physiological and non-physiological stimuli including chemical insults, virus infections and developmental cues, being essential in the maintenance of tissue development and homeostasis in the adult as well as in the regulation of immune responses [<xref rid="CIT0002" ref-type="bibr">2</xref>,<xref rid="CIT0048" ref-type="bibr">48</xref>]. The process of apoptosis shows a number of distinct characteristic morphological changes, including cell shrinkage and partial detachment from substratum, plasma membrane blebbing, chromatin condensation and intra-nucleosomal cleavage and ultimately cell fragmentation into apoptotic bodies which are phagocytosed without provoking an inflammatory response [<xref rid="CIT0037" ref-type="bibr">37</xref>,<xref rid="CIT0050" ref-type="bibr">50</xref>].</p><p>Apoptosis is genetically controlled and several viral gene products affect apoptosis by interacting directly with components of the highly conserved biochemical pathway which regulates cell death. Viruses may perform two functions. One is that viruses have evolved mechanisms to block the premature apoptosis of infected cells facilitating either the establishment and maintenance of persistent infection or prolonging the survival of lytically infected cells such that the production of progeny virus is maximized. The other function is that increasing number of viruses actively promote apoptosis, which is the culmination of a lytic infection and serving to spread virus progeny to neighbouring cells while evading the host inflammatory responses [<xref rid="CIT0037" ref-type="bibr">37</xref>]. Viral induction of apoptosis occurs when one or several cells in a living organism is infected with a virus leading to cell death. Cell death in organisms is necessary for normal development of cells and cell cycle maturation. It is also critical in maintaining the regular functions and activities of cells.</p><p>Apoptosis is not only a key event in biological homeostasis but is also involved in the pathogenesis of many human diseases including acquired immune deficiency syndrome (AIDS) [<xref rid="CIT0001" ref-type="bibr">1</xref>]. AIDS was first clinically observed from the US Centers for Disease Control in 1981, which is now defined as either a CD4+T-cell count below about 200 per microlitre blood, or the occurrence of specific diseases in association with certain HIV-related conditions and symptoms. Two years later the causative virus was identified by two separate research groups and afterwards named the human immunodeficiency virus (HIV) [<xref rid="CIT0029" ref-type="bibr">29</xref>]. HIV is a retrovirus that primarily infects vital cells of the human immune system such as helper T-cells (specifically CD4+T-cells), macrophages and dendritic cells. The HIV infection usually leads to a progressive decay in the functionality and number of CD4+T-cells (normally about 1000 per microlitre in blood), resulting in a consequent impairment in host immune defenses and increasing susceptibility to opportunistic infections and malignancies for patients.</p><p>What causes the progressive depletion of the CD4+T helper cells leading to immunodeficiency in HIV infection?</p><p>The gradual decline of CD4+T-cells in HIV-infected patients is one of the most fundamental and controversial issues in AIDS research [<xref rid="CIT0021" ref-type="bibr">21</xref>]. It has been reported that CD4+T-cell loss may be attributed to one of the following: (i) direct destruction by HIV cytopathic effects [<xref rid="CIT0033" ref-type="bibr">33</xref>]; (ii) apoptosis induced by HIV proteins, such as Env, Tat (Tatanus antitoxin), Nef, Vpu and Vpr, which were proposed by Gougeon <italic>et al.</italic> in the 1990s [<xref rid="CIT0019" ref-type="bibr">19</xref>,<xref rid="CIT0020" ref-type="bibr">20</xref>] and reviewed by Gougeon [<xref rid="CIT0018" ref-type="bibr">18</xref>]; (iii) excessive infection-induced immune cells activation drives CD4+T-cell depletion [<xref rid="CIT0014" ref-type="bibr">14</xref>]; (iv) HIV-1-induced apoptosis in bystander uninfected cells [<xref rid="CIT0001" ref-type="bibr">1</xref>]. Extensive body of studies has tried to address the phenomenon behind accelerated apoptosis in T-cells in HIV-infected patients since apoptosis has been suggested as another mechanism responsible for T-cell depletion since 1991 [<xref rid="CIT0001" ref-type="bibr">1</xref>,<xref rid="CIT0003" ref-type="bibr">3</xref>,<xref rid="CIT0005" ref-type="bibr">5</xref>,<xref rid="CIT0007" ref-type="bibr">7</xref>,<xref rid="CIT0014" ref-type="bibr">14</xref>,<xref rid="CIT0017" ref-type="bibr">17–22</xref>,<xref rid="CIT0033" ref-type="bibr">33</xref>,<xref rid="CIT0038" ref-type="bibr">38</xref>,<xref rid="CIT0048" ref-type="bibr">48</xref>,<xref rid="CIT0057" ref-type="bibr">57</xref>]. The role of bystander apoptosis induction in HIV infection and its role in disease progression is reviewed in [<xref rid="CIT0017" ref-type="bibr">17</xref>]. Bystander apoptosis of neighbouring uninfected cells appears to encompass an explanation for most of the phenomenon observed during HIV infection that leads to progression to AIDS and remains one of the leading hypothesis for CD4+T-cell loss [<xref rid="CIT0017" ref-type="bibr">17</xref>]. Analysis of blood cells derived from HIV-infected patients when cultured <italic>in vitro</italic> has revealed accelerated cell death infected as well as uninfected T-cells but, remarkably, the vast majority of the cells that undergo apoptosis are uninfected [<xref rid="CIT0047" ref-type="bibr">47</xref>].</p><p>Nowadays, most researches are focused on the elucidation of apoptotic mechanisms. The possibility that modulating cell death by targeting specific factors involved in the whole process could be the key for cure of progression of HIV infection to AIDS, which is the also motivation of the work in this paper. Mathematical models can provide some insights into the dynamics of HIV viral load <italic>in vivo</italic> and may play a significant role in the development of a better understanding of HIV/AIDS and drug therapies. Here we mainly focus on qualitative understanding of the impact of apoptosis on virus infection dynamics by analysing several models with or without time delay.</p><p>The outline of the paper is as follows. In the next section, we incorporate the activation-induced apoptosis into virus infection model and analyse the impact of apoptosis of uninfected cells. In Section 3, we take into account specific cytotoxic T lymphocytes (CTLs) immune in HIV infection model including apoptosis. By analysing dynamical properties of the model, we give the potential impact of apoptosis. In Section 4, we focus on the existence of Hopf bifurcation when the apoptosis activation delay is present. Detailed numerical simulations are also given. The discussion of the mathematical results and of their biological implications are presented in Section 5.</p></sec><sec id="S002"><label>2. </label><title>Positive impact of apoptosis in viral dynamics</title><p>Mathematical models have been of central importance for understanding the dynamics of populations in an epidemiological context. Since the early 1990s numerous epidemic models have been used to describe the dynamics between viral infections and the immune response [<xref rid="CIT0006" ref-type="bibr">6</xref>,<xref rid="CIT0008" ref-type="bibr">8</xref>,<xref rid="CIT0025" ref-type="bibr">25</xref>,<xref rid="CIT0026" ref-type="bibr">26</xref>,<xref rid="CIT0036" ref-type="bibr">36</xref>,<xref rid="CIT0039" ref-type="bibr">39</xref>,<xref rid="CIT0044" ref-type="bibr">44</xref>,<xref rid="CIT0046" ref-type="bibr">46</xref>,<xref rid="CIT0051" ref-type="bibr">51</xref>,<xref rid="CIT0053" ref-type="bibr">53</xref>,<xref rid="CIT0055" ref-type="bibr">55</xref>], particularly in the context of HIV infection [<xref rid="CIT0004" ref-type="bibr">4</xref>,<xref rid="CIT0009" ref-type="bibr">9</xref>,<xref rid="CIT0010" ref-type="bibr">10</xref>,<xref rid="CIT0012" ref-type="bibr">12</xref>,<xref rid="CIT0013" ref-type="bibr">13</xref>,<xref rid="CIT0023" ref-type="bibr">23</xref>,<xref rid="CIT0024" ref-type="bibr">24</xref>,<xref rid="CIT0027" ref-type="bibr">27</xref>,<xref rid="CIT0028" ref-type="bibr">28</xref>,<xref rid="CIT0030" ref-type="bibr">30</xref>,<xref rid="CIT0034" ref-type="bibr">34</xref>,<xref rid="CIT0035" ref-type="bibr">35</xref>,<xref rid="CIT0040" ref-type="bibr">40–43</xref>,<xref rid="CIT0045" ref-type="bibr">45</xref>,<xref rid="CIT0049" ref-type="bibr">49</xref>,<xref rid="CIT0052" ref-type="bibr">52</xref>,<xref rid="CIT0054" ref-type="bibr">54</xref>,<xref rid="CIT0058" ref-type="bibr">58</xref>]. The basic model for viral dynamics is given by the following simple three-dimensional system, <disp-formula-group id="M0001"><disp-formula><graphic xlink:href="tjbd-8-020-e001.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Here <italic>x</italic>(<italic>t</italic>) represents the concentration of uninfected or healthy cells at time <italic>t, y</italic>(<italic>t</italic>) represents the concentration of infected cells that produce virion at time <italic>t</italic> and <italic>v</italic>(<italic>t</italic>) represents the concentration of viruses at time <italic>t</italic>. λ is the rate at which new healthy cells are generated. <italic>d</italic> is the death rate of uninfected cells. β is the constant rate of infection of the healthy cells. <italic>a</italic> is the death rate of infected cells due either to the virus or to the immune system. Free viral particles are produced by infected cells at a rate <italic>ky</italic> and removed at a rate <italic>uv</italic>. All parameters <inline-formula><inline-graphic xlink:href="tjbd-8-020-m001.jpg"/></inline-formula> and <italic>u</italic> have positive values.</p><p>System (1) can be further simplified if we take into consideration that an average life span of viral particles is usually significantly shorter than one of infected cells. Therefore, it can be assumed that compared with a ‘slow’ variation of the infected cells level, the virus load <italic>v</italic>(<italic>t</italic>) relatively quickly reaches a quasi-equilibrium level. The equality <inline-formula><inline-graphic xlink:href="tjbd-8-020-m002.jpg"/></inline-formula> holds in the quasi-equilibrium state and hence <inline-formula><inline-graphic xlink:href="tjbd-8-020-m003.jpg"/></inline-formula>. This assumption is referred to as ‘separation of time scales’ and is in common use in the virus dynamics [<xref rid="CIT0024" ref-type="bibr">24</xref>,<xref rid="CIT0027" ref-type="bibr">27</xref>,<xref rid="CIT0028" ref-type="bibr">28</xref>,<xref rid="CIT0044" ref-type="bibr">44</xref>]. We have to stress that this assumption does not imply that the virus concentration <italic>v</italic>(<italic>t</italic>) remains constant; on the contrary, it is assumed to be proportional to the varying concentration of infected cells <italic>y</italic>(<italic>t</italic>). Accordingly, system (1) can now be reformulated as a system of two ordinary differential equations, <disp-formula-group id="M0002"><disp-formula><graphic xlink:href="tjbd-8-020-e002.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>Virus infection and replication are often associated with apoptosis and this effect is likely to be responsible for much of the pathology associated with infectious disease. Many of the viruses associated with oncogenic transformation have adopted strategies for blocking apoptosis highlighting the centrality of this effect in carcinogenesis [<xref rid="CIT0037" ref-type="bibr">37</xref>]. Indeed, these viral anti-apoptotic strategies can also contribute to the pathogenesis of virus infection and, in extreme situations, promote the oncogenic capacity of certain viruses. Some studies of the various mechanism used by viruses to suppress apoptosis have shed light on the fundamental biochemical pathways responsible for regulating programmed cell death [<xref rid="CIT0056" ref-type="bibr">56</xref>]. Hence, understanding the mechanisms by which viruses regulate apoptosis may lead to the development of novel therapies for infectious disease.</p><p>Apoptosis is an important biological process that eliminates selected cells for the benefit of the whole organism. The ‘decision’ for apoptosis can come from the cell itself, or to be induced from its surrounding environment. We consider two different mechanisms by which the healthy cell commits suicide by apoptosis. One mechanism is generated by signals arising within the cell. Another is triggered by death activators binding to receptors at the infected cell surface. We use the following two models to describe those mechanisms.</p><sec id="S002-S2001"><label>2.1 </label><title>Case I: apoptosis triggered by internal signals</title><p>One of the ways in which apoptosis occurs is by internal signals. When there is internal cellular damage, it triggers the release of a protein called BAX which punctures the mitochondrial membrane (Mitochondria handle cellular energy production). This causes Cytochrome C to leak from the mitochondria and bind with Apaf-1 (Apoptosis protease activating factor). The binding of Cytochrome C and Apaf-1 causes the formation of apoptosomes. This sets in motion a series of caspase formations which causes a structural breakdown in the cell and DNA destruction.</p><p>In this case, when apoptosis is taken into account in system (2), we have, <disp-formula-group id="M0003"><disp-formula><graphic xlink:href="tjbd-8-020-e003.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <italic>d</italic> is the death rate due to necrosis, and <italic>d</italic> <sub>α</sub> describes the death rate related to activation-induced apoptosis of uninfected cells.</p><p>The basic reproductive number of virus for system (3) is given by <disp-formula-group id="M0004"><disp-formula><graphic xlink:href="tjbd-8-020-e004.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> which describes the average number of infected cells generated from one infected cell at the beginning of the infection process. System (3) has two equilibria which are <inline-formula><inline-graphic xlink:href="tjbd-8-020-m004.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-020-m005.jpg"/></inline-formula> when <italic>R</italic> <sub>0</sub>>1.</p><p>The global dynamical properties of model (3) is clear, that is, all solutions of system (3) converge to the infection free steady state <inline-formula><inline-graphic xlink:href="tjbd-8-020-m006.jpg"/></inline-formula> when <italic>R</italic> <sub>0</sub>≤1 and the infected steady state <inline-formula><inline-graphic xlink:href="tjbd-8-020-m007.jpg"/></inline-formula> is globally asymptotically stable when <italic>R</italic> <sub>0</sub>>1. Obviously, the increasing of apoptosis parameter <italic>d</italic> <sub>α</sub> could lead to the basic reproductive number <italic>R</italic> <sub>0</sub> to less than 1, which means the virus load can be eliminated. In addition, there exists a threshold value <italic>d</italic> <sub>0</sub>, where <inline-formula><inline-graphic xlink:href="tjbd-8-020-m008.jpg"/></inline-formula>, and when <inline-formula><inline-graphic xlink:href="tjbd-8-020-m009.jpg"/></inline-formula>, the level of infected cells tend to zero. From the term <inline-formula><inline-graphic xlink:href="tjbd-8-020-m010.jpg"/></inline-formula>, we know that increasing in <italic>d</italic> <sub>α</sub> can decrease the load of infected cells. It shows that the apoptosis is useful to reduce the number of virus, which implies that apoptosis plays a positive role in preventing the spread of virus infection.</p></sec><sec id="S002-S2002"><label>2.2 </label><title>Case II: Apoptosis triggered by external signals</title><p>Another way that apoptosis can occur is through external signals. The previously mentioned protein called Fas on the cells’ surface can bind with another protein called tumour necrosis factor receptor. This binding will signal the cytoplasm and activate caspase 8 which in turn will set in motion a series of caspase formations. These series of caspase formations will ultimately cause phagocytosis (the cells being consumed by phagocytes). This process is often induced by T-cells which use the apoptosis process to destroy cells which are carrying viruses.</p><p>In the special case of lymphocytes, apoptosis plays an important role in optimizing the immune system by compensating lymphocytes proliferation through the elimination of cells that have become ill or ineffective. In HIV infection, apoptosis is a complex process that involves both HIV-infected and uninfected cells. HIV has evolved multiple mechanisms to promote survival for long enough to ensure a productive infection and this may be supported by the fact that infected cells do not undergo apoptosis as readily as uninfected bystander cells. This is also confirmed by direct evidence in lymph nodes where apoptosis was seen primarily in the uninfected bystander cells [<xref rid="CIT0015" ref-type="bibr">15</xref>]. As reported in [<xref rid="CIT0038" ref-type="bibr">38</xref>], lymph nodes of HIV-infected individuals contain a high percentage (with respect to uninfected individuals) of uninfected cells which are in an apoptotic state (that is which are ready to enter an apoptotic process).</p><p>As mentioned above, in HIV-infected persons, although both infected and uninfected cells undergo accelerated apoptosis, massive apoptosis was predominantly observed in uninfected CD4+T-cells present in lymph node, thymus or spleen. For the mathematical simplicity, we assume that all apoptosis-inducing factors of uninfected CD4+T-cells are directly correlated with the concentration of HIV-infected CD4+T-cells and since the chemical messengers inducing apoptosis are released by the infected CD4+T-cells, we further assume that the concentration of the chemical messengers inducing apoptosis is proportional to the concentration of the infected CD4+T-cells. Furthermore, when apoptosis of infected cells is not influenced by the presence of other cells, its effect is attributed to the death rate of infected cells. Consequently, it is more reasonable that the model (3) is modified as follows: <disp-formula-group id="M0005"><disp-formula><graphic xlink:href="tjbd-8-020-e005.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Here α is a positive parameter that represents the ‘apoptosis rate’ (also called apoptosis parameter) of uninfected cells in the presence of infected cells. It is reasonable to assume that α is bounded, that is, <inline-formula><inline-graphic xlink:href="tjbd-8-020-m011.jpg"/></inline-formula>. β is new per capita infection rate which describes both the probability of an infecting contact and the reproduction of virus. It should be pointed out that d <italic>x</italic> is a natural death rate of healthy CD4+T-cells. α <italic>xy</italic> is the death rate of healthy CD4+T-cells for the activation-induced apoptosis.</p><p>Consider system (5). There are two physically and biologically relevant non-negative equilibria, namely, <list list-type="bullet"><list-item><p> <inline-formula><inline-graphic xlink:href="tjbd-8-020-m012.jpg"/></inline-formula>, infection or disease-free equilibrium, at which all individuals are susceptible and the population remain in the absence of disease.</p></list-item><list-item><p> <inline-formula><inline-graphic xlink:href="tjbd-8-020-m013.jpg"/></inline-formula>, positive or endemic equilibrium, where <disp-formula id="UM0001"><graphic xlink:href="tjbd-8-020-u001.jpg" position="float" orientation="portrait"/></disp-formula> If <inline-formula><inline-graphic xlink:href="tjbd-8-020-m014.jpg"/></inline-formula>, then <italic>Ê</italic> <sub>1</sub> exists.</p></list-item></list> </p><p>It is noteworthy that the basic reproduction number of infected cells <italic>R˜</italic> <sub>0</sub> is independent of the apoptosis parameter α and the concentration of infected cells <italic>y</italic> <sub>1</sub> of <italic>Ê</italic> <sub>1</sub> is inversely proportional to the apoptosis parameter α.</p></sec><sec id="S002-S2003"><label>2.3 </label><title>Evolution of apoptosis parameter</title><p>We fix the parameter values as: <inline-formula><inline-graphic xlink:href="tjbd-8-020-m015.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-020-m016.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-020-m017.jpg"/></inline-formula>. By calculation, we have <inline-formula><inline-graphic xlink:href="tjbd-8-020-m018.jpg"/></inline-formula>, system (5) has an endemic steady state <inline-formula><inline-graphic xlink:href="tjbd-8-020-m019.jpg"/></inline-formula> and <italic>E</italic> <sub>1ˆ</sub> is globally asymptotically stable (<xref rid="F0001" ref-type="fig">Figure 1</xref>(a)). It can be expected that in the endemic state the healthy cells level <italic>x</italic> <sub>1</sub> of system (5) is identical to that of system (2), but the infected cells level <italic>y</italic> <sub>1</sub> of system (5) is lower than that of system (2). Furthermore, the infected cells level <italic>y</italic> <sub>1</sub> reduces gradually with the increasing of apoptosis parameter α, which is illustrated by numerical simulation (<xref rid="F0001" ref-type="fig">Figure 1</xref>(b)). <fig id="F0001" orientation="portrait" position="float"><label>Figure 1. </label><caption><p>(a) The time evolution of the trajectory of system (2) and system (5) for healthy cells <italic>x</italic>(<italic>t</italic>) and infected cells <italic>y</italic>(<italic>t</italic>), respectively. The red line corresponds to <italic>x</italic>(<italic>t</italic>) and blue to <italic>y</italic>(<italic>t</italic>) in system (5), while the dashed green line corresponds to <italic>x</italic>(<italic>t</italic>) and the dashed magenta to <italic>y</italic>(<italic>t</italic>) in system (2). From (a), we can observe the number of infected cells of system (5) is less than that of system (2) in the endemic state. (b) The number of the healthy cells <italic>x</italic> <sub>1</sub> and infected cells <italic>y</italic> <sub>1</sub> in the endemic state of system (5) varying with the increasing of the apoptosis parameter α. From (b), we can observe that the healthy cells <italic>x</italic> <sub>1</sub> does not vary with the increasing of α, however, the infected cells <italic>y</italic> <sub>1</sub> decreases with the increasing of α.</p></caption><graphic xlink:href="tjbd-8-020-g001"/></fig> </p><p>The dynamical properties of system (5) shows that <italic>R</italic> <sub>0˜</sub>, which defines the average number of secondary infections generated by a typical infectious individual in a completely susceptible population in a steady demographic state, is a threshold parameter for the global stability of the infection-free equilibrium <italic>Ê</italic> <sub>0</sub>.</p><p>If <inline-formula><inline-graphic xlink:href="tjbd-8-020-m020.jpg"/></inline-formula>, then on average an infected individual produces less than one new infected individual over the course of its infectious period, and the infection cannot be established among the individuals. Conversely, if <inline-formula><inline-graphic xlink:href="tjbd-8-020-m021.jpg"/></inline-formula>, then each infected individual produces, on average, more than one new infection, and the disease can invade the population. Meanwhile, it is easy to see that in the case that <inline-formula><inline-graphic xlink:href="tjbd-8-020-m022.jpg"/></inline-formula>, the levels of infected cell <italic>y</italic> <sub>1</sub> inversely correlate with levels of apoptosis α. Hence, it is possible to speculate that apoptosis may contribute to eliminating cells which might prove harmful if they were to survive during viral infection. Indeed, as reported by O'Brien [<xref rid="CIT0037" ref-type="bibr">37</xref>], by the method of apoptosis induction, viruses can induce host cell death while limiting inflammatory and other immune responses. Furthermore, promoting apoptosis may be used for treating diseases associated with a failure to trigger appropriate apoptosis [<xref rid="CIT0037" ref-type="bibr">37</xref>].</p></sec></sec><sec id="S003"><label>3. </label><title>Negative impact of apoptosis in HIV infection</title><p>It is commonly believed that the depletion of CD4+T-cells is the principal reason for the collapse of immune system. T-cell loss appears to be due to direct destruction by the virus or due to defective T-cell generation. Apoptosis has been suggested as another mechanism responsible for T-cell depletion during the progression of the HIV infection and an extensive body of recent literature is supporting this hypothesis [<xref rid="CIT0001" ref-type="bibr">1</xref>,<xref rid="CIT0003" ref-type="bibr">3</xref>,<xref rid="CIT0005" ref-type="bibr">5</xref>,<xref rid="CIT0007" ref-type="bibr">7</xref>,<xref rid="CIT0014" ref-type="bibr">14</xref>,<xref rid="CIT0017" ref-type="bibr">17–22</xref>,<xref rid="CIT0033" ref-type="bibr">33</xref>,<xref rid="CIT0038" ref-type="bibr">38</xref>,<xref rid="CIT0048" ref-type="bibr">48</xref>,<xref rid="CIT0057" ref-type="bibr">57</xref>]. When the level of CD4+T-cells drops from around 1000 cells per mm<sup>3</sup> (the normal level for a healthy individual) to about 200 cells per mm<sup>3</sup>, the cell-mediated immunity is lost and the body becomes progressively more vulnerable to opportunistic infections. The T helper cells are unusual in the sense that they have no cytotoxic or phagocytic activity towards pathogens, and they do not kill infected cells or pathogens. Instead, T helper cells are involved in activating and directing other immune cells. It is noteworthy that activation of helper T-cells requires a much weaker activation stimulus than activation of cytotoxic T-cells. The role of CD4+T-cells in regulating and amplifying the immune response is vital, and a decline in their number results in deficits in humoral and cell-mediated immunity, opening an opportunity for opportunistic infections.</p><p>The Th immune response can be differentiated according to which type of the response is activated: Type 1 Th responses are critical in controlling intracellular infections via CTLs-mediated mechanisms (cell-mediated responses), whereas Type 2 helper responses are characterized by the activation of B-cells, which produce neutralizing (killing) antibodies (humoral immunity) [<xref rid="CIT0032" ref-type="bibr">32</xref>]. The factors that determine which type of response will be activated are not fully understood [<xref rid="CIT0032" ref-type="bibr">32</xref>]; however, in general, Th1 responses are more effective against intracellular pathogens (viruses, including HIV, and bacteria that inside host cells), while Th2 responses are more effective against extracellular bacteria, parasites and toxins. For this reason, in this paper we concentrate on the cell-mediated response, disregarding humoral immunity.</p><p>In order to describe the interactions between CD4+T-cells, HIV and CTLs response, it is usually assumed that the proliferation of immune response agents is proportional to their current concentration, the concentration of infected cells or a product (or a more complicated nonlinear function) of both (see [<xref rid="CIT0039" ref-type="bibr">39</xref>,<xref rid="CIT0051" ref-type="bibr">51</xref>] for comparison and discussion of models). An apparent deficiency of such a model is that it disregards the role of CD4+T-cells in immune response. This model deficiency is acceptable for the majority of viral infections, where the CD4+T-cell level remains approximately constant throughout the course of the infection. For HIV infection, however, CD4+T-cells are the target cells, and a decrease in their levels affects the efficacy of immune system. The experimental findings by Borrow <italic>et al.</italic> [<xref rid="CIT0011" ref-type="bibr">11</xref>] show that CD40 ligand-mediated, which is a glycoprotein that is transiently expressed at high levels on the surface of CD4+T-cells when they are activated, interactions are involved in the generation of CTLs. In accordance with experimental findings and to overcome the deficiency of disregarding the role of CD4+T-cells in immune response, in this paper we assume that the establishment of a lasting CTLs response depends on CD4+T-cell help, and that HIV impairs T helper cell function. Thus, the activation and proliferation of CTLs depend on three concentrations, namely on the current concentration of CTLs, the concentration of infected cells (antigen-presenting cells) and the concentration of the healthy CD4+T-cells. This assumption was firstly introduced by Wodarz and Nowak [<xref rid="CIT0052" ref-type="bibr">52</xref>]. Afterwards, Huang <italic>et al.</italic> [<xref rid="CIT0024" ref-type="bibr">24</xref>] discussed a possible mechanism which eventually enables HIV to break from immune control by the model with this assumption via continuous mutations and evolution.</p><p>Based on the above assumption, we consider the following model with CTLs immune responses: <disp-formula-group id="M0006"><disp-formula><graphic xlink:href="tjbd-8-020-e006.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Here, <italic>x</italic>(<italic>t</italic>), <italic>y</italic>(<italic>t</italic>) and <italic>z</italic>(<italic>t</italic>) are the concentration of healthy CD4+T-cells, the infected cells, antigen-specific CTLs, respectively. The infected cells are killed by CTLs at a rate <italic>pyz</italic>. The specific CTLs proliferates at a rate <italic>cxyz</italic> (which is proportional to the levels of healthy CD4+T helper cells, antigen-presenting cells and CTLs) and decays at a rate <italic>bz</italic>.</p><p>It is noteworthy that in the form <italic>xyz</italic>, the multiplier <italic>y</italic>(<italic>t</italic>) is responsible for antigen-presenting, describing the level of antigen-presenting cells, which is assumed to be proportional to the infected cells level. The assumption that the concentration of antigen-presenting cells is proportional to the infected cells concentration is fairly common in mathematical immunology.</p><p>There are three physically and biologically relevant types of non-negative equilibria, namely <list list-type="bullet"><list-item><p> <inline-formula><inline-graphic xlink:href="tjbd-8-020-m023.jpg"/></inline-formula>, infection-free or disease-free equilibrium, at which all individuals are susceptible and the population remains in the absence of disease. <italic>E</italic> <sub>0</sub> always exists.</p></list-item><list-item><p> <inline-formula><inline-graphic xlink:href="tjbd-8-020-m024.jpg"/></inline-formula>, immune-absence equilibrium, where <disp-formula id="UM0002"><graphic xlink:href="tjbd-8-020-u002.jpg" position="float" orientation="portrait"/></disp-formula> If <inline-formula><inline-graphic xlink:href="tjbd-8-020-m025.jpg"/></inline-formula>, then <italic>E</italic> <sub>1</sub> is nonnegative and therefore biologically meaningful.</p></list-item><list-item><p> <inline-formula><inline-graphic xlink:href="tjbd-8-020-m026.jpg"/></inline-formula>, interior immune-presence equilibrium, where <disp-formula id="UM0003"><graphic xlink:href="tjbd-8-020-u003.jpg" position="float" orientation="portrait"/></disp-formula> If <inline-formula><inline-graphic xlink:href="tjbd-8-020-m027.jpg"/></inline-formula>(and hence <inline-formula><inline-graphic xlink:href="tjbd-8-020-m028.jpg"/></inline-formula>), or <inline-formula><inline-graphic xlink:href="tjbd-8-020-m029.jpg"/></inline-formula>, then <italic>E</italic>* is nonnegative and therefore biologically meaningful.</p></list-item></list> </p><p>Here, <italic>R</italic> <sub>1</sub> is called the basic reproduction number of infected cells of system (6) (that is, <italic>R</italic> <sub>1</sub> is an average number of infected cells produced by a single infected cell introduced into entirely healthy environment), and <italic>Q</italic> <sub>1</sub> is called the basic reproduction number of immune response of system (6) (that is, <italic>Q</italic> <sub>1</sub> is an average number of CTLs produced by a single CTLs introduced into a system where healthy and infected cells are at their equilibrium levels). Therefore, depending on values of <italic>R</italic> <sub>1</sub> and <italic>Q</italic> <sub>1</sub>(α), system (6) has three equilibria.</p><p>The basic reproduction number of immune response <italic>Q</italic> <sub>1</sub> decreases with increasing α. Furthermore, there is α<sub>0</sub> (<inline-formula><inline-graphic xlink:href="tjbd-8-020-m030.jpg"/></inline-formula>) such that <inline-formula><inline-graphic xlink:href="tjbd-8-020-m031.jpg"/></inline-formula>; <italic>Q</italic> <sub>1</sub>>1 for all <inline-formula><inline-graphic xlink:href="tjbd-8-020-m032.jpg"/></inline-formula>, and <italic>Q</italic> <sub>1</sub><1 for all <inline-formula><inline-graphic xlink:href="tjbd-8-020-m033.jpg"/></inline-formula> (<xref rid="F0002" ref-type="fig">Figure 2</xref>(a)). <fig id="F0002" orientation="portrait" position="float"><label>Figure 2. </label><caption><p>(a), (b) and (c), respectively, show the basic reproduction number of immune response <italic>Q</italic> <sub>1</sub> with respect to α, the number of healthy cells <italic>x</italic>* and the number of infected cells <italic>y</italic>* in the immune-presence equilibrium with respect to α. From (a), we can derive <italic>Q</italic> <sub>1</sub>=1 at α=α<sub>0</sub>, <italic>Q</italic> <sub>1</sub>>1 when 0<α<α<sub>0</sub> and <italic>Q</italic> <sub>1</sub><1 when α<sub>0</sub><α<α¯. (b) and (c) show the number of healthy cells decreases to a stable value with the increasing of α satisfying α<sub>0</sub><α<α¯, while that of infected cells increases.</p></caption><graphic xlink:href="tjbd-8-020-g002"/></fig> </p><p>When <inline-formula><inline-graphic xlink:href="tjbd-8-020-m034.jpg"/></inline-formula>, namely <italic>Q</italic> <sub>1</sub>>1, the immune presence equilibrium <italic>E</italic>* exists. In this case, the concentration of infected cells <italic>y</italic>* grows with increasing α, whereas the concentration of healthy cells decreases with increasing α (see <xref rid="F0002" ref-type="fig">Figure 2</xref>(b) and (c)). This interval of α might correspond to the asymptomatic stage of HIV infection, where the viral load gradually grows and the healthy cells decrease. Hence the apoptosis further accelerates the progression of the HIV infection to AIDS in the ranges <inline-formula><inline-graphic xlink:href="tjbd-8-020-m035.jpg"/></inline-formula>. When <inline-formula><inline-graphic xlink:href="tjbd-8-020-m036.jpg"/></inline-formula>, which means <italic>Q</italic> <sub>1</sub><1 and <italic>R</italic> <sub>1</sub>>1, the immune absence equilibrium <italic>E</italic> <sub>1</sub> exists and the concentration of viruses decreases with the increasing of α for <inline-formula><inline-graphic xlink:href="tjbd-8-020-m037.jpg"/></inline-formula>.</p><sec id="S003-S2001"><label>3.1 </label><title>Stability analysis of three equilibria states</title><p>To study the local stability of the steady states of model (6), we linearize the system at <inline-formula><inline-graphic xlink:href="tjbd-8-020-m038.jpg"/></inline-formula> and obtain the following Jacobian matrix: <disp-formula id="UM0004"><graphic xlink:href="tjbd-8-020-u004.jpg" position="float" orientation="portrait"/></disp-formula> <list list-type="simple"><list-item><p>(i) The characteristic equation of system (6) at <italic>E</italic> <sub>0</sub> is <disp-formula-group id="M0007"><disp-formula><graphic xlink:href="tjbd-8-020-e007.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Roots of this equation are <italic>r</italic> <sub>1</sub>=−<italic>b, r</italic> <sub>2</sub>=−<italic>d</italic> and <inline-formula><inline-graphic xlink:href="tjbd-8-020-m039.jpg"/></inline-formula>. Hence, <italic>R</italic> <sub>1</sub><1 is sufficient to ensure the asymptotic stability of <italic>E</italic> <sub>0</sub>. When <italic>R</italic> <sub>1</sub>>1, Equation (7) has one positive real root, therefore <italic>E</italic> <sub>0</sub> is unstable. When <italic>R</italic> <sub>1</sub>=1, Equation (7) has a zero solution.</p></list-item><list-item><p>(ii) The characteristic equation of system (6) at <italic>E</italic> <sub>1</sub> is <disp-formula-group id="M0008"><disp-formula><graphic xlink:href="tjbd-8-020-e008.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Here root <inline-formula><inline-graphic xlink:href="tjbd-8-020-m040.jpg"/></inline-formula> is negative when <italic>Q</italic> <sub>1</sub><1, positive when <italic>Q</italic> <sub>1</sub>>1, and zero when <italic>Q</italic> <sub>1</sub>=1. Furthermore, when <italic>E</italic> <sub>1</sub> exists, all coefficients of quadratic equation <disp-formula-group id="M0009"><disp-formula><graphic xlink:href="tjbd-8-020-e009.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> are positive, and hence both roots of Equation (9) have negative real parts. That is, <italic>Q</italic> <sub>1</sub><1 ensures that all eigenvalues of Equation (8) have negative real parts. Hence the immune-absence equilibrium state <italic>E</italic> <sub>1</sub> is locally asymptotically stable if <italic>R</italic> <sub>1</sub>>1 and <italic>Q</italic> <sub>1</sub><1, and is unstable if <italic>Q</italic> <sub>1</sub>>1. When <italic>Q</italic> <sub>1</sub>=1, it is a critical case.</p></list-item><list-item><p>(iii) The characteristic equation of system (6) at <italic>E</italic>* is <disp-formula-group id="M0010"><disp-formula><graphic xlink:href="tjbd-8-020-e010.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula id="UM0005"><graphic xlink:href="tjbd-8-020-u005.jpg" position="float" orientation="portrait"/></disp-formula> </p></list-item></list> </p><p>Since <disp-formula id="UM0006"><graphic xlink:href="tjbd-8-020-u006.jpg" position="float" orientation="portrait"/></disp-formula> by the Routh–Hurwitz theorem, all roots of (10) have negative real parts, and the immune-presence equilibrium state <italic>E</italic>* is locally asymptotically stable if <italic>Q</italic> <sub>1</sub>>1.</p><statement id="E0001"><label>Theorem 3.1 </label><p>Consider system (6). <list list-type="simple"><list-item><p>(i) The infection-free equilibrium state <italic>E</italic> <sub>0</sub> is locally asymptotically stable if <italic>R</italic> <sub>1</sub><1, is unstable if <italic>R</italic> <sub>1</sub>>1 and if <italic>R</italic> <sub>1</sub>=1, it is a critical case.</p></list-item><list-item><p>(ii) The immune-absence equilibrium state <italic>E</italic> <sub>1</sub> is locally asymptotically stable if <italic>R</italic> <sub>1</sub>>1 and <italic>Q</italic> <sub>1</sub><1, is unstable if <italic>Q</italic> <sub>1</sub>>1 and if <italic>Q</italic> <sub>1</sub>=1, it is a critical case.</p></list-item><list-item><p>(iii) The immune-presence equilibrium state <italic>E</italic>* is locally asymptotically stable if <italic>Q</italic> <sub>1</sub>>1.</p></list-item></list> </p></statement></sec><sec id="S003-S2002"><label>3.2 </label><title>Numerical simulations and biological meanings</title><p>It is shown in Theorem 3.1 that <italic>R</italic> <sub>1</sub> is a threshold parameter for the local stability of the infection-free equilibrium <italic>E</italic> <sub>0</sub>. For 0<<italic>R</italic> <sub>1</sub><1, the infection-free equilibrium state <italic>E</italic> <sub>0</sub> is unique equilibrium state of the system (6) and is locally asymptotically stable (it can be conjectured that it is the global attractor of the system: phase trajectories with non-negative initial conditions eventually converge to this equilibrium state). The concentration of healthy CD4+T-cells reaches its maximum possible level λ/<italic>d</italic> in this equilibrium state. When <inline-formula><inline-graphic xlink:href="tjbd-8-020-m041.jpg"/></inline-formula>, the model (6) has two equilibrium states: in addition to the infection-free equilibrium state <italic>E</italic> <sub>0</sub>, an immune-absence equilibrium state <italic>E</italic> <sub>1</sub> appears. For these ranges of α, the infection-free equilibrium loses its stability and turns into a saddle point, whereas the immune-absence equilibrium <italic>E</italic> <sub>1</sub> is asymptotically stable. The immune-absence equilibrium state corresponds to a situation where both healthy and infected CD4+T-cells are present, while the antigen-specific CTLs response is not activated. When <inline-formula><inline-graphic xlink:href="tjbd-8-020-m042.jpg"/></inline-formula> (and hence <italic>Q</italic> <sub>1</sub>>1 holds), the system (6) has three equilibria: in addition to the infection-free equilibrium state <italic>E</italic> <sub>0</sub> and the immune-absence equilibrium state <italic>E</italic> <sub>1</sub>, an asymptotically stable immune-presence equilibrium state <italic>E</italic>* where all three components of system are present appears in the positive quadrant. Here we define α<sub>0</sub> as the immunodeficiency threshold, which is a threshold parameter for the stability of the immune-presence equilibrium <italic>E</italic>.</p><p>Now we fix the parameter values as <disp-formula-group id="M0011"><disp-formula><graphic xlink:href="tjbd-8-020-e011.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> By calculation, we have <italic>R</italic> <sub>1</sub>=12.5 and <italic>Q</italic> <sub>1</sub>=1.23. There exists the immune-presence equilibrium state <inline-formula><inline-graphic xlink:href="tjbd-8-020-m043.jpg"/></inline-formula> and <italic>E</italic> is asymptotically stable (see <xref rid="F0003" ref-type="fig">Figure 3</xref>). <fig id="F0003" orientation="portrait" position="float"><label>Figure 3. </label><caption><p>(a) and (b) The time evolution of the trajectory of system (6) for healthy cells <italic>x</italic>(<italic>t</italic>), infected cells <italic>y</italic>(<italic>t</italic>) and antigen-specific CTLs <italic>z</italic>(<italic>t</italic>). In (a), the red line corresponds to <italic>x</italic>(<italic>t</italic>), blue line corresponds to <italic>y</italic>(<italic>t</italic>) and green line corresponds to <italic>z</italic>(<italic>t</italic>). The panel (b) is the phase plane of system (6). They show that the immune-presence equilibrium state <italic>E</italic>* is locally asymptotically stable for <italic>R</italic> <sub>1</sub>>1 and <italic>Q</italic> <sub>1</sub>>1.</p></caption><graphic xlink:href="tjbd-8-020-g003"/></fig> </p><p>It is well known that, during the second phase (the chronic or asymptomatic phase) of HIV infection, the virus within-host diversity increases and the number of host CD4+T-cells decreases because they are the primary target of virus. Further, the third phase or AIDS phase is characterized by a dramatic loss in CD4+T-cells and a strong increase of viral load. Clinically, the onset of AIDS is defined as the time point at which the CD4+T-cells count in the blood falls below 200 per microlitre. In this paper, the characteristics of the chronic phase and AIDS phase are obviously displayed with the variation of the apoptosis parameter α and the activation-induced apoptosis may promote HIV infection procession, that is, the effect of activation-induced apoptosis may be detrimental to the immune system. As we mentioned above, α<sub>0</sub> is the immunodeficiency threshold. When <inline-formula><inline-graphic xlink:href="tjbd-8-020-m044.jpg"/></inline-formula>, that is, <italic>Q</italic> <sub>1</sub>>1, all solutions of system (6) converge to the immune-presence equilibrium state <italic>E</italic>, and the number of infected CD4+T-cells increases with increasing α, whereas the number of the healthy CD4+T-cells and the number of specific CTLs decrease with increasing α.</p><p>At <inline-formula><inline-graphic xlink:href="tjbd-8-020-m045.jpg"/></inline-formula>, that is <italic>Q</italic> <sub>1</sub>=1, the asymptotically stable immune-presence equilibrium state <italic>E</italic> disappears. When <inline-formula><inline-graphic xlink:href="tjbd-8-020-m046.jpg"/></inline-formula>, solutions of system (6) converge to the immune-absence equilibrium state <italic>E</italic> <sub>1</sub>, which is asymptotically stable. Furthermore, as α increases further, the number of infected cells gradually decreases. In terms of theory, it seems to be that when <inline-formula><inline-graphic xlink:href="tjbd-8-020-m047.jpg"/></inline-formula>, that is, the immune system has not been activated, the apoptosis may benefit to the infected host. In fact, many viruses have adopted strategies for blocking apoptosis [<xref rid="CIT0037" ref-type="bibr">37</xref>,<xref rid="CIT0056" ref-type="bibr">56</xref>]. Hence, the level of infected cells may not be decreasing.</p></sec></sec><sec id="S004"><label>4. </label><title>Effects of time delay in apoptosis</title><p>Due to the complexity of delay differential equations, many scientists do not include delays in their models. However, many biological processes have inherent delays and models including them may lead to additional insights into the study of complicated biological processes. Earlier work in modelling dynamics between viruses and immune responses was presented by Nowak and Bangham [<xref rid="CIT0036" ref-type="bibr">36</xref>]. Tam [<xref rid="CIT0046" ref-type="bibr">46</xref>] incorporated a discrete delay into one of Nowak and Bangham's models. Recently, Zhu and Zou [<xref rid="CIT0058" ref-type="bibr">58</xref>] considered the dynamical properties of an HIV-1 infection model with CTLs immune response and an intracellular delay.</p><p>In HIV-infected persons, both infected and uninfected CD4+T-cells undergo accelerated apoptosis <italic>in vitro</italic> and <italic>in vivo</italic>. Apoptosis occurs mainly in bystander uninfected cells, whereas productively HIV-infected cells have evolved strategies to prevent or delay apoptosis in the context of immune activation [<xref rid="CIT0003" ref-type="bibr">3</xref>]. Furthermore, it was determined that intracellular expression of HIV-1 Tat was able to delay Fas-mediated apoptosis and this effect was due to the presence of the second exon which is the protein, and causes a persistent infection in the host [<xref rid="CIT0031" ref-type="bibr">31</xref>]. It is speculated that time delay of the apoptosis of uninfected CD4+T-cells may play an important part in the progression of HIV infection. Thus, we incorporate time delay of apoptosis into HIV infection model (6) where the delay describes the latent period from the contacted cells with viruses to activating apoptosis and obtain the following model: <disp-formula-group id="M0012"><disp-formula><graphic xlink:href="tjbd-8-020-e012.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Here τ is called the intracellular delay describing the phase in which target cells are infected until uninfected cells start apoptosis in presence of infected cells.</p><p>Systems (12) and (6) have the same equilibrium states. The basic infection reproductive number and the basic reproduction number of immune response of system (12) are also, respectively, <disp-formula id="UM0007"><graphic xlink:href="tjbd-8-020-u007.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Next, we discuss the local stability of equilibrium states and the existence of Hopf bifurcations of system (12) when τ>0.</p><sec id="S004-S2001"><label>4.1 </label><title>Positivity and boundedness</title><p>We denote by <italic>C</italic> the Banach space of continuous functions <inline-formula><inline-graphic xlink:href="tjbd-8-020-m048.jpg"/></inline-formula> equipped with the sup-norm <disp-formula id="UM0008"><graphic xlink:href="tjbd-8-020-u008.jpg" position="float" orientation="portrait"/></disp-formula> where <inline-formula><inline-graphic xlink:href="tjbd-8-020-m049.jpg"/></inline-formula>. Further, let <disp-formula id="UM0009"><graphic xlink:href="tjbd-8-020-u009.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>The initial condition of system (12) is given as <disp-formula-group id="M0013"><disp-formula><graphic xlink:href="tjbd-8-020-e013.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <inline-formula><inline-graphic xlink:href="tjbd-8-020-m050.jpg"/></inline-formula>.</p><p>The following proposition establishes the non-negativity and boundedness of the solutions of (12) with (13).</p><statement id="E0002"><label>Proposition 4.1 </label><p>Let <inline-formula><inline-graphic xlink:href="tjbd-8-020-m051.jpg"/></inline-formula> be any solution of system (12). Then under the initial conditions (13), all solutions <inline-formula><inline-graphic xlink:href="tjbd-8-020-m052.jpg"/></inline-formula> are non-negative on [0,+∞) and ultimately bounded.</p></statement><p> <italic>Proof</italic> If <italic>x</italic>(<italic>t</italic>) were to lose its non-negativity on some local existence interval [0, <italic>T</italic>) for some constant <italic>T</italic>>0, there would be the first time at <italic>t</italic> <sub>1</sub>>0 such that <italic>x</italic>(<italic>t</italic> <sub>1</sub>)=0. By the first equation of (12) we have <inline-formula><inline-graphic xlink:href="tjbd-8-020-m053.jpg"/></inline-formula>. That means <italic>x</italic>(<italic>t</italic>)<0 for <inline-formula><inline-graphic xlink:href="tjbd-8-020-m054.jpg"/></inline-formula>, where ϵ is an arbitrarily small positive constant. This leads to a contradiction. It follows that <italic>x</italic>(<italic>t</italic>) is always positive. Further, from the second and the third equations in (12), we have, respectively, <disp-formula id="UM0010"><graphic xlink:href="tjbd-8-020-u010.jpg" position="float" orientation="portrait"/></disp-formula> Then, it is easy to see that <italic>y</italic>(<italic>t</italic>) and <italic>z</italic>(<italic>t</italic>) are non-negative on [0, <italic>T</italic>).</p><p>For <italic>t</italic>∈[0, <italic>T</italic>), we have from (12) that <inline-formula><inline-graphic xlink:href="tjbd-8-020-m055.jpg"/></inline-formula>. The well-known comparison principle implies that <italic>x</italic>(<italic>t</italic>) is bounded on [0, <italic>T</italic>), i.e. <inline-formula><inline-graphic xlink:href="tjbd-8-020-m056.jpg"/></inline-formula>. We again have from Equation (12) that on [0, <italic>T</italic>), <disp-formula-group id="M0014"><disp-formula><graphic xlink:href="tjbd-8-020-e014.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> which implies there exists <inline-formula><inline-graphic xlink:href="tjbd-8-020-m057.jpg"/></inline-formula> Since <inline-formula><inline-graphic xlink:href="tjbd-8-020-m058.jpg"/></inline-formula>, it has <disp-formula id="UM0011"><graphic xlink:href="tjbd-8-020-u011.jpg" position="float" orientation="portrait"/></disp-formula> Then by <disp-formula id="UM0012"><graphic xlink:href="tjbd-8-020-u012.jpg" position="float" orientation="portrait"/></disp-formula> we have <disp-formula-group id="M0015"><disp-formula><graphic xlink:href="tjbd-8-020-e015.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Thus there exists <inline-formula><inline-graphic xlink:href="tjbd-8-020-m059.jpg"/></inline-formula> Therefore by comparison principle, <italic>y</italic>(<italic>t</italic>) and <italic>z</italic>(<italic>t</italic>) are also bounded on [0, <italic>T</italic>). Boundedness of the solution <inline-formula><inline-graphic xlink:href="tjbd-8-020-m060.jpg"/></inline-formula> implies that the local existence interval [0, <italic>T</italic>) can be continued to <italic>T</italic>=+∞. This proves that the solution <inline-formula><inline-graphic xlink:href="tjbd-8-020-m061.jpg"/></inline-formula> exists and is non-negative on [0,+∞). The inequality (14) implies that <italic>x</italic>(<italic>t</italic>)+<italic>y</italic>(<italic>t</italic>) is ultimately bounded, and so are <italic>x</italic>(<italic>t</italic>), <italic>y</italic>(<italic>t</italic>). By the inequality (15), <italic>z</italic>(<italic>t</italic>) is also ultimately bounded. This completes the proof.</p><p>To study the local stability of the steady states of model (12), we linearize the system at <inline-formula><inline-graphic xlink:href="tjbd-8-020-m062.jpg"/></inline-formula> and obtain the following characteristic equation: <disp-formula id="UM0013"><graphic xlink:href="tjbd-8-020-u013.jpg" position="float" orientation="portrait"/></disp-formula> </p></sec><sec id="S004-S2002"><label>4.2 </label><title>Stability analysis of the infection-free equilibrium</title><p>The characteristic equation of system (12) at the infection-free equilibrium <italic>E</italic> <sub>0</sub> is <disp-formula-group id="M0016"><disp-formula><graphic xlink:href="tjbd-8-020-e016.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Roots of Equation (16) are <italic>r</italic> <sub>1</sub>=−<italic>b, r</italic> <sub>2</sub>=−<italic>d</italic> and <inline-formula><inline-graphic xlink:href="tjbd-8-020-m063.jpg"/></inline-formula>, hence <italic>R</italic> <sub>1</sub><1 is sufficient to ensure the asymptotic stability of <italic>E</italic> <sub>0</sub>. When <italic>R</italic> <sub>1</sub>>1, Equation (16) has one positive real root, thus <italic>E</italic> <sub>0</sub> is unstable. When <italic>R</italic> <sub>1</sub>=1, Equation (16) has a zero solution. Therefore, we have the following theorem.</p><statement id="E0003"><label>Theorem 4.2 </label><p>Consider system (12). <list list-type="simple"><list-item><p>(i) If <italic>R</italic> <sub>1</sub><1, then the infection-free equilibrium <italic>E</italic> <sub>0</sub> is locally asymptotically stable for any time delay τ>0, and the disease cannot invade the population.</p></list-item><list-item><p>(ii) If <italic>R</italic> <sub>1</sub>>1, then the infection-free equilibrium <italic>E</italic> <sub>0</sub> is unstable for any time delay τ>0, and invasion is always possible.</p></list-item></list> </p></statement></sec><sec id="S004-S2003"><label>4.3 </label><title>Stability analysis of the immune-absence equilibrium</title><p>The characteristic equation of system (12) at the immune-absence equilibrium <italic>E</italic> <sub>1</sub> is <disp-formula-group id="M0017"><disp-formula><graphic xlink:href="tjbd-8-020-e017.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula id="UM0014"><graphic xlink:href="tjbd-8-020-u014.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Here the root <inline-formula><inline-graphic xlink:href="tjbd-8-020-m064.jpg"/></inline-formula> is negative when <italic>Q</italic> <sub>1</sub><1, positive when <italic>Q</italic> <sub>1</sub>>1 and zero when <italic>Q</italic> <sub>1</sub>=1. Now, we consider the equation <disp-formula-group id="M0018"><disp-formula><graphic xlink:href="tjbd-8-020-e018.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>Since <italic>r</italic>=0 is not a root of Equation (18) for any τ>0 and all roots have negative real parts for τ=0, when <italic>Q</italic> <sub>1</sub><1, as the delay τ increases, the root of Equation (18) can only enter the right-half plane in the complex plane by crossing the imaginary axis except the origin. If <inline-formula><inline-graphic xlink:href="tjbd-8-020-m065.jpg"/></inline-formula> is a purely imaginary root of Equation (18), then <disp-formula id="UM0015"><graphic xlink:href="tjbd-8-020-u015.jpg" position="float" orientation="portrait"/></disp-formula> From which, we have that <disp-formula-group id="M0019"><disp-formula><graphic xlink:href="tjbd-8-020-e019.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Noting that <inline-formula><inline-graphic xlink:href="tjbd-8-020-m066.jpg"/></inline-formula>, it follows that <disp-formula-group id="M0020"><disp-formula><graphic xlink:href="tjbd-8-020-e020.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula id="UM0016"><graphic xlink:href="tjbd-8-020-u016.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Letting <italic>s</italic>=ω<sup>2</sup> gives <disp-formula-group id="M0021"><disp-formula><graphic xlink:href="tjbd-8-020-e021.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Denote <disp-formula id="UM0017"><graphic xlink:href="tjbd-8-020-u017.jpg" position="float" orientation="portrait"/></disp-formula> Equation (21) has at least one positive real root in two circumstances. <list list-type="bullet"><list-item><p>If <inline-formula><inline-graphic xlink:href="tjbd-8-020-m067.jpg"/></inline-formula>, since the leading coefficient is positive, then there is a positive real root.</p></list-item><list-item><p>If <inline-formula><inline-graphic xlink:href="tjbd-8-020-m068.jpg"/></inline-formula>, the roots of Equation (21) are <disp-formula id="UM0018"><graphic xlink:href="tjbd-8-020-u018.jpg" position="float" orientation="portrait"/></disp-formula> and there are two simple positive roots if and only if <italic>A</italic> <sub>1</sub><0 and <inline-formula><inline-graphic xlink:href="tjbd-8-020-m069.jpg"/></inline-formula>.</p></list-item></list> </p><p>Without loss of generality, we assume that Equation (21) has two positive roots denoted by <italic>s</italic> <sub>1</sub> and <italic>s</italic> <sub>2</sub>, respectively. Then Equation (20) has two positive roots <inline-formula><inline-graphic xlink:href="tjbd-8-020-m070.jpg"/></inline-formula>. From Equation (19), we have <disp-formula id="UM0019"><graphic xlink:href="tjbd-8-020-u019.jpg" position="float" orientation="portrait"/></disp-formula> where <italic>k</italic>=1, 2; <italic>j</italic>=0, 1, … . Then <inline-formula><inline-graphic xlink:href="tjbd-8-020-m071.jpg"/></inline-formula> is a pair of purely imaginary roots of Equation (20) with <inline-formula><inline-graphic xlink:href="tjbd-8-020-m072.jpg"/></inline-formula>.</p><p>Define <disp-formula id="UM0020"><graphic xlink:href="tjbd-8-020-u020.jpg" position="float" orientation="portrait"/></disp-formula> where <disp-formula id="UM0021"><graphic xlink:href="tjbd-8-020-u021.jpg" position="float" orientation="portrait"/></disp-formula> Denote <inline-formula><inline-graphic xlink:href="tjbd-8-020-m073.jpg"/></inline-formula>, which is a positive root of Equation (21).</p><p>Assume <list list-type="simple"><list-item><p>(<italic>H</italic> <sub>1</sub>) <italic>A</italic> <sub>0</sub><0 or <italic>A</italic> <sub>0</sub>>0, <italic>A</italic> <sub>1</sub><0 and <inline-formula><inline-graphic xlink:href="tjbd-8-020-m074.jpg"/></inline-formula>;</p></list-item><list-item><p>(<italic>H</italic> <sub>2</sub>) <italic>h</italic>′(<italic>s</italic> <sub>0</sub>)>0.</p></list-item></list> </p><p>By implicit differentiation of Equation (18) with respect to τ, we obtain <disp-formula id="UM0022"><graphic xlink:href="tjbd-8-020-u022.jpg" position="float" orientation="portrait"/></disp-formula> Hence <disp-formula id="UM0023"><graphic xlink:href="tjbd-8-020-u023.jpg" position="float" orientation="portrait"/></disp-formula> If the hypothesis (<italic>H</italic> <sub>2</sub>) is satisfied and noting Equation (19), we have <disp-formula id="UM0024"><graphic xlink:href="tjbd-8-020-u024.jpg" position="float" orientation="portrait"/></disp-formula> Then <disp-formula id="UM0025"><graphic xlink:href="tjbd-8-020-u025.jpg" position="float" orientation="portrait"/></disp-formula> Therefore, for delay τ<τ<sub>0</sub>, roots of Equation (18) have negative real parts. When τ=τ<sub>0</sub>, there exist a pair of purely imaginary roots for Equation (18). When τ>τ<sub>0</sub>, Equation (18) has at least one positive real root.</p><p>From the above analysis, we have the following theorem.</p><statement id="E0004"><label>Theorem 4.3 </label><p>Consider system (12), when <italic>R</italic> <sub>1</sub>>1, <italic>Q</italic> <sub>1</sub><1 and the conditions (<italic>H</italic> <sub>1</sub>) and (<italic>H</italic> <sub>2</sub>) hold, <list list-type="simple"><list-item><p>(i) if <inline-formula><inline-graphic xlink:href="tjbd-8-020-m075.jpg"/></inline-formula> then the immune-absence equilibrium state <italic>E</italic> <sub>1</sub> is locally asymptotically stable;</p></list-item><list-item><p>(ii) if <inline-formula><inline-graphic xlink:href="tjbd-8-020-m076.jpg"/></inline-formula> then the immune-absence equilibrium state <italic>E</italic> <sub>1</sub> is unstable and system (12) undergoes a Hopf bifurcation at <italic>E</italic> <sub>1</sub> when τ=τ<sub>0</sub>.</p></list-item></list> </p></statement></sec><sec id="S004-S2004"><label>4.4 </label><title>Stability analysis of the immune-presence equilibrium</title><p>The characteristic equation of system (12) at the immune-presence equilibrium <italic>E</italic>* is <disp-formula-group id="M0022"><disp-formula><graphic xlink:href="tjbd-8-020-e022.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula id="UM0026"><graphic xlink:href="tjbd-8-020-u026.jpg" position="float" orientation="portrait"/></disp-formula> It is found from Theorem 3.1 that when τ=0, the roots of Equation (22) have negative real roots. By the continuous dependence of roots of Equation (22) on the parameters, it follows that there exists a τ<sub>0</sub>>0 such that for <inline-formula><inline-graphic xlink:href="tjbd-8-020-m077.jpg"/></inline-formula>, all the roots of Equation (22) will satisfy Re(<italic>r</italic>)<0 and when τ=τ<sub>0</sub>, Re(<italic>r</italic>)=0.</p><p>To determine this τ<sub>0</sub>, we assume <inline-formula><inline-graphic xlink:href="tjbd-8-020-m078.jpg"/></inline-formula> is a solution of Equation (22) and separate real and imaginary parts, and obtain <disp-formula id="UM0027"><graphic xlink:href="tjbd-8-020-u027.jpg" position="float" orientation="portrait"/></disp-formula> From which, we have that <disp-formula-group id="M0023"><disp-formula><graphic xlink:href="tjbd-8-020-e023.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula id="UM0028"><graphic xlink:href="tjbd-8-020-u028.jpg" position="float" orientation="portrait"/></disp-formula> Noting that <inline-formula><inline-graphic xlink:href="tjbd-8-020-m079.jpg"/></inline-formula>, it follows that <disp-formula-group id="M0024"><disp-formula><graphic xlink:href="tjbd-8-020-e024.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula id="UM0029"><graphic xlink:href="tjbd-8-020-u029.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Letting <italic>s</italic>=ω<sup>2</sup> gives <disp-formula-group id="M0025"><disp-formula><graphic xlink:href="tjbd-8-020-e025.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Denote <disp-formula id="UM0030"><graphic xlink:href="tjbd-8-020-u030.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>We utilized the Sturm chain which is proposed in [<xref rid="CIT0016" ref-type="bibr">16</xref>] to extend the sufficient conditions to preserve that the polynomial (25) has at least one positive real root. Since <italic>h</italic> <sub>0</sub>>0 and the polynomial (25) is of odd degree, we are guaranteed that Equation (25) has a negative real root. From Theorem 1 of [<xref rid="CIT0016" ref-type="bibr">16</xref>], we obtain the sufficient conditions to ensure one positive real root for the polynomial (25). That is, <list list-type="bullet"><list-item><p> <italic>h</italic> <sub>0</sub>, <italic>h</italic> <sub>1</sub> and <italic>h</italic> <sub>2</sub> are not all positive;</p></list-item><list-item><p> <inline-formula><inline-graphic xlink:href="tjbd-8-020-m080.jpg"/></inline-formula>;</p></list-item><list-item><p> <inline-formula><inline-graphic xlink:href="tjbd-8-020-m081.jpg"/></inline-formula>.</p></list-item></list> </p><p>The above conditions can also ensure that Equation (25) has two positive real roots. We denote them by <italic>s˜</italic> <sub>1</sub> and <italic>s˜</italic> <sub>2</sub>, respectively. Then Equation (24) has two positive roots <inline-formula><inline-graphic xlink:href="tjbd-8-020-m082.jpg"/></inline-formula>. From Equation (23), we have <disp-formula id="UM0031"><graphic xlink:href="tjbd-8-020-u031.jpg" position="float" orientation="portrait"/></disp-formula> where <italic>k</italic>=1, 2; <italic>j</italic>=0, 1, … . Then <inline-formula><inline-graphic xlink:href="tjbd-8-020-m083.jpg"/></inline-formula> is a pair of purely imaginary roots of Equation (24) with <inline-formula><inline-graphic xlink:href="tjbd-8-020-m084.jpg"/></inline-formula>.</p><p>Define <disp-formula id="UM0032"><graphic xlink:href="tjbd-8-020-u032.jpg" position="float" orientation="portrait"/></disp-formula> where <disp-formula id="UM0033"><graphic xlink:href="tjbd-8-020-u033.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Denote <inline-formula><inline-graphic xlink:href="tjbd-8-020-m085.jpg"/></inline-formula>, which is a positive root of Equation (25).</p><p>Assume <list list-type="simple"><list-item><p>(<italic>H</italic> <sub>3</sub>) <italic>h</italic> <sub>0</sub>, <italic>h</italic> <sub>1</sub> and <italic>h</italic> <sub>2</sub> are not all positive;</p></list-item><list-item><p>(<italic>H</italic> <sub>4</sub>) <inline-formula><inline-graphic xlink:href="tjbd-8-020-m086.jpg"/></inline-formula>; <inline-formula><inline-graphic xlink:href="tjbd-8-020-m087.jpg"/></inline-formula>;</p></list-item><list-item><p>(<italic>H</italic> <sub>5</sub>) <inline-formula><inline-graphic xlink:href="tjbd-8-020-m088.jpg"/></inline-formula>.</p></list-item></list> </p><p>By implicit differentiation of Equation (22) with respect to τ, we obtain <disp-formula id="UM0034"><graphic xlink:href="tjbd-8-020-u034.jpg" position="float" orientation="portrait"/></disp-formula> Hence, <italic>q</italic> <sub>2</sub>=0 <disp-formula id="UM0035"><graphic xlink:href="tjbd-8-020-u035.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>If the hypothesis (<italic>H</italic> <sub>3</sub>) is satisfied and noting Equation (23), we have <disp-formula id="UM0036"><graphic xlink:href="tjbd-8-020-u036.jpg" position="float" orientation="portrait"/></disp-formula> Then <disp-formula id="UM0037"><graphic xlink:href="tjbd-8-020-u037.jpg" position="float" orientation="portrait"/></disp-formula> Therefore, when the delay <inline-formula><inline-graphic xlink:href="tjbd-8-020-m089.jpg"/></inline-formula>, the roots of Equation (22) have negative real parts. When <inline-formula><inline-graphic xlink:href="tjbd-8-020-m090.jpg"/></inline-formula>, there exist a pair of purely imaginary roots for Equation (22), and other roots have no negative real parts.</p><p>From the above analysis, we have the following theorem:</p><statement id="E0005"><label>Theorem 4.4 </label><p>Consider system (12), when <italic>Q</italic> <sub>1</sub>>1 and conditions (<italic>H</italic> <sub>3</sub>)–(<italic>H</italic> <sub>5</sub>) hold, we have <list list-type="simple"><list-item><p>(i) if <inline-formula><inline-graphic xlink:href="tjbd-8-020-m091.jpg"/></inline-formula> then the immune-presence equilibrium <italic>E</italic>* is asymptotically stable;</p></list-item><list-item><p>(ii) if <inline-formula><inline-graphic xlink:href="tjbd-8-020-m092.jpg"/></inline-formula> then the immune-presence equilibrium <italic>E</italic>* is unstable and system (12) undergoes a Hopf bifurcation at <italic>E</italic>* when <inline-formula><inline-graphic xlink:href="tjbd-8-020-m093.jpg"/></inline-formula>.</p></list-item></list> </p></statement></sec><sec id="S004-S2005"><label>4.5 </label><title>Numerical simulations</title><p>By using suitable numerical methods, we demonstrate that stability change would occur at the branch of the equilibria <italic>E</italic> <sub>1</sub> and <italic>E</italic>* of system (12), leading to Hopf bifurcations.</p><p>First, we fix the parameter values as <disp-formula-group id="M0026"><disp-formula><graphic xlink:href="tjbd-8-020-e026.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> By direct calculation, we have <italic>R</italic> <sub>1</sub>=6.67, <italic>Q</italic> <sub>1</sub>=0.57, <italic>A</italic> <sub>1</sub>=−1.0222, <italic>A</italic> <sub>0</sub>=0.2408, <inline-formula><inline-graphic xlink:href="tjbd-8-020-m094.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-020-m095.jpg"/></inline-formula>. Note that the conditions <italic>H</italic> <sub>1</sub> and <italic>H</italic> <sub>2</sub> hold. We obtain <inline-formula><inline-graphic xlink:href="tjbd-8-020-m096.jpg"/></inline-formula> by calculation, and the simulation displays <italic>E</italic> <sub>1</sub>=(30, 1.8889, 0) is asymptotically stable when τ=1.4 (<xref rid="F0004" ref-type="fig">Figure 4</xref>(a)), while it loses its stability and a periodic solution emerges when τ=1.6 (<xref rid="F0004" ref-type="fig">Figure 4</xref>(b)). <fig id="F0004" orientation="portrait" position="float"><label>Figure 4. </label><caption><p>The diagrams show the time evolution of trajectory of system (12) including the time delay of apoptosis for healthy cells <italic>x</italic>(<italic>t</italic>), infected cells <italic>y</italic>(<italic>t</italic>) and antigen-specific CTLs <italic>z</italic>(<italic>t</italic>) for <italic>R</italic> <sub>1</sub>>1 and <italic>Q</italic> <sub>1</sub><1. In (a), it is shown that when τ<τ<sub>0</sub>, the solution of Equation (12) tends to the immune-absence equilibrium state <italic>E</italic> <sub>1</sub> for the given initial value. In (b), it is shown that when τ>τ<sub>0</sub>, the solution of Equation (12) no longer tends to the equilibrium state <italic>E</italic> <sub>1</sub>. Instead, a periodic solution appears.</p></caption><graphic xlink:href="tjbd-8-020-g004"/></fig> </p><p>Next we choose the parameter values given by (11). We have <italic>R</italic> <sub>1</sub>=12.5, <italic>Q</italic> <sub>1</sub>=1.23 and three equilibrium states exist. In this case, we have <inline-formula><inline-graphic xlink:href="tjbd-8-020-m097.jpg"/></inline-formula>, <italic>h</italic> <sub>1</sub>=26.8773, <italic>h</italic> <sub>0</sub>=0.3468, <inline-formula><inline-graphic xlink:href="tjbd-8-020-m098.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-020-m099.jpg"/></inline-formula>, which make the conditions (<italic>H</italic> <sub>3</sub>−<italic>H</italic> <sub>5</sub>) hold. By Theorem 4.4, the first Hopf bifurcation occur at <inline-formula><inline-graphic xlink:href="tjbd-8-020-m100.jpg"/></inline-formula>. We use a Matlab package called DDE-BIFTOOL to draw the graph of Hopf branches in <xref rid="F0005" ref-type="fig">Figure 5</xref>. We use a Matlab package DDE23 to find numerical solution to system (12) with τ=0.45 and 0.6. As shown in <xref rid="F0006" ref-type="fig">Figure 6</xref> we observe that <inline-formula><inline-graphic xlink:href="tjbd-8-020-m101.jpg"/></inline-formula> is asymptotically stable when τ=0.45, while it loses its stability and a periodic solution emerges when τ=0.6. Comparing <xref rid="F0004" ref-type="fig">Figure 4</xref> with <xref rid="F0003" ref-type="fig">Figure 3</xref> we see that increasing the time delay brings rich dynamical behaviour. <fig id="F0005" orientation="portrait" position="float"><label>Figure 5. </label><caption><p>Bifurcation diagram of system (12) with parameter values given in (11).</p></caption><graphic xlink:href="tjbd-8-020-g005"/></fig> <fig id="F0006" orientation="portrait" position="float"><label>Figure 6. </label><caption><p>(a) and (b) The time evolution of trajectory of system (12) including the time delay of apoptosis for healthy cells <italic>x</italic>(<italic>t</italic>), infected cells <italic>y</italic>(<italic>t</italic>) and antigen-specific CTLs <italic>z</italic>(<italic>t</italic>), for <italic>R</italic> <sub>1</sub>>1 and <italic>Q</italic> <sub>1</sub>>1. In (a), it is shown that when τ<τ˜<sub>0</sub>, the solution of system (12) tends to the equilibrium <italic>E</italic>* for the given initial value. In (b), it is shown that when τ>τ˜<sub>0</sub>, the solution of Equation (12) no longer tends to the equilibrium <italic>E</italic>*. Instead, a periodic solution appears.</p></caption><graphic xlink:href="tjbd-8-020-g006"/></fig> </p></sec></sec><sec id="S005"><label>5. </label><title>Discussion and conclusion</title><p>As a result of HIV infection, CD4+T-cells loss may be due to direct viral killing infected cells or killing infected CD4+T-cells by CTLs. Recently apoptosis has been suggested to be another mechanism responsible for CD4+T-cells depletion. In this paper, we propose several models which include the activation-induced apoptosis phenomenon and analyse the influence of apoptosis on viral infection dynamics, especially HIV infection dynamics.</p><p>In model (3), we use the constant <italic>d</italic> <sub>α</sub> to describe the death rate of uninfected cells related to activation-induced apoptosis. The analysis for system (3) shows that there exists a threshold <italic>d</italic> <sub>0</sub> when <italic>R</italic> <sub>0</sub>=1. When <italic>d</italic> <sub>α</sub> is below this threshold, virus is able to invade and persist within a host. Once <italic>d</italic> <sub>α</sub> becomes more than this threshold, the level of infected cells gradually decreases with increasing of <italic>d</italic> <sub>α</sub> and finally infected cells tend to vanish. Further, we postulate that the concentration of chemical messengers released by infected cells and inducing apoptosis is proportional to that of the infected cells and obtain model (5). The analysis for system (5) illustrates that when the basic reproduction number of virus is more than 1, the endemic equilibrium is stable and the number of infected cells at this equilibrium decreases with the increasing of apoptosis parameter. These theoretical analyses for models (3) and (5) illustrate that an increase in apoptosis can bring about a fall in the number of infected cells, which implies that apoptosis plays a positive role in preventing virus infection.</p><p>In HIV infection, immune response has been shown to be universal and necessary to eliminate or control the disease. To recover from a viral infection, the CTLs will clear away the infected cells to prevent further viral replications. Since CTLs play a critical role in antiviral defense by attacking cells infected with HIV, we consider the influence of apoptosis in HIV infection model with the immune response. Here, we assume that the activation of proliferation of CTLs depends on the concentration of CTLs, the concentration of infected CD4+T-cells and the concentration of healthy CD4+T-cells, and we obtain model (6). According to the qualitative analysis for model (6), we find that there is a threshold, namely the immunodeficiency threshold α<sub>0</sub>, which is specific for HIV dynamics. When the apoptosis parameter is below this threshold, the immune system is effective in virus control, and all solutions of system (6) converge to the immune-presence equilibrium in which the number of infected CD4+T-cells increases with rise of the apoptosis parameter (although the concentration of healthy CD4+T-cells and CTLs decrease). Furthermore, once the apoptosis parameter becomes beyond this threshold, the immune response vanishes and the level of infected cell decreases with the raise of apoptosis parameter. From the above results, we find that the apoptosis phenomenon may accelerate the depletion of CD4+T-cells and destroy the immune response for the ranges <inline-formula><inline-graphic xlink:href="tjbd-8-020-m102.jpg"/></inline-formula>, whereas for <inline-formula><inline-graphic xlink:href="tjbd-8-020-m103.jpg"/></inline-formula>, the apoptosis may boost the rate of decline in infected cells. Therefore, the activation-induced apoptosis may aggravate or comfort the HIV infection for some ranges of apoptosis parameter.</p><p>In reality, during HIV infection, the intracellular expression of full-length Tat is able to delay Fas-mediated apoptosis, hence there exists a time lag from viruses contacted to activating apoptosis. Hence, we incorporate time delay of apoptosis into model (6). We study the impact of the time delay in apoptosis by analysing model (12). By using a combination of bifurcation theory and numerical simulations, we find that there exists the limit cycle with increased time delay. By associated transcendental characteristic equations for the immune-free equilibrium of system (12) with the delay, we show that there exists the critical value denoted by τ<sub>0</sub>. When τ is less than this critical value, the infection-absence equilibrium of system (12) is locally asymptotically stable; when τ is more than τ<sub>0</sub>, this equilibrium is unstable and a Hopf bifurcation occurs at τ=τ<sub>0</sub>. When <italic>Q</italic> <sub>1</sub> is larger than 1, we obtain results for the immune-present equilibrium similar to those we obtained for the immune-free equilibrium. Therefore, the time delay in apoptosis could lead to very complicated dynamics including stable periodic solution, which gives additional insights into the study of the apoptosis effect on the HIV infection.</p><p>The principal conclusion in this paper is that apoptosis is helpful for inhibition of the virus level by cells itself, while to some extent it is dangerous for HIV infection as it may cause fall in CD4+T-cells which could lead to collapse of CTLs immune system. Furthermore apoptosis delay can give rise to a viral oscillation in the host.</p></sec>
Cell mechanics: from single scale-based models to multiscale modelling	Could not extract abstract	<contrib contrib-type="author"><name><surname>Stadtländer</surname><given-names>Christian T.K.-H.</given-names></name><xref ref-type="aff" rid="AF1"> <sup>a</sup> </xref><xref ref-type="corresp" rid="cor1"> <sup>*</sup> </xref></contrib><aff id="AF1"><label><sup>a</sup></label><institution><named-content content-type="institution-name">St. Paul</named-content></institution>, <named-content content-type="state">MN</named-content>, <country>USA</country></aff>	Journal of Biological Dynamics	<p>Biological cells are considered as the building units of life. It was Robert Hooke who coined the word <italic>cell</italic> in histological description of cork and other specimens using coarse optical lenses, as published in his <italic>Micrographia</italic> in 1665 [<xref rid="CIT0010" ref-type="bibr">10</xref>]. A few years later, Anthonie van Leeuwenhoek used a self-built microscope to observe living microorganisms which he called <italic>animalcules</italic> (‘little animals’) which are known today as protozoa, bacteria, and other forms of life [<xref rid="CIT0006" ref-type="bibr">6</xref>]. Since these early years of observation, researchers have come a long way to decipher many features of life, ranging from the structural and functional organization of single- and multi-cell organisms to entire populations, communities, and ecosystems. Researchers recognized that the mathematical language is also a powerful tool for explaining biological phenomena. More specifically, the dynamics of biological systems can be described, simulated, and often predicted in the form of mathematical formulas, algorithms, and computer models. In recent years, computational biology (which includes a quantitative data analysis approach, theoretical methods, mathematical modelling, and computational simulation techniques) has become essential in biological research. This is because modern experimental methods (e.g. advanced imaging techniques, biochemical, genetic, immunological, and molecular analyses) are generating massive data sets at an unprecedented rate, detail, and complexity.</p><p>The book by Chauvière, Preziosi, and Verdier focuses on an area of research known as <italic>cell mechanics</italic>. Here, investigators attempt to illuminate how cells sense and identify, interact with each other, and respond to the physical properties of the environment in which they live. Cell mechanics, which draws investigators from various disciplines such as biology, physics, mathematics, computer science, and engineering, has proven to be indispensable for elucidating mechanisms involved in many aspects of cell dynamics, including cell structure, movement, and adhesion [<xref rid="CIT0002" ref-type="bibr">2–4</xref>] embryogenesis and embryonics [<xref rid="CIT0007" ref-type="bibr">7</xref>], wound healing and tissue engineering [<xref rid="CIT0001" ref-type="bibr">1</xref>,<xref rid="CIT0011" ref-type="bibr">11</xref>], viable cell growth arrest and apoptosis [<xref rid="CIT0005" ref-type="bibr">5</xref>,<xref rid="CIT0008" ref-type="bibr">8</xref>], as well as the development of diseases such as asthma, glaucoma, and cancer [<xref rid="CIT0009" ref-type="bibr">9</xref>].</p><p>The editors point out in the preface that the book is the result of many years of collaboration among researchers of the European Research and Training Network on a topic entitled ‘Modeling, Mathematical Methods and Computer Simulation of Tumour Growth and Therapy’. The book contains reviews that describe experimental and applied mathematical approaches to illuminate how cells behave at various scales. The editors emphasize that it became clear while researching these biological phenomena from different points of view that ‘multiscale problems are ubiquitous and fundamental in cell mechanics’.</p><p> <italic>Cell Mechanics: From Single Scale-Based Models to Multiscale Modelling</italic> is divided into four major parts: ‘From Subcellular to Cellular Properties’ (Part I), ‘Single Cell Migration Modelling’ (Part II), ‘Mechanical Effects of Environment on Cell Behaviour’ (Part III), and ‘From Cellular to Multicellular Models’ (Part IV). These parts contain a total of 15 chapters which have been written by 50 contributors. In addition to the chapters, the preface, and information about the editors and contributors, the book also contains a table of contents and an index which I both tested and found to be functional.</p><p>In the first three chapters, which make up Part I, the authors describe issues related to the movement of cells. Chapter 1 introduces the reader to the structural and mechanical properties of the cytoskeletal network which consists of a dynamic assembly of macromolecules forming actin filaments, intermediate filaments, and microtubules, and interacting with or controlled by proteins, cross-linkers, and molecular motors. The authors review data from rheology (stress–strain relationship) studies of single living cells (e.g. myoblasts, epithelial cells, macrophages) at different time and length scales, and present a semi-phenomenological model for cell rheology that can predict the mechanical response of a cell submitted to a controlled stress. In the second chapter, the authors point out that most living cells are able to perform a directed motion, for example, by swimming in a liquid environment, by crawling on a solid support, or by squeezing through a three-dimensional matrix of fibres. They look at the actin-based cell propulsion system and describe the interplay between material properties (the physical mechanisms) and growth processes (i.e. actin polymerization and organization). The third chapter is about cell motility and cancer. The authors describe tumourigenesis as a multistep process and review tumour suppressor genes and the molecular mechanisms involved in cell migration, adhesion, and invasion. They discuss methods for <italic>in vitro</italic> cell migration studies, including two- and three-dimensional migration assays.</p><p>Part II (Chapters 4–6) is about the mathematical modelling of single cell migration events. More specifically, the authors of the fourth chapter review the biology of cell polarization and locomotion/migration, events that play a central role in the development and maintenance of tissues in multicellular organisms. They mention that the polarization of a cell defines the direction of migration as well as the cell division axis, and is thus responsible for the three-dimensional structures of tissues, organs, and the whole organism. The authors review various models (e.g. the cellular Potts model and the ‘reactive flow’ model) and then propose a generalized continuum model useful for studying the coupling of cytoplasm (cytoskeletal) and adhesion dynamics. They demonstrate that relatively simple laws for the small-scale mechanics and kinetics of events (e.g. filament polymerization, pushing/sliding, binding/pulling on adhesion sites) can be combined into a nonlinearly coupled system of differential equations which model the polarization and migration behaviour on the large-scale cell level. In the fifth chapter, the authors describe the movement via lamellipodia, which are thin membrane-bound, mostly filamentous actin-containing sheets of cytoplasm that cells protrude at the front end and retract at the rear. They show a multiphase evolution model for lamellipodia with arbitrary shape that allows relating computationally the structure and dynamics of the actin network to the traction forces and shape changes that constitute the fascinating amoeboid movement. The authors of Chapter 6 reiterate that the motility/migration of living cells is a complex and highly integrated process in which cytoskeletal assembly, actin turnover, the contractility of actomyosin, and adhesion dynamics are all closely interlinked and regulated by the properties of the extracellular environment. They present a computational framework (numerical simulations), constructed from an existing mathematical model of fibroblast cell deformations, for the investigation of the coupling of these processes.</p><p>The third part contains four chapters in which the authors discuss cell–environment interactions from the mechanical point of view. In Chapter 7, a model is described which demonstrates how a flow field under varying shear stresses can affect the dynamic behaviour of a cell (represented by an adhesive microbead) in close contact with a wall. The authors perform microscale calculations of bond-formation, -advection, and -breakage to predict macroscale cell motion. They identify three different regimes which may overlap for some parameter values: (1) when the sphere adheres to the wall, the bonds prevent both sliding and translational motions; (2) when the sphere tank-treads on the wall, the bonds prevent sliding motion; and (3) when the sphere is free from adhesive forces, most of the bonds are broken. The eighth chapter provides an interesting discussion about our understanding of adhesion sites as mechanosensitive cellular elements. Adhesion sites are considered as ‘clusters of membrane-associated proteins constituting a discrete physiochemical link between the cytoskeleton and the extracellular environment’. The authors discuss the physicochemical mechanism(s) by which cells sense and adapt to the mechanical properties of the extracellular environment.</p><p>Chapter 9 is about tumour cell migration. This process, which occurs during the formation of metastases, requires a link between adhesion anchoring and the reorganization of the cytoskeleton. As an experimental model, the authors use migrating T24 cancer cells on polyacrylamide gel substrates with different stiffnesses and discuss the results in light of observations that such cancer cells exert less traction than other cell types. The tenth chapter is entitled ‘Single Cell Imaging of Calcium in Response to Mechanical Stimulation’. The authors discuss the importance of calcium ions (Ca<sup>2+</sup>) in the regulation of many cellular activities and provide design strategies for Ca<sup>2+</sup> imaging. Furthermore, they describe how the mechanical stimuli or physical forces can be perceived by cells and transduced into biochemical responses – a process known as mechano-transduction.</p><p>The final part (Part IV) is entitled ‘From Cellular to Multicellular Models’. The authors of Chapter 11 describe the mathematical framework for modelling the migration of a cell population (i.e. the collective movement of cells) in the extracellular matrix (ECM). They look at the heterogeneity/anisotropy of the ECM medium and examine events such as chemotaxis, haptotaxis, and repellent quorum sensing migration. They explain that cell migration is an essential feature of many physiological and pathological phenomena in biology, such as embryonic development, tissue homeostasis, immune response, and wound healing, as well as metastasis dissemination and tumour invasion. The authors point out that the characteristics of the spreading of a cell population can vary greatly and is depending on both the local micro-environment and the function of the migrating cell(s).</p><p>The following three chapters are about mathematical modelling of cell populations as it relates to developmental biology and cancer (Chapter 12), embryogenesis (Chapter 13), and cancer invasion (Chapter 14). In more detail, the authors of the twelfth chapter emphasize that ‘From the earliest embryonic stages to the complexity of the adult, the ability of cell populations to adhere to each other or the surrounding ECM is of critical importance to the survival of the organism’. They demonstrate, for example, the mathematical modelling of cell adhesion, cohesion through adhesion, and cell–cell sorting. In Chapter 13, the authors attempt to quantitatively describe processes in embryo morphogenesis, using a multiscale approach. They believe that ‘Models must span from subcellular mechanisms of cellular mechanics, adhesion, and polarized cell behaviours to a mechanical understanding of substantial portions of the embryo and its environment’. This, again, shows that mathematical modelling is complex and requires knowledge and skills in mathematics, physics, computation, and biology, as well as many adjacent fields. The modelling steps from benign tumour to invasive cancer are laid out in Chapter 14. Here, the authors look at tumour growth and angiogenesis, as well as later stages of cancer, including invasion and intravasation.</p><p>The final chapter of Part IV (Chapter 15) is devoted to the description of the Delaunay Object Dynamics (DOD) method. The authors explain that the DOD platform allows investigators to separate the phenotype of a cell seen in experiments (e.g. surface markers, signalling cascades, etc.) from more universal biophysical phenotypes (e.g. surface area, cell division, etc.). This provides the opportunity to uncover subtle effects of the biophysical cell features on tissue organization. The DOD is essentially a tool which can be used to examine the behaviour of a large number of highly motile cells in certain tissues, each having its own phenotype dynamics and mechanical properties. The authors illustrate this method with a simulation of fast-migrating cells in secondary lymphoid tissue.</p><p>This book is comprehensive and discusses many issues important in biological dynamics in general and in cell mechanics in particular. The authors demonstrate the great value of mathematical tools for examining biological phenomena at different scales. What I like most about this book is the range of topics selected by the editors and discussed in a remarkable way by the groups of authors who contributed to the individual chapters. I also like that each part of the book contains a brief summary of the contents of all associated chapters. Based on my observations as a book reviewer, it is rare to find such an approach, but it is an excellent idea.</p><p>The reader will find the numerous illustrations useful as they sufficiently support the information provided in the text. I will describe two specific examples: The first example is Figure P1 (Preface, p. x) which illustrates the multiscale view of cell mechanics. This schematic describes the subcellular scale, cell scale, and macroscopic scale (biological tissue). I believe this schematic is well placed and useful as it allows the reader to envision the different levels of investigation and modelling. The second example is Figure 6.1 (p. 161). This figure shows three interacting components that contribute to cell motility: adhesion, cytoskeleton, and extracellular environment. Inserting this figure in the text helps the reader understand the complexity of cell motility and the various biological components that need to be (are) considered for biological investigation and for constructing a computational framework.</p><p>I noticed that few illustrations are presented in colour but are somewhat oddly placed without legends in the middle of Chapter 14 (between pp. 398 and 399). A specific example is the set of three images depicting the adhesion of T24 cells to substrates with various stiffnesses. This figure is shown in Chapter 9 in black-and-white (Figure 9.10, p. 259) and then again in colour in Chapter 14. I believe showing this figure once in colour with legend in Chapter 9 would have been a better choice. The reference sections are placed at the end of each chapter and contain between 23 and 81 entries, which should be sufficient for readers who want to access additional information.</p><p>In conclusion, I believe that Chauvière, Preziosi, and Verdier have edited a useful text for the instructor, student, and researcher interested in cell mechanics and mathematical modelling. The book provides considerable value with comprehensive reviews of biological and mathematical concepts, methodologies, and applications. I recommend <italic>Cell Mechanics: From Single Scale-Based Models to Multiscale Modelling</italic> to the reader who wishes to gain an insight into this fascinating field.</p>
Associations between <italic>Mycobacterium avium</italic> subsp. <italic>paratuberculosis</italic> antibodies in bulk tank milk, season of sampling and protocols for managing infected cows	<sec><title>Background</title><p>The objective of this study was to identify associations between the concentration of <italic>Mycobacterium avium</italic> subsp. <italic>paratuberculosis</italic> (MAP) antibodies in bulk milk and potential risk factors in herd management and herd characteristics, explaining high MAP antibody titers in milk. An extensive questionnaire was administered to 292 organic and conventional dairy farms from New York, Wisconsin and Oregon. Bulk milk samples were taken from each farm for MAP enzyme-linked immunosorbent assay (ELISA). A general linear model was constructed with MAP ELISA value as the outcome variable and the management factors and herd characteristics as independent variables, while at the same time controlling for the study design variables of state, herd size, and production system (organic or conventional). High bulk tank MAP ELISA value may be due to either a high prevalence of MAP in a herd with many cows contributing to the antibody titer or due to a few infected cows that produce large quantities of antibodies.</p></sec><sec><title>Results</title><p>Results of the regression models indicated that bulk milk ELISA value was associated with season of sampling and the presence or absence of protocols for managing MAP-positive cows. The concentration of MAP antibodies in bulk milk varied seasonally with a peak in the summer and low concentrations in the winter months. When compared to farms that had never observed clinical Johne’s disease, keeping MAP-positive cows or only culling them after a period of delay was associated with an increase in optical density.</p></sec><sec><title>Conclusions</title><p>The seasonal variation in MAP antibody titers, with a peak in the summer, may be due to a seasonal increase in MAP-bacterial load. Additionally, seasonal calving practices may contribute to seasonal fluctuations in MAP antibody titers in bulk tank milk. Keeping MAP-positive cows increases the antibody titer in bulk milk, likely due to direct antibody production in the infected cow and indirect triggering of antibody production in herdmates.</p></sec>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Cazer</surname><given-names>Casey L</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>clcazer@gmail.com</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Mitchell</surname><given-names>Rebecca M</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>rmm257@gmail.com</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Cicconi-Hogan</surname><given-names>Kellie M</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>kmc277@cornell.edu</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Gamroth</surname><given-names>Michael</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>mike.gamroth@oregonstate.edu</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Richert</surname><given-names>Roxann M</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>rxweix@gmail.com</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Ruegg</surname><given-names>Pamela L</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>plruegg@wisc.edu</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Schukken</surname><given-names>Ynte H</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>yschukken@cornell.edu</email></contrib>	BMC Veterinary Research	<sec><title>Background</title><p>Johne’s disease, a chronic disease caused by infection with <italic>Mycobacterium avium</italic> subsp. <italic>paratuberculosis</italic> (MAP), costs the US dairy industry $200 to $250 million annually due to increased cow replacement costs and reduction in milk production [<xref ref-type="bibr" rid="B1">1</xref>] and also decreased fertility in high-shedding animals [<xref ref-type="bibr" rid="B2">2</xref>]. The control of Johne’s disease requires good herd management practices, such as preventing fecal contamination of feed and water and testing replacement cattle for MAP. Good management procedures focus on reducing transmission and the introduction of MAP into the herd [<xref ref-type="bibr" rid="B3">3</xref>]. Because MAP infected cows may not show clinical signs during their productive lifetime [<xref ref-type="bibr" rid="B4">4</xref>], it is important to test many cows in a herd to properly assess MAP infection prevalence.</p><p>A simple, quick test that provides an estimate of herd-level MAP prevalence would allow herd managers to respond by changing their management strategies, thus improving their probability of eliminating and preventing MAP infection in the long term. MAP surveillance and monitoring has been proposed as an ideal testing strategy to ensure that infection pressures are low while keeping the cost of testing low [<xref ref-type="bibr" rid="B5">5</xref>]. Herd-level MAP-prevalence testing often involves pooled fecal samples used for culture or PCR. However pooled sample strategies are still time-consuming because individual cows or environmental areas must be sampled [<xref ref-type="bibr" rid="B6">6</xref>].</p><p>The magnitude of an ELISA test result for MAP antibodies in the milk of individual cows has been reported to be related to the likelihood of an animal testing positive on a fecal culture for MAP [<xref ref-type="bibr" rid="B7">7</xref>]. Collins et al. [<xref ref-type="bibr" rid="B7">7</xref>] also reported that the level of MAP shedding, considered a measure of the stage of infection, was directly related to the ability of an individual milk ELISA test to detect an infected animal. The Parachek commercial ELISA tests have been found to have a high specificity and a sensitivity ranging from 21 to 67% on individual milk samples [<xref ref-type="bibr" rid="B8">8</xref>]. When used on bulk tank milk, the Pourquier ELISA test had a sensitivity of 57% [<xref ref-type="bibr" rid="B9">9</xref>]. Additionally, van Weering et al. [<xref ref-type="bibr" rid="B10">10</xref>] demonstrated that certified MAP-negative herds had a low sample/positive (S/P) ratio on bulk milk Pourquier ELISA tests and they showed that the likelihood of a herd having a MAP-infected animal increased with increasing bulk milk ELISA S/P ratio. Finally, bulk milk ELISA tests have been shown to perform similarly to serum ELISA tests at the herd level, with a sensitivity of 56 to 83%, when fecal culture is used as a reference [<xref ref-type="bibr" rid="B11">11</xref>]. The sensitivity can be improved by using modified protocols [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B12">12</xref>]. Together, these published results on bulk milk and individual cow milk provide a logical validation for the use of bulk milk ELISA corrected optical density (OD) as a continuous outcome value to scale the risk of MAP infection in the lactating herd.</p><p>Determining associations between management factors and bulk milk ELISA values will provide a simple, efficient method for farms to respond to herd MAP infections by changing their management practices or understanding risks associated with certain management practices. However, few studies have used bulk milk ELISA to identify herd-level risk factors for increased MAP prevalence, while studies relating individual animal fecal or serum test results to management practices are more common (e.g. [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B14">14</xref>]).</p><p>The objective of this study, therefore, is to relate bulk tank milk ELISA optical density with potential risk factors for MAP prevalence in 233 farms. Our aim was to determine if particular management practices and herd characteristics are associated with increased bulk milk ELISA values.</p></sec><sec sec-type="results"><title>Results</title><p>The median herd size was 57 for organic farms and 70 for conventional farms included in the multivariate model. Herd size ranged from 20 to 723 on the organic farms and 26 to 535 on the conventional farms. Organic farms produced an average of 14,900 lbs of milk per cow per year whereas conventional farms in the model produced an average of 20,637 lbs per cow annually. All 170 of the organic farms grazed their cattle whereas 25 conventional herds grazed and 38 conventional farms did not graze. All of the non-grazing farms allowed heifers to graze pasture.</p><p>The corrected OD, for all herds, ranged from -0.098 to 0.37 with a mean of -0.023 and a standard deviation of 0.047. The range was -0.098 to 0.17 for herds included in the multivariate model. The mean of the included herds was -0.024 and the standard deviation was 0.037. There were two farms that had a corrected OD greater than 0.17 and these were excluded from the multivariate model due to missing data.</p><p>All of the variables initially included in the multivariate model are summarized in Table <xref ref-type="table" rid="T1">1</xref>. The stepwise order of removal from the final regression model was as follows: written plan for Johne’s disease (P = 0.95), parity (P = 0.89), calving area (P = 0.83), Jersey herd (P = 0.82), spreading manure (P = 0.75), source of drinking water (P = 0.72), open farm (P = 0.59), sine seasonality (P = 0.48), average yield (P = 0.41), heifer and cow contact on pasture (P = 0.39), Johne’s program (P = 0.43), and calf housing (P = 0.38). The variables remaining in the final multivariate model are summarized in Table <xref ref-type="table" rid="T2">2</xref>.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>List of variables included in GLMSELECT procedure to produce final multivariate model</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Variable</bold></th><th align="left"><bold>Type</bold></th><th align="left"><bold>Description</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Production<hr/></td><td align="left" valign="bottom">Binary<hr/></td><td align="left" valign="bottom">Organic or Conventional<hr/></td></tr><tr><td align="left" valign="bottom">State<hr/></td><td align="left" valign="bottom">Nominal<hr/></td><td align="left" valign="bottom">NY, OR, or WI<hr/></td></tr><tr><td align="left" valign="bottom">Herd size<hr/></td><td align="left" valign="bottom">Ordinal<hr/></td><td align="left" valign="bottom">< 100, 100 to 200, or > 200<hr/></td></tr><tr><td align="left" valign="bottom">Parity<hr/></td><td align="left" valign="bottom">Continuous<hr/></td><td align="left" valign="bottom">The average lactation of all lactating cows at the time of the survey<hr/></td></tr><tr><td align="left" valign="bottom">Average yield<hr/></td><td align="left" valign="bottom">Continuous<hr/></td><td align="left" valign="bottom">Average annual milk production per cow<hr/></td></tr><tr><td align="left" valign="bottom">Calf housing<hr/></td><td align="left" valign="bottom">Nominal<hr/></td><td align="left" valign="bottom">Always housed in individual pens or areas; sometimes housed in groups or with contact to cows; always housed in groups or with contact to cows<hr/></td></tr><tr><td align="left" valign="bottom">Description of calving area<hr/></td><td align="left" valign="bottom">Nominal<hr/></td><td align="left" valign="bottom">Dedicated calving area; area shared with lactating cows; or area sometimes shared with sick cows<hr/></td></tr><tr><td align="left" valign="bottom">Heifer and cow contact on pasture<hr/></td><td align="left" valign="bottom">Nominal<hr/></td><td align="left" valign="bottom">Heifers grazing a pasture with or after the cows; before the cows; or heifers and cows do not graze the same pasture<hr/></td></tr><tr><td align="left" valign="bottom">Jersey<hr/></td><td align="left" valign="bottom">Binary<hr/></td><td align="left" valign="bottom">Mostly Jersey or not mostly Jersey herd<hr/></td></tr><tr><td align="left" valign="bottom">Johne’s procedures<hr/></td><td align="left" valign="bottom">Nominal<hr/></td><td align="left" valign="bottom">Procedures for MAP-positive cows: cull after calving or dry off; cull immediately; keep; never observed clinical Johne’s disease<hr/></td></tr><tr><td align="left" valign="bottom">Johne’s program<hr/></td><td align="left" valign="bottom">Binary<hr/></td><td align="left" valign="bottom">Participation in Johne’s program<hr/></td></tr><tr><td align="left" valign="bottom">Open farm<hr/></td><td align="left" valign="bottom">Binary<hr/></td><td align="left" valign="bottom">Some or no entering animals<hr/></td></tr><tr><td align="left" valign="bottom">Seasonality sine<hr/></td><td align="left" valign="bottom">Continuous<hr/></td><td align="left" valign="bottom"><disp-formula><mml:math id="M1" name="1746-6148-9-234-i1" overflow="scroll"><mml:mrow><mml:mi mathvariant="normal">Sin</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>2</mml:mn><mml:mi mathvariant="normal">π</mml:mi><mml:mfrac><mml:mi mathvariant="normal">Day</mml:mi><mml:mn>365</mml:mn></mml:mfrac></mml:mrow></mml:mfenced></mml:mrow></mml:math></disp-formula>, where day is a continuous variable from 1 to 365<hr/></td></tr><tr><td align="left" valign="bottom">Seasonality cosine<hr/></td><td align="left" valign="bottom">Continuous<hr/></td><td align="left" valign="bottom"><disp-formula><mml:math id="M2" name="1746-6148-9-234-i2" overflow="scroll"><mml:mrow><mml:mi mathvariant="normal">Cos</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>2</mml:mn><mml:mi mathvariant="normal">π</mml:mi><mml:mfrac><mml:mi mathvariant="normal">Day</mml:mi><mml:mn>365</mml:mn></mml:mfrac></mml:mrow></mml:mfenced></mml:mrow></mml:math></disp-formula>, where day is a continuous variable from 1 to 365<hr/></td></tr><tr><td align="left" valign="bottom">Source of drinking water<hr/></td><td align="left" valign="bottom">Nominal<hr/></td><td align="left" valign="bottom">Primary source of drinking water for 60 days prior to survey: well; municipal water; or surface water<hr/></td></tr><tr><td align="left" valign="bottom">Spreading manure<hr/></td><td align="left" valign="bottom">Binary<hr/></td><td align="left" valign="bottom">Manure spreading on pasture or fields that will be consumed by animals<hr/></td></tr><tr><td align="left">Written plan for Johne’s disease</td><td align="left">Binary</td><td align="left">Written herd plan for managing Johne’s disease</td></tr></tbody></table></table-wrap><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Summary of linear regression model</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="left" valign="bottom"> <hr/></th><th align="left" valign="bottom"> <hr/></th><th colspan="3" align="center" valign="bottom"><bold>Multivariate model</bold><hr/></th></tr><tr><th align="left"><bold>Variable</bold></th><th align="left"><bold>Description</bold></th><th align="center"><bold>β Coefficient</bold></th><th align="center"><bold>Standard error</bold></th><th align="center"><bold>P-value</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Intercept<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="center" valign="bottom">-0.045<hr/></td><td align="center" valign="bottom">0.008<hr/></td><td align="center" valign="bottom">< .001<hr/></td></tr><tr><td align="left" valign="bottom">Production<hr/></td><td align="left" valign="bottom">Conventional<hr/></td><td align="center" valign="bottom">-0.003<hr/></td><td align="center" valign="bottom">0.004<hr/></td><td align="center" valign="bottom">0.400<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">Organic<hr/></td><td align="center" valign="bottom">0<hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom">-<hr/></td></tr><tr><td align="left" valign="bottom">Herd size<hr/></td><td align="left" valign="bottom">< 100<hr/></td><td align="center" valign="bottom">0.010<hr/></td><td align="center" valign="bottom">0.007<hr/></td><td align="center" valign="bottom">0.175<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">100 to 200<hr/></td><td align="center" valign="bottom">0.003<hr/></td><td align="center" valign="bottom">0.008<hr/></td><td align="center" valign="bottom">0.734<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">> 200<hr/></td><td align="center" valign="bottom">0<hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom">-<hr/></td></tr><tr><td align="left" valign="bottom">State<hr/></td><td align="left" valign="bottom">NY<hr/></td><td align="center" valign="bottom">0.003<hr/></td><td align="center" valign="bottom">0.004<hr/></td><td align="center" valign="bottom">0.463<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">OR<hr/></td><td align="center" valign="bottom">0.013<hr/></td><td align="center" valign="bottom">0.007<hr/></td><td align="center" valign="bottom">0.069<hr/></td></tr><tr><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">WI<hr/></td><td align="center" valign="bottom">0<hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom">-<hr/></td></tr><tr><td rowspan="4" align="left" valign="bottom">Johne’s procedures<hr/></td><td align="left" valign="bottom">Cull after calving or dry off<hr/></td><td align="center" valign="bottom">0.020<hr/></td><td align="center" valign="bottom">0.009<hr/></td><td align="center" valign="bottom">0.025<hr/></td></tr><tr><td align="left" valign="bottom">Cull immediately<hr/></td><td align="center" valign="bottom">0.010<hr/></td><td align="center" valign="bottom">0.005<hr/></td><td align="center" valign="bottom">0.035<hr/></td></tr><tr><td align="left" valign="bottom">Keep<hr/></td><td align="center" valign="bottom">0.018<hr/></td><td align="center" valign="bottom">0.007<hr/></td><td align="center" valign="bottom">0.019<hr/></td></tr><tr><td align="left" valign="bottom">Never observed clinical Johne’s disease<hr/></td><td align="center" valign="bottom">0<hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom">-<hr/></td></tr><tr><td align="left">Seasonality</td><td align="left"><disp-formula><mml:math id="M3" name="1746-6148-9-234-i3" overflow="scroll"><mml:mrow><mml:mi mathvariant="italic">Cosine</mml:mi><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>2</mml:mn><mml:mi>π</mml:mi><mml:mfrac><mml:mi mathvariant="italic">Day</mml:mi><mml:mn>365</mml:mn></mml:mfrac></mml:mrow></mml:mfenced></mml:mrow></mml:math></disp-formula></td><td align="center">-0.030</td><td align="center">0.003</td><td align="center">< .001</td></tr></tbody></table><table-wrap-foot><p>Linear regression model on <italic>Mycobacterium avium</italic> subsp. <italic>paratuberculosis</italic> ELISA optical density from bulk tank milk samples of 233 farms. Multivariate model was built by backwards stepwise elimination and started with all variables listed in Table <xref ref-type="table" rid="T1">1</xref>.</p></table-wrap-foot></table-wrap><p>The final multivariate model residuals were approximately normal with a slight skew to the right. However, it may be expected that the PROC GLM procedure still produces valid results when the assumption of error normality is mildly violated [<xref ref-type="bibr" rid="B15">15</xref>].</p><p>None of the included study design variables that were forced into the models were significant in the final multivariate model (Table <xref ref-type="table" rid="T2">2</xref>). The cosine seasonality curve was very significant (P < 0.0001). It demonstrated a peak in OD during the summer months and a trough during the winter months. The regression coefficient of -0.03 indicates the maximum drop in winter of the corrected OD and the maximum increase of +0.03 in summer. The seasonality is most evident in the farms sampled in late 2010 through 2011 (Figure <xref ref-type="fig" rid="F1">1</xref>). Seasonal calving farms and non-grazing farms were well distributed across the sampling time frame (Figure <xref ref-type="fig" rid="F1">1</xref>). There were 46 farms in the final multivariate model that indicated they used seasonal calving practices. Of these 46 farms, 42 responded that they tried to calve in the spring, 15 tried to calve in the summer, 21 in the fall and 5 in the winter. Seasonal farms did not necessarily calve over just one season, and all farms with fall or winter calving also had at least one additional calving season. The bulk milk ELISA value of seasonal calving farms differed with season of sampling, with a maximum value in the spring and continually decreasing to a minimum in the winter (Figure <xref ref-type="fig" rid="F1">1</xref>).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Seasonality in ELISA optical density.</bold> Bulk tank milk optical density of <italic>Mycobacterium avium</italic> subsp. <italic>paratuberculosis</italic> ELISA by date of sampling. Solid blue line fitted from cosine seasonality variable with all other variables set to baseline. Herds with seasonal calving practices are red dots. Open circles represent farms that do not graze their cattle. Red open circles are seasonal calving farms that do not graze. Solid dots represent all other farms included in final multivariate model.</p></caption><graphic xlink:href="1746-6148-9-234-1"/></fig><p>The procedure for managing MAP-positive cows was the only other significant variable in the multivariate model. Compared to farms that had never observed clinical Johne’s disease, farms that kept MAP-positive cows had a significant increase of 0.018 in their bulk milk OD (P = 0.019). Farms that culled MAP-positive cows also had significant increases in bulk milk ELISA value when compared to farms without a history of Johne’s disease but those that culled the cow immediately had a much smaller increase (β = 0.010, P = 0.035) than those that culled the cow after a period of delay, such as after her next calving (β = 0.020, P = 0.025). The final multivariate model explained 40% of the variation in bulk tank ELISA OD.</p></sec><sec sec-type="discussion"><title>Discussion</title><p>Bulk milk could have a high concentration of MAP antibodies because a dairy herd has a high prevalence of MAP infected cows or because a few infected cows produce large quantities of MAP antibodies. Variation in milk MAP antibody concentration within groups of fecal-positive and groups of fecal-negative cows has been observed due to a small number of cows producing a high level of antibodies and changes in antibody production within individual cows over time [<xref ref-type="bibr" rid="B16">16</xref>]. A study by van Weering et al. [<xref ref-type="bibr" rid="B10">10</xref>] showed that a 100-fold dilution of a positive individual milk ELISA sample may still result in a positive ELISA result, indicating that a few positive cows may impact the bulk milk ELISA result. Given the relatively small herd sizes in our study, it is clear that a few cows producing high concentrations of antibodies may influence bulk milk ELISA titers. The precise impact of a single ELISA positive cow would depend on the herd size, milk production, and the difference between the individual milk antibody concentration of a particular cow and the average antibody concentration of the rest of the herd. Therefore, some farms may have higher ELISA values than average because they have higher than average MAP infection rates or because they have some cows producing high concentrations of MAP antibodies. Careful comparison of individual cow milk ELISA tests and cow milk production level to a bulk tank test would be required to differentiate between farms that have several infected, average antibody-producing cows, and those that have a limited number of animals with high antibody titers.</p><p>In our model, the location of a farm (NY, OR, or WI) was not significantly associated with corrected OD. However, other studies have found that herds in the Midwest are more likely to be positive for Johne’s disease [<xref ref-type="bibr" rid="B14">14</xref>] and observe clinical signs of Johne’s disease in their herd [<xref ref-type="bibr" rid="B17">17</xref>]. Herd size, another study design variable, despite not being significantly associated with ELISA results in this study has previously been found to be positively associated with herd infection status, with larger herds having a greater risk [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B18">18</xref>]. Additionally, Wells and Wagner [<xref ref-type="bibr" rid="B14">14</xref>] observed a positive association between group housing for calves and the herd infection status whereas our model did not find a significant association between types of calf housing and ELISA result. Spreading manure on forage fields [<xref ref-type="bibr" rid="B13">13</xref>] and open water sources [<xref ref-type="bibr" rid="B19">19</xref>] have also been shown to be associated with a higher risk of MAP infection, but were not significantly associated with corrected OD in our model. Finally, other studies have demonstrated an increased risk of Johne’s disease with high parity and Jersey breed cows [<xref ref-type="bibr" rid="B18">18</xref>], although those variables did not remain in our multivariate model.</p><p>The significant seasonal effect, with MAP antibodies highest in the summer and lowest in the winter (Figure <xref ref-type="fig" rid="F1">1</xref>), represents a change in antibody secretion into bulk milk across the seasons. Seasonal calving could account for this periodic change in antibodies because milk antibodies are, on average, greatest at the beginning and end of lactation [<xref ref-type="bibr" rid="B16">16</xref>]. A farm that uses seasonal calving would see a herd-level increase in bulk milk antibodies, including antibodies against MAP, during or shortly after the calving season when most of their cows are just starting to lactate. This could result in high optical densities from the detection of MAP-specific and non-specific antibodies. The bulk milk ELISA value of seasonal calving farms, the majority of which indicated that they attempt to calve their cows in the spring, was greatest during the spring. This supports a days-in-milk dependent change in antibody secretion as a potential explanation for the seasonal trend in MAP antibodies in milk (Figure <xref ref-type="fig" rid="F1">1</xref>). Thus, seasonal calving could explain some of the seasonal variation but not all since only 46 of the 233 farms in the multivariate model used seasonal calving and non-seasonal calving farms still show a seasonal trend in OD (Figure <xref ref-type="fig" rid="F1">1</xref>). The seasonal variation also exists in non-grazing conventional farms (Figure <xref ref-type="fig" rid="F1">1</xref>), suggesting that the seasonal variation in MAP antibodies is not limited to grazing farms, which may have inherent seasonality in their calvings due to seasonal nutritional differences regardless of planned seasonal calving practices. Additionally, the seasonality variable remains significant (P < 0.0001) after removing all grazing farms from the final multivariate model.</p><p>Another possible explanation for the seasonal change in antibodies is a seasonal fluctuation in MAP load. It is unlikely that MAP prevalence changes seasonally on any given farm but the MAP load in the environment or in the cows could potentially change with the seasons. A previous study found a higher prevalence of MAP-positive carcasses, as determined by ileum and lymph node cultures and PCR, in the spring than in other seasons [<xref ref-type="bibr" rid="B20">20</xref>]. Additionally, an increase in viable MAP isolated from retail milk in the summer has been shown [<xref ref-type="bibr" rid="B21">21</xref>]. Humoral responses are known to occur in subclinical MAP infections, which results in activated B cells producing antibodies [<xref ref-type="bibr" rid="B22">22</xref>]. It has previously been suggested that MAP exposure can trigger antibody production in infection-resistant adult cattle, possibly resulting in an increased MAP ELISA titer [<xref ref-type="bibr" rid="B7">7</xref>]. Thus it is possible that the increase in MAP load of individual animals during the spring and summer could result in an increased humoral immune response in herdmates and therefore increase antibody levels in milk.</p><p>If bulk milk MAP ELISA is used as an indicator of MAP infection, it needs to be corrected for the seasonal changes in MAP milk antibodies. Consistently sampling during only one season may need to be recommended in order to compare ELISA results across time at one farm or among farms. Our results would need to be confirmed by similar studies to make such recommendations with more confidence.</p><p>Both culling and keeping cows known to be MAP-infected are associated with an increase in bulk milk ELISA value compared to herds without a history of clinical Johne’s disease, which suggests that farms with protocols for MAP-positive cows in place are more likely infected with MAP and possibly have a higher MAP prevalence. The larger increase in bulk milk OD associated with keeping MAP-positive cows or culling them after calving compared to the small increase in OD associated with culling MAP-positive cows immediately was not statistically significant (P = 0.25). However, it has previously been suggested that culling cows immediately is the best method for controlling MAP antibodies in bulk milk. Lu et al. [<xref ref-type="bibr" rid="B23">23</xref>] demonstrated the importance of culling positive animals immediately after detection in order to control MAP transmission. MAP-positive cows that are kept in the herd for any period of time could increase the bulk milk MAP antibody titer by producing MAP antibodies in their milk and by re-exposing other cows, which may then begin secreting MAP antibodies into milk. This would be particularly true for cows with a progressive course of disease, in which antibodies rise rapidly as the cow progresses from moderate to heavy fecal shedding [<xref ref-type="bibr" rid="B24">24</xref>].</p></sec><sec sec-type="conclusions"><title>Conclusion</title><p>In summary, increased MAP antibody concentrations in bulk milk were associated with season of sampling and the protocols that were used on the farm for managing MAP-positive cows. The association between the season of sampling and MAP antibody concentration could be the result of seasonal variations in MAP in cows and in the environment, and could be related to seasonal calving practices. This seasonal change in MAP antibodies in bulk milk will need to be considered when using bulk milk ELISA as a MAP surveillance tool. The significance of protocols to manage MAP-positive cows pointed towards the importance of culling known MAP-positive cows immediately.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Study and survey</title><p>Data came from a large cross-sectional study of 292 farms conducted between March 2009 and May 2011 that focused on comparing organic and conventional dairy farms. The study has previously been described in more detail [<xref ref-type="bibr" rid="B25">25</xref>]. Briefly, to be included in the study, farms had at least 20 lactating cows and had been shipping milk for 2 full years prior to the study; organic farms must have been shipping certified organic milk for these 2 years. Organic herds in New York, Oregon, and Wisconsin were identified through county extension agents, personal contacts and organic certifying organizations. Conventional herds were identified from lists of licensed dairy farmers, which were obtained from the departments of agriculture in New York, Oregon and Wisconsin, and were located within 50 miles of the identified organic farms. Conventional farms were size-category matched to organic farms based on three herd-size groups: 20 to 99 cows, 100 to 199 cows, and more than 200 cows. Conventional to organic herd ratios were found to be different in each state so farms were matched accordingly: 3 organic to 1 conventional in NY, 1 organic to 1 conventional in OR, and 2 organic to 1 conventional in WI. Conventional herds were designated as non-grazing if less than 30% of the lactating cows’ dry matter intake came from pasture during the grazing season. Non-grazing farms could allow heifers to graze pasture.</p><p>A questionnaire was used on all farms to identify farm management practice and herd performance. The questionnaire has been described in detail [<xref ref-type="bibr" rid="B25">25</xref>] and is available online [<xref ref-type="bibr" rid="B26">26</xref>]. All the interviews were conducted by a single person within each state. The interviewers had been trained in administering and scoring the questionnaire in a consistent manner across states. The Institutional Review Board at Oregon State University approved the use of human subjects for the questionnaire, reference number 3995. Herds that indicated that they were seasonal calving herds were further classified by the season in which they tried to calve their cows. Seasonal calving farms were able to designate one or more seasons as a calving season on the questionnaire but the questionnaire did not require specifying a main or primary calving season. Study personnel collected bulk tank samples on the same visit as questionnaire administration. Bulk milk samples were sent to Quality Milk Production Services at Cornell University (Ithaca, NY). Samples were tested for <italic>Salmonella spp</italic>., <italic>Listeria monocytogenes</italic>, Shiga toxin producing <italic>E. coli</italic>, <italic>Mycoplasma bovis</italic>, Bovine Virus Diarrhea virus, antibodies against <italic>Mycobacterium avium</italic> subsp. <italic>paratuberculosis</italic> (MAP) and mastitis-causing bacteria. Samples were then sent to Dairy One Cooperative in Ithaca, NY for standard milk quality assays.</p></sec><sec><title>Enzyme linked immunosorbent assay</title><p>Bulk tank samples were analyzed using the commercially available Parachek ELISA (product number 63308) according to the directions provided by the manufacturer (Prionics, Zurich, Switzerland). Briefly, 100 μL of each bulk milk sample was diluted with 100 μL of Green Diluent, containing <italic>Mycobacterium phlei</italic>, and incubated at room temperature for 30-60 minutes. Then 100 μL of each sample and 100 μL of the manufacturer-provided positive and negative controls were added to microtitre plates coated with <italic>M. paratuberculosis</italic>. The plates were shaken and incubated at room temperature for 30 minutes. Plates were washed 6 times with wash buffer at room temperature. Then 100 μL of conjugate reagent (Horseradish peroxidase labeled anti-bovine Ig) was added to each well. The plates were again shaken, incubated at room temperature for 30 minutes, and washed 6 times with wash buffer. Next 100 μL of enzyme substrate solution (DMSO) was added to each well and the plates were incubated and shaken at room temperature until the positive controls reached an optical density of .35 to .40 with a 620-650 nm filter. Finally, 50 μL of enzyme stopping solution (0.5 M H<sub>2</sub>SO<sub>4</sub>) was mixed into each well and the absorbance of each well was read with a 450 nm filter. The Parachek ELISA optical density result is reported as a numerical value, which is classified as positive or negative in relation to a cut-off value equal to the average negative control plus 0.10. The average optical density of the two negative controls on each plate was subtracted from the sample optical densities of the same plate [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B27">27</xref>] to account for inter-plate variation. This can result in the corrected optical density being less than 0 [<xref ref-type="bibr" rid="B27">27</xref>]. This corrected optical density value (optical density minus average negative control) was used as the continuous outcome variable for statistical analysis.</p></sec><sec><title>Statistical analysis</title><p>Descriptive analyses were performed on independent and dependent variables included in the multivariate model. The key outcome variable was the corrected bulk milk ELISA optical density. Management factors and other variables of interest were identified in the dataset based on an <italic>a-priori</italic> rationale that they were associated with MAP antibodies in milk (Table <xref ref-type="table" rid="T1">1</xref>). These variables of interest included aspects of herd management (production system, written plan for Johne’s disease, procedures for MAP-positive cows, participation in a Johne’s program), herd descriptors (state, herd size, average parity, average yield, Jerseys), and Johne’s disease specific risk factors (spreading manure, contact between heifers and cows on pasture, calf housing, calving area, open farm, source of drinking water). Variables to describe seasonality were developed using sine and cosine functions as previously described: <inline-formula><mml:math id="M4" name="1746-6148-9-234-i4" overflow="scroll"><mml:mrow><mml:mo>sine</mml:mo><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>2</mml:mn><mml:mi>π</mml:mi><mml:mfrac><mml:mi mathvariant="italic">Day</mml:mi><mml:mn>365</mml:mn></mml:mfrac></mml:mrow></mml:mfenced></mml:mrow></mml:math></inline-formula> and <inline-formula><mml:math id="M5" name="1746-6148-9-234-i5" overflow="scroll"><mml:mrow><mml:mo>cosine</mml:mo><mml:mfenced open="(" close=")"><mml:mrow><mml:mn>2</mml:mn><mml:mi>π</mml:mi><mml:mfrac><mml:mi mathvariant="italic">Day</mml:mi><mml:mn>365</mml:mn></mml:mfrac></mml:mrow></mml:mfenced></mml:mrow></mml:math></inline-formula>[<xref ref-type="bibr" rid="B28">28</xref>]. Stepwise backwards least-squares linear regression (PROC GLMSELECT, SAS 9.3, Inst. Inc., 2011) was used to build a multivariate model with a constant sample size of 233 farms. Fifty-nine farms were excluded from this analysis due to missing data from one or more variables evaluated for inclusion in the multivariate model. The study design variables (state, herdsize size category, production system) were forced into the model (INCLUDE option) because they determined the inclusion of herds in the study. All other variables were selected based on statistical significance at the 0.05 level. The final multivariate model was in the form:</p><p><disp-formula><mml:math id="M6" name="1746-6148-9-234-i6" overflow="scroll"><mml:mtable columnalign="left"><mml:mtr><mml:mtd><mml:mi>CORRECTED</mml:mi><mml:mo>_</mml:mo><mml:mi>OD</mml:mi><mml:mo>=</mml:mo><mml:mi mathvariant="normal">β</mml:mi><mml:mn>0</mml:mn><mml:mo>+</mml:mo><mml:mi mathvariant="normal">β</mml:mi><mml:mn>1</mml:mn><mml:mfenced open="(" close=")"><mml:mrow><mml:mi mathvariant="italic">HERDSIZE</mml:mi><mml:mo>_</mml:mo><mml:mo><</mml:mo><mml:mn>100</mml:mn></mml:mrow></mml:mfenced></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mspace width="11em"/><mml:mo>+</mml:mo><mml:mi mathvariant="normal">β</mml:mi><mml:mn>2</mml:mn><mml:mo stretchy="true">(</mml:mo><mml:mi mathvariant="italic">HERDSIZE</mml:mi><mml:mo>_</mml:mo><mml:mn>100</mml:mn><mml:mo>-</mml:mo><mml:mn>200</mml:mn><mml:mo stretchy="true">)</mml:mo></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mspace width="11em"/><mml:mo>+</mml:mo><mml:mi mathvariant="normal">β</mml:mi><mml:mn>3</mml:mn><mml:mo stretchy="true">(</mml:mo><mml:mi mathvariant="italic">STATE</mml:mi><mml:mo>_</mml:mo><mml:mi mathvariant="italic">NY</mml:mi><mml:mo stretchy="true">)</mml:mo><mml:mo>+</mml:mo><mml:mi mathvariant="normal">β</mml:mi><mml:mn>4</mml:mn><mml:mo stretchy="true">(</mml:mo><mml:mi mathvariant="italic">STATE</mml:mi><mml:mo>_</mml:mo><mml:mi mathvariant="italic">OR</mml:mi><mml:mo stretchy="true">)</mml:mo></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mspace width="11em"/><mml:mo>+</mml:mo><mml:mi mathvariant="normal">β</mml:mi><mml:mn>5</mml:mn><mml:mfenced open="(" close=")"><mml:mrow><mml:mi mathvariant="italic">PRODUCTION</mml:mi><mml:mo>_</mml:mo><mml:mi mathvariant="italic">SYSTEM</mml:mi></mml:mrow></mml:mfenced></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mspace width="11em"/><mml:mo>+</mml:mo><mml:mi mathvariant="normal">β</mml:mi><mml:mn>6</mml:mn><mml:mo stretchy="true">(</mml:mo><mml:mi mathvariant="italic">COSINE</mml:mi><mml:mo>_</mml:mo><mml:mi mathvariant="italic">SEASON</mml:mi><mml:mo stretchy="true">)</mml:mo></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mspace width="11em"/><mml:mo>+</mml:mo><mml:mi mathvariant="normal">β</mml:mi><mml:mn>7</mml:mn><mml:mo stretchy="true">(</mml:mo><mml:mi mathvariant="italic">SINE</mml:mi><mml:mo>_</mml:mo><mml:mi mathvariant="italic">SEASON</mml:mi><mml:mo stretchy="true">)</mml:mo><mml:mo>+</mml:mo><mml:mstyle displaystyle="true"><mml:munderover><mml:mi>∑</mml:mi><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>n</mml:mi></mml:munderover><mml:mrow><mml:msub><mml:mi mathvariant="normal">β</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:msub><mml:mi>X</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow></mml:mstyle><mml:mo>+</mml:mo><mml:mi mathvariant="normal">ϵ</mml:mi></mml:mtd></mml:mtr></mml:mtable></mml:math></disp-formula></p><p>where OD is optical density, β<sub>0</sub> is the intercept, β<sub>1</sub> and β<sub>2</sub> are the coefficients of the herd size variable, β<sub>3</sub> and β<sub>4</sub> are the coefficients of the state variable, β<sub>5</sub> is the coefficient of the production system variable, β<sub>6</sub> and β<sub>7</sub> are the coefficients of the cosine and sine season variables respectively, β<sub>i</sub> represents the coefficients of other risk factor variables X<sub>i</sub> (i = 1 to n), and ϵ is a random error term assumed to be normally distributed with mean 0. This assumption was validated by examining the distribution of residuals.</p></sec></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>CLC: performed the statistical analysis and drafted the manuscript. RMM: performed the statistical analysis and drafted the manuscript. KMCH: designed and carried out the survey. MG: designed the survey. RMR: designed and carried out the survey. PLR: designed the survey. YHS: assisted with statistical analysis and survey design, and helped draft the manuscript. All authors read and approved the final manuscript.</p></sec>
<italic>Anaplasma phagocytophilum</italic> strains from voles and shrews exhibit specific <italic>ankA</italic> gene sequences	<sec><title>Background</title><p><italic>Anaplasma phagocytophilum</italic> is a Gram-negative bacterium that replicates obligate intracellularly in neutrophils. It is transmitted by <italic>Ixodes</italic> spp. ticks and causes acute febrile disease in humans, dogs, horses, cats, and livestock. Because <italic>A. phagocytophilum</italic> is not transmitted transovarially in <italic>Ixodes</italic> spp., it is thought to depend on reservoir hosts to complete its life cycle. In Europe, <italic>A. phagocytophilum</italic> was detected in roe deer, red deer, wild boars, and small mammals. In contrast to roe deer, red deer and wild boars have been considered as reservoir hosts for granulocytic anaplasmosis in humans, dogs, and horses according to <italic>groESL</italic>- and <italic>ankA</italic>-based genotyping. <italic>A. phagocytophilum</italic> variants infecting small mammals in Europe have not been characterized extensively to date.</p></sec><sec><title>Results</title><p>We amplified the total <italic>ankA</italic> open reading frames of 27 strains from voles and shrews. The analysis revealed that they harboured <italic>A. phagocytophilum</italic> strains that belonged to a distinct newly described <italic>ankA</italic> gene cluster. Further, we provide evidence that the heterogeneity of <italic>ankA</italic> gene sequences might have arisen via recombination.</p></sec><sec><title>Conclusions</title><p>Based on <italic>ankA</italic>-based genotyping voles and shrews are unlikely reservoir hosts for granulocytic anaplasmosis in humans, dogs, horses, and livestock in Europe.</p></sec>	<contrib contrib-type="author" id="A1"><name><surname>Majazki</surname><given-names>Juliana</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>ylona@web.de</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Wüppenhorst</surname><given-names>Nicole</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>nicole.wueppenhorst@hu.hamburg.de</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Hartelt</surname><given-names>Kathrin</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>kathrin.hartelt@rps.bwl.de</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Birtles</surname><given-names>Richard</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>R.J.Birtles@salford.ac.uk</email></contrib><contrib contrib-type="author" corresp="yes" id="A5"><name><surname>von Loewenich</surname><given-names>Friederike D</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>friederike.loewenich@uniklinik-freiburg.de</email></contrib>	BMC Veterinary Research	<sec><title>Background</title><p><italic>Anaplasma phagocytophilum</italic> is a Gram-negative bacterium that replicates obligate intracellularly in neutrophils [<xref ref-type="bibr" rid="B1">1</xref>]. It is tick-transmitted and causes acute febrile disease in humans [<xref ref-type="bibr" rid="B2">2</xref>], in companion animals such as dogs [<xref ref-type="bibr" rid="B3">3</xref>], horses [<xref ref-type="bibr" rid="B4">4</xref>], and cats [<xref ref-type="bibr" rid="B5">5</xref>] as well as in livestock such as sheep and cattle [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref>]. The main vector of <italic>A. phagocytophilum</italic> in Europe is <italic>Ixodes ricinus</italic>, whereas it is primarily transmitted by <italic>I. scapularis</italic> and <italic>I. pacificus</italic> in North America and by <italic>I. persulcatus</italic> in Asia [<xref ref-type="bibr" rid="B2">2</xref>].</p><p>Evidence exists that the naturally circulating <italic>A. phagocytophilum</italic> strains show a considerable degree of host adaptation, because they are not equally infectious for different animal species [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B7">7</xref>-<xref ref-type="bibr" rid="B9">9</xref>]. The molecular characterization using major surface protein 2 (<italic>msp2</italic>) pseudogene sequences [<xref ref-type="bibr" rid="B10">10</xref>] as well as the <italic>ankA</italic> gene [<xref ref-type="bibr" rid="B11">11</xref>] has shown that strains originating from humans, dogs, and horses are homologous. Furthermore, horses and dogs are susceptible to infection with human <italic>A. phagocytophilum</italic> isolates [<xref ref-type="bibr" rid="B12">12</xref>-<xref ref-type="bibr" rid="B14">14</xref>].</p><p>At least in <italic>Ixodes</italic> spp. ticks <italic>A. phagocytophilum</italic> is not transmitted transovarially [<xref ref-type="bibr" rid="B15">15</xref>]. Therefore, it is thought to depend on reservoir hosts to complete its life cycle. In North America, based on molecular characterization and experimental infections small mammals such as white-footed mice [<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B17">17</xref>], chipmunks [<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B19">19</xref>], and squirrels [<xref ref-type="bibr" rid="B19">19</xref>] were reported as probable reservoirs for granulocytic anaplasmosis in humans, horses, and dogs. In contrast, the impact of white-tailed deer and woodrats was questioned [<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B20">20</xref>,<xref ref-type="bibr" rid="B21">21</xref>]. In Europe, <italic>A. phagocytophilum</italic> was detected amongst others in roe deer [<xref ref-type="bibr" rid="B22">22</xref>,<xref ref-type="bibr" rid="B23">23</xref>], red deer [<xref ref-type="bibr" rid="B23">23</xref>], wild boars [<xref ref-type="bibr" rid="B24">24</xref>], hedgehogs [<xref ref-type="bibr" rid="B25">25</xref>], and other small mammals [<xref ref-type="bibr" rid="B26">26</xref>].</p><p>The 16S rRNA gene has been used most often for strain characterization. However, it was shown that it is not informative enough to delineate distinct <italic>A. phagocytophilum</italic> genotypes [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B27">27</xref>-<xref ref-type="bibr" rid="B29">29</xref>]. Based on <italic>groESL</italic> and <italic>ankA</italic> gene sequences red deer [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B30">30</xref>] and wild boar [<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B32">32</xref>] were considered as reservoir hosts for granulocytic anaplasmosis in humans, dogs, and horses. In contrast, roe deer harboured <italic>A. phagocytophilum</italic> strains which mostly belonged to clearly separated <italic>groESL</italic>[<xref ref-type="bibr" rid="B30">30</xref>] and <italic>ankA</italic>[<xref ref-type="bibr" rid="B11">11</xref>] gene clusters.</p><p>Apart from using the 16S rRNA gene the <italic>A. phagocytophilum</italic> variants infecting small mammals in Europe have not been typed extensively to date. We therefore amplified the total <italic>ankA</italic> open reading frame (ORF) of 27 strains from voles and shrews captured in Germany as well as the UK and compared them to 221 <italic>ankA</italic> sequences determined earlier [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B27">27</xref>]. We here show that they harboured <italic>A. phagocytophilum</italic> strains that belonged to a distinct newly described <italic>ankA</italic> gene cluster. Therefore, voles and shrews are unlikely reservoir hosts for granulocytic anaplasmosis in humans, dogs, horses, and livestock in Europe.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Samples</title><p>27 <italic>A. phagocytophilum</italic> positive DNA samples from voles and shrews were investigated. 22 had been prepared earlier from the lung of voles captured in Germany [<xref ref-type="bibr" rid="B33">33</xref>]. Five had been purified from the blood of two voles [<xref ref-type="bibr" rid="B34">34</xref>] and three shrews [<xref ref-type="bibr" rid="B35">35</xref>] from the United Kingdom. The 16S rRNA and <italic>ankA</italic> gene sequences obtained here were compared to 221 sequences from humans, a great variety of animals, and <italic>I. ricinus</italic> ticks from previous studies [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B27">27</xref>]. Furthermore, seven additional samples from three humans, one dog, one horse, one cow, and one sheep were included. Table <xref ref-type="table" rid="T1">1</xref> shows host species and geographic origin of the samples.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p><bold>Host species and geographic origin of </bold><bold> <italic>A. phagocytophilum </italic> </bold><bold>positive samples (n =34)</bold></p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Sample</bold></th><th align="left"><bold>Origin</bold></th><th align="left"><bold>Sample</bold></th><th align="left"><bold>Origin</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom"><italic>Myodes glareolus</italic><hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"><italic>Microtus arvalis</italic><hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">2/99<hr/></td><td align="left" valign="bottom">Germany<hr/></td><td align="left" valign="bottom">79/99<hr/></td><td align="left" valign="bottom">Germany<hr/></td></tr><tr><td align="left" valign="bottom">23/99<hr/></td><td align="left" valign="bottom">Germany<hr/></td><td align="left" valign="bottom">151/99<hr/></td><td align="left" valign="bottom">Germany<hr/></td></tr><tr><td align="left" valign="bottom">42/99<hr/></td><td align="left" valign="bottom">Germany<hr/></td><td align="left" valign="bottom">220/99<hr/></td><td align="left" valign="bottom">Germany<hr/></td></tr><tr><td align="left" valign="bottom">92/99<hr/></td><td align="left" valign="bottom">Germany<hr/></td><td align="left" valign="bottom"><italic>Sorex araneus</italic><hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">106/99<hr/></td><td align="left" valign="bottom">Germany<hr/></td><td align="left" valign="bottom">S1<hr/></td><td align="left" valign="bottom">UK<hr/></td></tr><tr><td align="left" valign="bottom">129/99<hr/></td><td align="left" valign="bottom">Germany<hr/></td><td align="left" valign="bottom">S2<hr/></td><td align="left" valign="bottom">UK<hr/></td></tr><tr><td align="left" valign="bottom">159/99<hr/></td><td align="left" valign="bottom">Germany<hr/></td><td align="left" valign="bottom">S3<hr/></td><td align="left" valign="bottom">UK<hr/></td></tr><tr><td align="left" valign="bottom">240/00<hr/></td><td align="left" valign="bottom">Germany<hr/></td><td align="left" valign="bottom"><italic>Homo sapiens</italic><hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">241/00<hr/></td><td align="left" valign="bottom">Germany<hr/></td><td align="left" valign="bottom">Human 96HE27<hr/></td><td align="left" valign="bottom">USA<hr/></td></tr><tr><td align="left" valign="bottom">252/00<hr/></td><td align="left" valign="bottom">Germany<hr/></td><td align="left" valign="bottom">Human 98HE4<hr/></td><td align="left" valign="bottom">USA<hr/></td></tr><tr><td align="left" valign="bottom">278/00<hr/></td><td align="left" valign="bottom">Germany<hr/></td><td align="left" valign="bottom">Human HGE-1<sup></sup><hr/></td><td align="left" valign="bottom">USA<hr/></td></tr><tr><td align="left" valign="bottom">289/00<hr/></td><td align="left" valign="bottom">Germany<hr/></td><td align="left" valign="bottom"><italic>Canis lupus familiaris</italic><hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">331/00<hr/></td><td align="left" valign="bottom">Germany<hr/></td><td align="left" valign="bottom">Dog Martin<sup></sup><hr/></td><td align="left" valign="bottom">USA<hr/></td></tr><tr><td align="left" valign="bottom">338/00<hr/></td><td align="left" valign="bottom">Germany<hr/></td><td align="left" valign="bottom"><italic>Equus caballus</italic><hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">354/00<hr/></td><td align="left" valign="bottom">Germany<hr/></td><td align="left" valign="bottom">Horse 32 FR<hr/></td><td align="left" valign="bottom">Switzerland<hr/></td></tr><tr><td align="left" valign="bottom">362/00<hr/></td><td align="left" valign="bottom">Germany<hr/></td><td align="left" valign="bottom"><italic>Bos taurus</italic><hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">414/00<hr/></td><td align="left" valign="bottom">Germany<hr/></td><td align="left" valign="bottom">Cow A262<hr/></td><td align="left" valign="bottom">Germany<hr/></td></tr><tr><td align="left" valign="bottom">426/00<hr/></td><td align="left" valign="bottom">Germany<hr/></td><td align="left" valign="bottom"><italic>Ovis aries</italic><hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">523/00<hr/></td><td align="left" valign="bottom">Germany<hr/></td><td align="left" valign="bottom">sheep F1480<hr/></td><td align="left" valign="bottom">Germany<hr/></td></tr><tr><td align="left" valign="bottom"><italic>Microtus agrestis</italic><hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">F1<hr/></td><td align="left" valign="bottom">UK<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td></tr><tr><td align="left">F6</td><td align="left">UK</td><td align="left"> </td><td align="left"> </td></tr></tbody></table><table-wrap-foot><p><sup></sup>[<xref ref-type="bibr" rid="B36">36</xref>], <sup>*</sup>[<xref ref-type="bibr" rid="B37">37</xref>].</p></table-wrap-foot></table-wrap></sec><sec><title>PCR analyses and sequencing</title><p>1 to 2 μl of DNA were used as template in a 50 μl reaction mixture containing 50 mM KCl, 20 mM Tris–HCl (pH 8.4), 2 mM MgCl<sub>2</sub>, 0.2 mM desoxynucleoside triphosphates, 0.4 μM of each primer, and 0.2 μl (1U) of <italic>Taq</italic> DNA Polymerase (Invitrogen, Karlsruhe, Germany). PCRs were performed using the GeneAmp PCR System 9700 (Applied Biosystems, Darmstadt, Germany) under the following conditions: initial denaturation at 94°C for 3 min, 40 cycles consisting of denaturation at 94°C for 30 s, annealing at the predicted melting temperature of the primers minus 4°C for 30 s, extension at 72°C for 30 s per amplification of 500 bp, and a final extension at 72°C for 10 min. Nested PCR amplification and sequencing of the <italic>A. phagocytophilum</italic> 16S rRNA gene [<xref ref-type="bibr" rid="B27">27</xref>,<xref ref-type="bibr" rid="B38">38</xref>] and of the <italic>ankA</italic> gene clusters I [<xref ref-type="bibr" rid="B39">39</xref>] and IV [<xref ref-type="bibr" rid="B11">11</xref>] were performed as described previously. Nested PCR amplification and sequencing of the <italic>ankA</italic> gene cluster V was achieved as shown in Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Table S1. The sequence of the complete ORF was obtained by assembling the sequences of the six nested PCR products. Nucleotide sequences of primers (Metabion, Martinsried, Germany) are summarized in Additional file <xref ref-type="supplementary-material" rid="S2">2</xref>: Table S2. Nested PCR products were directly sequenced bidirectionally using a 3130 Genetic Analyzer (Applied Biosystems) and the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems).</p></sec><sec><title>Data analysis</title><p>Sequences were edited and assembled with the SeqMan program of the DNASTAR package (Lasergene, Madison, WI). For phylogenetic analysis of the 16S rRNA or <italic>ankA</italic> gene sequences the program MEGA 5.1 [<xref ref-type="bibr" rid="B40">40</xref>] was used. Sequences were aligned by ClustalW applying the IUB matrix (16S rRNA gene) or codon-aligned applying the PAM (Dayhoff) matrix. Tree construction was achieved by the neighbor-joining method with the complete deletion option using the Jukes-Cantor matrix for nucleotide sequences and the PAM (Dayhoff) matrix for protein sequences, respectively. Bootstrap analysis was conducted with 1,000 replicates. Average distances within and net average distances between <italic>ankA</italic> gene clusters were computed using the same parameters as for tree construction. Protein sequences were analyzed for Pfam domain matches (<ext-link ext-link-type="uri" xlink:href="http://pfam.sanger.ac.uk/">http://pfam.sanger.ac.uk/</ext-link>) and for tyrosine kinase group phosphorylation sites (<ext-link ext-link-type="uri" xlink:href="http://scansite.mit.edu/">http://scansite.mit.edu/</ext-link>). Nucleotide consensus sequences were calculated for each <italic>ankA</italic> gene cluster with consensus maker v2.0.0 using the most common character and breaking ties with IUPAC characters (<ext-link ext-link-type="uri" xlink:href="http://www.hiv.lanl.gov/content/sequence/HIV/HIVTools.html">http://www.hiv.lanl.gov/content/sequence/HIV/HIVTools.html</ext-link>). The consensus sequences were codon-aligned by ClustalW applying the PAM (Dayhoff) matrix. The alignment was analyzed for recombination by Recco [<xref ref-type="bibr" rid="B41">41</xref>] with the Hamming mutation cost matrix and gap extension costs of 0.2. Events with seq <italic>p</italic>-values of < 0.5 and savings ≥ 5 were regarded as significant.</p></sec><sec><title>Accession numbers</title><p>GenBank nucleotide accession numbers of 16S rRNA and <italic>ankA</italic> gene sequences are shown in Table <xref ref-type="table" rid="T2">2</xref>.</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>GenBank nucleotide accession numbers</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Sample</bold></th><th align="left"><bold>16S rRNA</bold></th><th align="left"><bold> <italic>ankA</italic> </bold></th><th align="left"><bold>Sample</bold></th><th align="left"><bold>16S rRNA</bold></th><th align="left"><bold> <italic>ankA</italic> </bold></th></tr></thead><tbody valign="top"><tr><td colspan="3" align="left" valign="bottom"><bold>Voles</bold><hr/></td><td colspan="3" align="left" valign="bottom"><bold>Voles</bold><hr/></td></tr><tr><td align="left" valign="bottom">2/99<hr/></td><td align="left" valign="bottom">KC740418<hr/></td><td align="left" valign="bottom">KC740451<hr/></td><td align="left" valign="bottom">426/00<hr/></td><td align="left" valign="bottom">KC740435<hr/></td><td align="left" valign="bottom">KC740468<hr/></td></tr><tr><td align="left" valign="bottom">23/99<hr/></td><td align="left" valign="bottom">KC740419<hr/></td><td align="left" valign="bottom">KC740452<hr/></td><td align="left" valign="bottom">523/00<hr/></td><td align="left" valign="bottom">KC740436<hr/></td><td align="left" valign="bottom">KC740469<hr/></td></tr><tr><td align="left" valign="bottom">42/99<hr/></td><td align="left" valign="bottom">KC740420<hr/></td><td align="left" valign="bottom">KC740453<hr/></td><td align="left" valign="bottom">F1<hr/></td><td align="left" valign="bottom">KC740437<hr/></td><td align="left" valign="bottom">KC740470<hr/></td></tr><tr><td align="left" valign="bottom">79/99<hr/></td><td align="left" valign="bottom">KC740439<hr/></td><td align="left" valign="bottom">KC740472<hr/></td><td align="left" valign="bottom">F6<hr/></td><td align="left" valign="bottom">KC740438<hr/></td><td align="left" valign="bottom">KC740471<hr/></td></tr><tr><td align="left" valign="bottom">92/99<hr/></td><td align="left" valign="bottom">KC740421<hr/></td><td align="left" valign="bottom">KC740454<hr/></td><td colspan="3" align="left" valign="bottom"><bold>Shrews</bold><hr/></td></tr><tr><td align="left" valign="bottom">106/99<hr/></td><td align="left" valign="bottom">KC740422<hr/></td><td align="left" valign="bottom">KC740455<hr/></td><td align="left" valign="bottom">S1<hr/></td><td align="left" valign="bottom">KC740442<hr/></td><td align="left" valign="bottom">KC740475<hr/></td></tr><tr><td align="left" valign="bottom">129/99<hr/></td><td align="left" valign="bottom">KC740423<hr/></td><td align="left" valign="bottom">KC740456<hr/></td><td align="left" valign="bottom">S2<hr/></td><td align="left" valign="bottom">KC740443<hr/></td><td align="left" valign="bottom">KC740476<hr/></td></tr><tr><td align="left" valign="bottom">151/99<hr/></td><td align="left" valign="bottom">KC740440<hr/></td><td align="left" valign="bottom">KC740473<hr/></td><td align="left" valign="bottom">S3<hr/></td><td align="left" valign="bottom">KC740444<hr/></td><td align="left" valign="bottom">KC740477<hr/></td></tr><tr><td align="left" valign="bottom">159/99<hr/></td><td align="left" valign="bottom">KC740424<hr/></td><td align="left" valign="bottom">KC740457<hr/></td><td colspan="3" align="left" valign="bottom"><bold>Humans</bold><hr/></td></tr><tr><td align="left" valign="bottom">220/99<hr/></td><td align="left" valign="bottom">KC740441<hr/></td><td align="left" valign="bottom">KC740474<hr/></td><td align="left" valign="bottom">Human 96HE27<hr/></td><td align="left" valign="bottom">KC740446<hr/></td><td align="left" valign="bottom">KC740478<hr/></td></tr><tr><td align="left" valign="bottom">240/00<hr/></td><td align="left" valign="bottom">KC740425<hr/></td><td align="left" valign="bottom">KC740458<hr/></td><td align="left" valign="bottom">Human 98HE4<hr/></td><td align="left" valign="bottom">KC740447<hr/></td><td align="left" valign="bottom">KC740479<hr/></td></tr><tr><td align="left" valign="bottom">241/00<hr/></td><td align="left" valign="bottom">KC740426<hr/></td><td align="left" valign="bottom">KC740459<hr/></td><td align="left" valign="bottom">Human HGE-1<hr/></td><td align="left" valign="bottom">KC740445<hr/></td><td align="left" valign="bottom">KC740480<hr/></td></tr><tr><td align="left" valign="bottom">252/00<hr/></td><td align="left" valign="bottom">KC740427<hr/></td><td align="left" valign="bottom">KC740460<hr/></td><td colspan="3" align="left" valign="bottom"><bold>Dog</bold><hr/></td></tr><tr><td align="left" valign="bottom">278/00<hr/></td><td align="left" valign="bottom">KC740428<hr/></td><td align="left" valign="bottom">KC740461<hr/></td><td align="left" valign="bottom">Dog Martin<hr/></td><td align="left" valign="bottom">KC740448<hr/></td><td align="left" valign="bottom">KC740481<hr/></td></tr><tr><td align="left" valign="bottom">289/00<hr/></td><td align="left" valign="bottom">KC740429<hr/></td><td align="left" valign="bottom">KC740462<hr/></td><td colspan="3" align="left" valign="bottom"><bold>Horse</bold><hr/></td></tr><tr><td align="left" valign="bottom">331/00<hr/></td><td align="left" valign="bottom">KC740430<hr/></td><td align="left" valign="bottom">KC740463<hr/></td><td align="left" valign="bottom">Horse 32 FR<hr/></td><td align="left" valign="bottom">JN247407<hr/></td><td align="left" valign="bottom">JN247406<hr/></td></tr><tr><td align="left" valign="bottom">338/00<hr/></td><td align="left" valign="bottom">KC740431<hr/></td><td align="left" valign="bottom">KC740464<hr/></td><td colspan="3" align="left" valign="bottom"><bold>Cow</bold><hr/></td></tr><tr><td align="left" valign="bottom">354/00<hr/></td><td align="left" valign="bottom">KC740432<hr/></td><td align="left" valign="bottom">KC740465<hr/></td><td align="left" valign="bottom">Cow A262<hr/></td><td align="left" valign="bottom">KC740449<hr/></td><td align="left" valign="bottom">KC740482<hr/></td></tr><tr><td align="left" valign="bottom">362/00<hr/></td><td align="left" valign="bottom">KC740433<hr/></td><td align="left" valign="bottom">KC740466<hr/></td><td colspan="3" align="left" valign="bottom"><bold>Sheep</bold><hr/></td></tr><tr><td align="left">414/00</td><td align="left">KC740434</td><td align="left">KC740467</td><td align="left">Sheep F1480</td><td align="left">KC740450</td><td align="left">KC740483</td></tr></tbody></table></table-wrap></sec></sec><sec sec-type="results"><title>Results</title><sec><title>16S rRNA gene sequences</title><p>Seven of the 16S rRNA gene sequences from voles contained ambiguous nucleotides, indicating multiple infections with several 16S rRNA genotypes, a phenomenon that was observed already earlier in animal and tick samples [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B27">27</xref>]. 14 of the 27 small mammals (11 voles and three shrews) harboured an <italic>A. phagocytophilum</italic> variant identical to [GenBank: M73220]. This genotype is widespread mainly in ruminants, but was also detected in voles and shrews [<xref ref-type="bibr" rid="B34">34</xref>,<xref ref-type="bibr" rid="B35">35</xref>,<xref ref-type="bibr" rid="B42">42</xref>]. Two 16S rRNA gene sequences were identical to [GenBank: AY082656] that was found in voles in the United Kingdom [<xref ref-type="bibr" rid="B43">43</xref>], whereas two matched [GenBank: GU236577] originating from red deer in Germany [<xref ref-type="bibr" rid="B11">11</xref>]. Additionally, one vole was infected with an <italic>A. phagocytophilum</italic> variant identical to [GenBank: AY281785] and one with a new variant, respectively.</p></sec><sec><title><italic>ankA</italic> sequences</title><p>Due to the pronounced dissimilarity of the <italic>ankA</italic> gene from voles and shrews to the known <italic>ankA</italic> gene clusters I, II, III, and IV described earlier [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B27">27</xref>], a new set of primers had to be developed for amplification and sequencing of the complete ORF (Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>: Table S1). Despite the <italic>ankA</italic> sequence from one sheep that belonged to cluster IV, all other six samples from humans and animals analysed during this study were part of cluster I. The obtained <italic>ankA</italic> gene sequences from voles and shrews were 99.8% identical to each other at the nucleotide level and 99.6% similar at the protein level. The comparison to 221 sequences (12 from humans, 43 from dogs, 10 from horses, two from cats, 53 from sheep, four from cattle, 47 from roe deer, 12 from red deer, 15 from European bison, 23 from <italic>I. ricinus</italic> ticks) described earlier [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B27">27</xref>] indicated that the 27 samples from voles and shrews belonged to a new <italic>ankA</italic> gene cluster V and revealed nucleotide identities of 59.6% to 68.1% and amino acid similarities of 36.6% to 51.6% to the known clusters (Table <xref ref-type="table" rid="T3">3</xref>). The sequences of <italic>ankA</italic> gene cluster V showed the lowest identities and similarities to all other <italic>ankA</italic> gene clusters indicating that they were most distantly related. The sequences most closely related to <italic>ankA</italic> gene cluster V were those from <italic>ankA</italic> gene cluster IV. However, their identity at the nucleotide level was limited to 68.1% and their similarity at the amino acid level to 51.6% (Table <xref ref-type="table" rid="T3">3</xref>).</p><table-wrap position="float" id="T3"><label>Table 3</label><caption><p><bold>Net average identities and similarities** between the different </bold><bold> <italic>ankA </italic> </bold><bold>gene clusters in percent</bold></p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="left"> </th><th align="center"><bold>Cluster I</bold></th><th align="center"><bold>Cluster II</bold></th><th align="center"><bold>Cluster III</bold></th><th align="center"><bold>Cluster IV</bold></th><th align="center"><bold>Cluster V</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom"><bold>Cluster I</bold><hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">85.4<hr/></td><td align="center" valign="bottom">74.6<hr/></td><td align="center" valign="bottom">69.6<hr/></td><td align="center" valign="bottom">61.3<hr/></td></tr><tr><td align="left" valign="bottom"><bold>Cluster II</bold><hr/></td><td align="center" valign="bottom"><italic>78.4</italic><hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">83.3<hr/></td><td align="center" valign="bottom">73.5<hr/></td><td align="center" valign="bottom">63.0<hr/></td></tr><tr><td align="left" valign="bottom"><bold>Cluster III</bold><hr/></td><td align="center" valign="bottom"><italic>60.5</italic><hr/></td><td align="center" valign="bottom"><italic>71.2</italic><hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">65.7<hr/></td><td align="center" valign="bottom">59.6<hr/></td></tr><tr><td align="left" valign="bottom"><bold>Cluster IV</bold><hr/></td><td align="center" valign="bottom"><italic>59.8</italic><hr/></td><td align="center" valign="bottom"><italic>63.3</italic><hr/></td><td align="center" valign="bottom"><italic>48.3</italic><hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">68.1<hr/></td></tr><tr><td align="left"><bold>Cluster V</bold></td><td align="center"><italic>45.0</italic></td><td align="center"><italic>44.7</italic></td><td align="center"><italic>36.6</italic></td><td align="center"><italic>51.6</italic></td><td align="center"> </td></tr></tbody></table><table-wrap-foot><p>At the nucleotide level (roman), *at the protein level (italics).</p></table-wrap-foot></table-wrap><p>A search against the Pfam domain database demonstrated that all AnkA sequences from voles and shrews contained ankyrin repeats. Furthermore, multiple tyrosine phosphorylation sites were predicted by Scansite (<ext-link ext-link-type="uri" xlink:href="http://scansite.mit.edu/">http://scansite.mit.edu/</ext-link>) at their C-terminal end, one of them displaying a classical EPIYA motif [<xref ref-type="bibr" rid="B44">44</xref>]. As described for AnkA clusters I and IV [<xref ref-type="bibr" rid="B11">11</xref>], the abundant tyrosine phosphorylation sites seemed to be arisen by duplication of direct repeats (Additional file <xref ref-type="supplementary-material" rid="S3">3</xref>: Figure S1).</p></sec><sec><title>Phylogenetic analysis</title><p>A neighbor-joining tree was constructed from the 34 <italic>ankA</italic> gene sequences obtained during this study and 221 sequences (12 from humans, 43 from dogs, 10 from horses, two from cats, 53 from sheep, four from cattle, 47 from roe deer, 12 from red deer, 15 from European bison, 23 from <italic>I. ricinus</italic> ticks) described earlier. As shown in Figure <xref ref-type="fig" rid="F1">1</xref>b, the <italic>A. phagocytophilum</italic> strains from voles and shrews were located on a distinct major branch that was supported by a high bootstrap value of 99%. As described previously [<xref ref-type="bibr" rid="B11">11</xref>], sequences from humans, dogs, horses, and cats were found exclusively in <italic>ankA</italic> gene cluster I. Sequences from sheep, cattle, red deer, and European bison were more heterogenous and belonged with the exception of one red deer sequence to <italic>ankA</italic> gene clusters I and IV. In contrast, sequences from roe deer were almost exclusively found in <italic>ankA</italic> gene clusters II and III. With the exception of <italic>ankA</italic> gene clusters III and V, sequences from <italic>I. ricinus</italic> ticks were scattered around the tree as expected. Using AnkA amino acid sequences similar results were obtained (data not shown). In contrast, on a tree calculated from the 16S rRNA gene sequences, no clear clustering was observed (Figure <xref ref-type="fig" rid="F1">1</xref>a).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Phylogenetic tree of the 16S rRNA (a) and </bold><bold><italic>ankA </italic></bold><bold>(b) gene sequences inferred using the neighbor-joining method.</bold> Only bootstrap values exceeding 95% are shown. The scale bar indicates the number of nucleotide substitutions per site. <bold>(a)</bold> Sequences with ambiguous nucleotides were not included. The final data set contained 497 positions. <bold>(b)</bold> Only bootstrap values of major branches are shown. The final data set contained 2947 positions. Roman numerals indicate <italic>ankA</italic> gene clusters. Symbols: (light green circle) dog, (red circle) human, (light blue circle) horse, (yellow circle) cat, (inverted blue triangle) sheep, (inverted gray triangle) bison, (inverted pink triangle) cow, (orange diamond) red deer, (brown diamond) roe deer, (inverted blue green triangle) vole/shrew, (inverted black triangle) tick.</p></caption><graphic xlink:href="1746-6148-9-235-1"/></fig></sec><sec><title>Recombination analysis</title><p>It is possible that the striking diversity of <italic>ankA</italic> gene sequences could have developed via recombination. To test this hypothesis, we generated nucleotide consensus sequences for each <italic>ankA</italic> gene cluster. A codon-based alignment of the five consensus sequences was created and analyzed applying the Recco method [<xref ref-type="bibr" rid="B41">41</xref>]. Because the sequences contained many repeats near their 3′ ends, the alignment was uncertain in the respective region and contained many gaps. Recco is subject to bias when analysing alignments with large gaps. We therefore further analyzed alignments without repeats as well as alignments without repeats and without any gaps. The results were compared to the analysis including repeats and all gaps. Whilst there was a tendency for Recco to report more and possibly spurious recombination events in the alignment containing repeats and gaps, we could confirm several recombination events with high confidence. Figure <xref ref-type="fig" rid="F2">2</xref> shows the conservative solutions from the analysis without repeats and gaps. Each solution is defined by the calculated recombination breakpoints and the sequence most similar to the putative recombinant between the breakpoints.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Recombination analysis using the Recco method.</bold> The <italic>ankA</italic> open reading frame of clusters I (red), II (yellow), III (green), IV (light blue), and V (dark blue) is shown. Underneath clusters II, IV, and V the conservative solutions from the analysis without repeats and gaps is demonstrated. Each solution is defined by the calculated recombination breakpoints and the sequence most similar to the putative recombinant between the breakpoints. The hypotheses were generated using the Recco method [<xref ref-type="bibr" rid="B41">41</xref>].</p></caption><graphic xlink:href="1746-6148-9-235-2"/></fig></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>In Europe, the reservoir hosts for <italic>A. phagocytophilum</italic> have not been clearly defined to date. The molecular characterization of <italic>A. phagocytophilum</italic> strains using <italic>groESL</italic> and <italic>ankA</italic> gene sequences revealed that red deer [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B30">30</xref>] and wild boar [<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B32">32</xref>] might harbour variants that cause granulocytic anaplasmosis in humans, dogs, and horses. Small mammals were considered as reservoir hosts too, but it was shown that voles were infected with <italic>msp4</italic> genotypes that differed from those of <italic>I. ricinus</italic> ticks [<xref ref-type="bibr" rid="B34">34</xref>]. Because <italic>I. ricinus</italic> is the main vector of granulocytic anaplasmosis in humans and domestic animals in Europe [<xref ref-type="bibr" rid="B2">2</xref>], voles rather seem to be involved in a separate enzootic cycle probably with <italic>I. trianguliceps</italic> as tick vector [<xref ref-type="bibr" rid="B34">34</xref>]. This is in line with our observation that voles and shrews harboured <italic>A. phagocytophilum</italic> strains that belonged to a newly defined distinct <italic>ankA</italic> gene cluster. Interestingly, we did not find sequences from <italic>I. ricinus</italic> ticks to cluster with those from voles and shrews supporting the hypothesis that <italic>A. phagocytophilum</italic> strains circulating in these small rodents are part of a completely separate ecology [<xref ref-type="bibr" rid="B34">34</xref>]. Similarly, the <italic>groESL</italic> variants in voles and shrews from the Asian part of Russia were found to be clearly separated phylogenetically from all other analyzed strains [<xref ref-type="bibr" rid="B42">42</xref>]. This is in contrast to the USA, where small rodents such as the white-footed mouse appear to be reservoir hosts for granulocytic anaplasmosis [<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B17">17</xref>]. Our results from the <italic>ankA</italic>-based phylogeny indicate that voles and shrews harbour <italic>A. phagocytophilum</italic> strains that might not be infectious for humans, dogs, horses, and livestock. However, other rodents species apart from those investigated here, could serve as reservoir hosts in Europe.</p><p>The AnkA protein is suggested to be secreted into host cells via the VirB/VirD-dependent type IV secretion system (T4SS) of <italic>A. phagocytophilum</italic>[<xref ref-type="bibr" rid="B45">45</xref>,<xref ref-type="bibr" rid="B46">46</xref>]. After translocation it is tyrosine phosphorylated and thought to disturb host cell signalling via protein-protein interactions mediated by its ankyrin repeats [<xref ref-type="bibr" rid="B45">45</xref>,<xref ref-type="bibr" rid="B46">46</xref>]. At its C-terminal end AnkA typically contains one classical EPIYA and multiple EPIYA-related motifs [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B47">47</xref>] that undergo tyrosine phosphorylation [<xref ref-type="bibr" rid="B47">47</xref>]. EPIYA motifs of bacterial effector proteins often show numerous duplications [<xref ref-type="bibr" rid="B44">44</xref>]. We described this phenomenon before especially for AnkA clusters I and IV [<xref ref-type="bibr" rid="B11">11</xref>] and show here that this is also true for the AnkA cluster V associated with voles and shrews (Additional file <xref ref-type="supplementary-material" rid="S3">3</xref>: Figure S1).</p><p>For the effector protein CagA of <italic>Helicobacter pylori</italic>, it was shown that its EPIYA motifs expanded via point mutation and recombination [<xref ref-type="bibr" rid="B48">48</xref>]. Our analysis of the five <italic>ankA</italic> consensus sequences revealed that the marked diversity of AnkA could have arisen via recombination as well (Figure <xref ref-type="fig" rid="F2">2</xref>). However, it was not possible to determine which sequences were the ancestral ones. It has been suggested that the diversification of EPIYA motifs may lead to altered or extended target-protein binding capacities [<xref ref-type="bibr" rid="B44">44</xref>]. Therefore, a specific AnkA could mediate a distinct host tropism of a particular <italic>A. phagocytophilum</italic> isolate and be involved in host adapation. Accordingly, variability between strains from different host species was found mainly in the surface-exposed components of the T4SS of <italic>A. phagocytophilum</italic>[<xref ref-type="bibr" rid="B49">49</xref>].</p><p>If the <italic>ankA</italic> gene is indeed involved in host adaptation driven by recombination, the <italic>ankA</italic>-based phylogeny could be disturbed by the fact that one single recombination event can introduce multiple nucleotide exchanges at once. Therefore, other more conserved loci should be used to proof the phylogenetic separation of <italic>A. phagocytophilum</italic> strains from voles and shrews described here. Nevertheless, their marked dissimilarity to all other strains investigated, indicates a long evolutionary distance. As sequence data alone are not able to prove different biological strain properties, in vivo experiments should address whether <italic>A. phagocytophilum</italic> isolates from voles and shrews are infectious for humans, dogs, horses, and livestock.</p><p>Although there might be some sampling error in our data set, voles and shrews are unlikely reservoir hosts for granulocytic anaplasmosis in humans, dogs, horses, and livestock in Europe based on <italic>ankA</italic> genotyping.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>Voles and shrews harbour <italic>A. phagocytophilum</italic> strains that contain <italic>ankA</italic> gene sequences belonging to the newly described cluster V that might have arisen via recombination. Because cluster V <italic>ankA</italic> sequences were restricted to voles and shrews, they are unlikely to serve as reservoir hosts for granulocytic anaplasmosis in humans, dogs, horses, and livestock in Europe.</p></sec><sec><title>Consent</title><p>For Germany, permission to trap rodents using snap traps was given by the District Government Stuttgart, Germany [<xref ref-type="bibr" rid="B50">50</xref>]. For the United Kingdom, protocols for the handling and sampling of wild small mammals were approved by the University of Liverpool Committee on Research Ethics and were conducted in compliance with the terms and conditions of licenses awarded under the UK Government Animals (Scientific Procedures) Act, 1986 [<xref ref-type="bibr" rid="B34">34</xref>].</p><p>The samples of human and domestic animal origin were obtained as part of routine diagnostic evaluation. Informed consent was obtained from the patients and owners, respectively. Human samples 96 HE27 and 98 HE4 were kindly provided by Stephen J. Dumler (The Johns Hopkins School of Medicine, Baltimore, MD), human HGE-1 [<xref ref-type="bibr" rid="B36">36</xref>] and dog Martin [<xref ref-type="bibr" rid="B37">37</xref>] samples by Ulrike G. Munderloh (University of Minnesota, St. Paul, MN), horse sample 32 FR by Daniel Schaarschmidt-Kiener (Laboratory at Zugersee, Hünenberg, Switzerland) and cow A262 and sheep F1480 samples by Martin Ganter (University of Veterinary Medicine, Hannover, Germany).</p></sec><sec><title>Availability of supporting data</title><p>All supporting data are included as additional files.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>KH and RB did the sampling of the small mammal material and isolated the DNA. JM, NW, and FDvL performed DNA amplification and sequencing. FDvL carried out the data analysis and drafted the manuscript. The manuscript was critically read by JM, NW, KH and RB. All authors approved its final version.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional file 1: Table S1</title><p>Primers used for amplification and sequencing of the complete ORF of <italic>ankA</italic> gene cluster V.</p></caption><media xlink:href="1746-6148-9-235-S1.doc"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S2"><caption><title>Additional file 2: Table S2</title><p>Nucleotide sequences of the primers used for amplification and sequencing of the complete ORF of <italic>ankA</italic> gene cluster V.</p></caption><media xlink:href="1746-6148-9-235-S2.doc"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S3"><caption><title>Additional file 3: Figure S1</title><p>Composition of the C-terminal end of cluster V AnkA. The composition of the C-terminal end of cluster V AnkA from 27 voles and shrews is shown. Homologous protein domains are displayed in same colors. Tyrosine phosphorylation motifs predicted by Scansite (<ext-link ext-link-type="uri" xlink:href="http://scansite.mit.edu/">http://scansite.mit.edu/</ext-link>) are indicated.</p></caption><media xlink:href="1746-6148-9-235-S3.ppt"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
Detection of <italic>Brucella abortus</italic> DNA and RNA in different stages of development of the sucking louse <italic>Haematopinus tuberculatus</italic>	<sec><title>Background</title><p>Brucellosis is considered the world’s most widespread zoonotic infection. It causes abortion and sterility in livestock leading to serious economic losses and has even more serious medical impact in humans, since it can be a trigger to more than 500,000 infections per year worldwide. The aim of this study was to evaluate the role of <italic>Haematopinus tuberculatus</italic>, a louse that can parasitize several ruminants, as a new host of brucellosis. Louse specimens were collected from seropositive and seronegative water buffaloes and divided in 3 developmental stages: adults, nymphs and nits. All samples were separately screened for <italic>Brucella</italic> spp. DNA and RNA detection by Real Time PCR. In particular, primers and probes potentially targeting the 16S rRNA and the <italic>Brucella</italic> Cell Surface 31 kDalton Protein (<italic>bcsp31</italic>) genes were used for Real Time PCR and buffalo β <italic>actin</italic> was used as a housekeeping gene to quantify host DNA in the sample. A known amount of <italic>B. abortus</italic> purified DNA was utilized for standard curve preparation and the target DNA amount was divided by the housekeeping gene amount to obtain a normalized target value. A further molecular characterization was performed for <italic>Brucella</italic> strain typing and genotyping by the Bruce-ladder, AMOS-PCR and MLVA assays. Data were statistically analysed by ANOVA.</p></sec><sec><title>Results</title><p><italic>Brucella abortus</italic> DNA and RNA were detected in all developmental stages of the louse, suggesting the presence of viable bacteria. Data obtained by MLVA characterization support this finding, since the strains present in animals and the relative parasites were not always identical, suggesting bacterial replication. Furthermore, the detection of <italic>Brucella</italic> DNA and RNA in nits samples demonstrate, for the first time, a trans-ovarial transmission of the bacterium into the louse.</p></sec><sec><title>Conclusions</title><p>These findings identified <italic>H. tuberculatus</italic> as a new host of brucellosis. Further studies are needed to establish the role of this louse in the epidemiology of the disease, such as vector or reservoir.</p></sec>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Neglia</surname><given-names>Gianluca</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>neglia@unina.it</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Veneziano</surname><given-names>Vincenzo</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>vincenzo.veneziano@unina.it</email></contrib><contrib contrib-type="author" id="A3"><name><surname>De Carlo</surname><given-names>Esterina</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>esterina.decarlo@cert.izsmportici.it</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Galiero</surname><given-names>Giorgio</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>giorgio.galiero@cert.izsmportici.it</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Borriello</surname><given-names>Giorgia</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>giorgia.borriello@cert.izsmportici.it</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Francillo</surname><given-names>Matteo</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>francillomatteo@gmail.com</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Campanile</surname><given-names>Giuseppe</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>giucampa@unina.it</email></contrib><contrib contrib-type="author" id="A8"><name><surname>Zicarelli</surname><given-names>Luigi</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>zicarell@unina.it</email></contrib><contrib contrib-type="author" id="A9"><name><surname>Manna</surname><given-names>Laura</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>laumanna@unina.it</email></contrib>	BMC Veterinary Research	<sec><title>Background</title><p>In recent years an intensification of livestock production systems was observed in many countries, increasing the risk for zoonosis transmission [<xref ref-type="bibr" rid="B1">1</xref>]. Among these, brucellosis, an infection caused by bacteria of the genus <italic>Brucella</italic>, represents one of the main zoonosis worldwide. It causes abortion and sterility in livestock leading to serious economic losses [<xref ref-type="bibr" rid="B2">2</xref>] and has even more serious medical impact in humans, leading to more than 500,000 infections per year worldwide [<xref ref-type="bibr" rid="B3">3</xref>]. Brucellosis has only been controlled and sometimes eradicated in animal reservoirs in developed world by applying strict veterinary hygiene measures, such as control tests, culling infected animals and environment sanitization [<xref ref-type="bibr" rid="B3">3</xref>]. Its eradication is even more difficult in developing countries, because of limited resources to indemnify farmers and their emotional attachment to the animals [<xref ref-type="bibr" rid="B4">4</xref>]. Furthermore, the existence of mammalian wildlife reservoirs of <italic>Brucella</italic> is an obstacle to brucellosis eradication in some countries [<xref ref-type="bibr" rid="B5">5</xref>]. Brucellosis is endemic in most areas of the world due to the difficulties in the application of control/eradication programs and/or other unknown environmental factors may have influenced the spread of the disease.</p><p>Recently, great attention has been focused on the role that some insects can play as reservoirs and vectors of many diseases [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref>]. Cheville et al. [<xref ref-type="bibr" rid="B8">8</xref>] observed that face flies have only a limited capacity to act as short-term carriers of <italic>B. abortus</italic>, since the bacteria did not replicate in the flies, and some bacteria were found to be degraded by secondary lysosomes of the midgut epithelium. The role of some lice to carry <italic>Brucella</italic> spp. was also hypothesized [<xref ref-type="bibr" rid="B9">9</xref>] in Egypt, although the authors failed to detect these bacteria by PCR. Eighteen species of bloodsucking arthropods were identified as natural <italic>Brucella</italic> carriers and 20 species proved to be susceptible to brucellosis infection under experimental conditions [<xref ref-type="bibr" rid="B10">10</xref>]. However, no studies have been performed on the sucking louse <italic>Haematopinus tuberculatus</italic> (Figure <xref ref-type="fig" rid="F1">1</xref>A-<xref ref-type="fig" rid="F1">1</xref>E)<italic>,</italic> Phylum Arthropoda, Class Insecta, Order Phthiraptera, Suborder Anoplura, Family <italic>Haematopinidae</italic>. It has a worldwide distribution, since it has been reported in Asia, Africa, Australia and South America [<xref ref-type="bibr" rid="B11">11</xref>]. In Europe it has been described in Albania, Macedonia, France, England and Italy [<xref ref-type="bibr" rid="B12">12</xref>-<xref ref-type="bibr" rid="B14">14</xref>]. <italic>H. tuberculatus</italic> lives as a permanent ectoparasite and undergoes a simple life cycle. Transition from egg to three nymphal instars to adults (Figure <xref ref-type="fig" rid="F2">2</xref>), in optimal environmental conditions, is completed on the host in 21–27 days [<xref ref-type="bibr" rid="B15">15</xref>]. Cattle [<xref ref-type="bibr" rid="B11">11</xref>], camel, bison and water buffalo are susceptible to lice infestation [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B15">15</xref>]. Water buffalo is better adapted to satisfy animal protein demand in tropical countries, where 98% of the world population is bred, but it is also an important milk producer in some developed countries, like Italy, where its breeding has reached a great level of innovation, similar to that in cattle [<xref ref-type="bibr" rid="B16">16</xref>]. Recently, it has been demonstrated that <italic>H. tuberculatus</italic> is as potential vector of <italic>Anaplasma marginale</italic>[<xref ref-type="bibr" rid="B17">17</xref>]. It may therefore have a similar role in the transmission of other diseases agents, such as <italic>Brucella</italic> spp. In some areas, such as Southern Italy, brucellosis is still endemic despite the application of an eradication program based on a test-and-slaughter approach. Recently, diagnostic molecular techniques have been successfully utilized to identify <italic>Brucella</italic> DNA at genus, species and even biovar levels [<xref ref-type="bibr" rid="B18">18</xref>-<xref ref-type="bibr" rid="B20">20</xref>]. Real-time PCR constitutes a further technological improvement for the molecular identification and quantification of the genus <italic>Brucella</italic> and for the differentiation of its species. This is a rapid, sensitive and specific diagnostic tool, characterized by a low risk of cross-contamination [<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B22">22</xref>]. Furthermore, <italic>Brucella</italic> detection by PCR-based methods is simpler, faster and less hazardous than conventional methods.</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold> <italic>Haematopinus tuberculatus </italic> </bold><bold>at different stages: A- female, B- male, C- third stage nymph, D- second stage nymph, E- first stage nymph.</bold></p></caption><graphic xlink:href="1746-6148-9-236-1"/></fig><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold> <italic>Haematopinus tuberculatus</italic> </bold><bold>: hatching phase of a nit with the emergence of a nymph.</bold></p></caption><graphic xlink:href="1746-6148-9-236-2"/></fig><p>The aim of this study was to detect the presence of <italic>Brucella</italic> spp. DNA and RNA in different developmental stages of <italic>H. tuberculatus</italic> in order to evaluate its possible role as vector of this bacterium.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Ethics</title><p>The investigation was approved by the Animal Ethics Committee of the University of Naples, Federico II.</p></sec><sec><title>Farms and animals</title><p>The animals involved in this study had a naturally-acquired louse infestation. They were bred in a commercial farm located in the South of Italy, where an eradication program based on a test-and-slaughter approach was applied by the Italian Veterinary Health Service. In order to carry out a taxonomic identification, a significant number of lice (about 50) were collected in each farm before the beginning of the trial from 5 randomly selected adult water buffaloes.</p><p>Louse specimens were examined on slides under optical (Leica DM 750 HD) and dissection microscopes (Leica EZ4 HD). Species determination was based on morphological keys previously proposed by several authors [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B15">15</xref>].</p><p>The study was performed on 72 adult water buffaloes bred in six farms located in the South of Italy. Thirty-six infected animals belonging to three farms in which the presence of <italic>Brucella</italic> spp. was detected by the Italian Veterinary Health Service within the National Brucellosis Eradication Program. According to this program, the farms were subjected to periodical controls (every 21 days) from the Italian Veterinary Health Service consisting of conventional serological tests, such as Rose Bengal Test (RBT) and Complement Fixation Test (CFT), on blood samples for the detection of anti-<italic>Brucella</italic> antibodies [<xref ref-type="bibr" rid="B23">23</xref>]. The remaining 36 animals originated from three different farms, historically brucellosis-free for at least 20 years, where all the animals were subjected to the same controls every 6 months.</p><p>According to RBT and CFT results, 36 seropositive water buffaloes (CFT titre ≥160 I.U.), were selected from the three infected farms (12 water buffaloes for each sampling). Simultaneously, 36 seronegative water buffaloes were randomly selected from the three brucellosis-free farms.</p></sec><sec><title>Louse identification and collection</title><p>A total of 6 samples of lice was collected from each animal using an entomological pin and fixed in 70% ethanol. Two samples containing ten adults, two samples containing ten nymphs and two samples containing thirty nits of <italic>H. tuberculatus</italic> were collected in each tube. Three samples (one sample representing each stage) were utilized for DNA extraction, while the remaining were stored at - 80°C for mRNA extraction.</p><p>All louse stages were analyzed separately by real-time PCR.</p></sec><sec><title>Blood and tissue collection</title><p>Two aliquots of blood samples were collected from the jugular vein of each selected water buffalo. All the samples were taken to the laboratory within two hours of collection. The first was utilized to obtain serum for carrying out RBT and CFT analyses. The second aliquot of whole blood was collected in tubes with EDTA and stored at −20°C until DNA extraction and real-time PCR assay was performed as described below. Seropositive water buffaloes were progressively eliminated according to the Italian brucellosis eradication program, and mammary lymph-nodes were sampled during slaughtering.</p></sec><sec><title>DNA extraction</title><p>QIAamp Blood Kit was used to extract DNA from 200 μl of blood samples, mammary lymph-nodes and 100 μl of 10<sup>9</sup> CFU/ml <italic>Brucella abortus</italic> cultures. DNA extraction from <italic>H. tuberculatus</italic> samples was performed by using the QIAamp DNA Mini Kit (Qiagen, Santa Clarita, CA, USA), according to the supplier’s instructions with some modifications. In particular, the lice were cut, placed in an eppendorf tube, and incubated in the lysis buffer with 50 μl of proteinase K (20 mg/ml) overnight at 56°C. DNA was eluted with 100 μl of the supplied buffer pre-heated at 70°C. The concentration and purity of extracted DNA was assessed by measuring spectrophotometrically the absorbance at 260 nm and 280 nm, respectively, and by gel electhrophoresis.</p></sec><sec><title>Molecular characterization and MLVA analysis</title><p>Molecular techniques were carried out on new samples collected from one seropositive farm. In particular, five seropositive water buffaloes were randomly chosen and, from each animal, a new double lice sampling was performed, the first one consisting of the collection of ten adults, ten nymphs and thirty nits of <italic>H. tuberculatus</italic> individually stored; the second one consisting of the collection of ten adults, ten nymphs and thirty nits of <italic>H. tuberculatus</italic> pooled according to the developmental stage in three separate tubes.</p><p><italic>Brucella</italic> strain typing was performed by the Bruce-ladder and AMOS-PCR assays [<xref ref-type="bibr" rid="B22">22</xref>] carried out on the DNA extracted from mammary lymph-nodes and parasite pools. The DNA extracted from lymph-nodes and individual parasite samples was analyzed by the MLVA-16 typing technique, as elsewhere described [<xref ref-type="bibr" rid="B20">20</xref>,<xref ref-type="bibr" rid="B24">24</xref>-<xref ref-type="bibr" rid="B26">26</xref>]. The 16 primer pairs were divided into two groups: panel 1 (<italic>loci</italic> Bruce06, Bruce08, Bruce11, Bruce12, Bruce42, Bruce43, Bruce45, and Bruce55) and panel 2 (<italic>loci</italic> Bruce04, Bruce07, Bruce09, Bruce16, Bruce18, Bruce19, Bruce21, and Bruce30). Amplifications were initiated by denaturing the sample for 3 min at 94°C, followed by 30 cycles at 94°C for 30 s, 60°C for 30 s, and 72°C for 50 s. After the last cycle samples were incubated for an additional 7 min at 72°C before they were stored at 4°C. All the forward primers were labeled with a fluorophore (either FAM or Vic or Ned or Pet). PCR products were mixed together in a ratio 1:1:1:1 to obtain four different mixtures each one containing 4 amplicons labeled with 4 different fluorophores. The mixtures were then denatured in presence of Hi-Di formamide and analyzed by capillary electrophoresis with a 310 Genetic Analyzer (Applied Biosystems, Foster City, CA) equipped with a 47 cm long and 50 μm section capillary filled with the separation medium POP-4 polymer. PCR products relative to the <italic>loci</italic> Bruce06, Bruce11 and Bruce42 were also stained with ethidium bromide and resolved by 2.5% agarose gel electrophoresis to visualize eventual amplicons greater than 500 bp.</p></sec><sec><title>RNA isolation and production of cDNAs</title><p>Total RNA was extracted from 100 μl of 10<sup>9</sup><italic>Brucella abortus</italic> cultures and louse samples by using the RNAeasy Mini Kit (Qiagen, Santa Clarita, CA, USA), according to manufacturer’s instructions. The RNA was resuspended in 100 μl of diethyl pyrocarbonate (DEPC) treated water, and stored at −80°C until use.</p><p>Synthesis of cDNA was performed by using a reverse transcription system (Im Prom II Reverse Transcription System Promega, Madison, WI, USA).</p></sec><sec><title>Primers and probes</title><p>All DNA and cDNA samples were tested by real time PCR by using designed primers and probes potentially targeting the 16S rRNA and the <italic>Brucella</italic> Cell Surface 31kDalton Protein (<italic>bcsp31</italic>) genes, which are highly conserved in 6 species of the genus <italic>Brucella</italic>[<xref ref-type="bibr" rid="B27">27</xref>-<xref ref-type="bibr" rid="B29">29</xref>]. Buffalo β <italic>actin</italic> was used as a housekeeping gene [<xref ref-type="bibr" rid="B30">30</xref>] to quantify host DNA in the sample. All primers and probes were designed by Primer Express Software (Applied Biosystems), according to technical parameters indicating a low level of penalty coupling factor (Table <xref ref-type="table" rid="T1">1</xref>). The fluorogenic probes were synthesized by using a FAM reporter molecule attached to the 5′ end, and a TAMRA quencer linked to the 3′ end (Applied Biosystems, Foster City, CA, USA).</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p><bold>Original real time PCR primers and probes used in this study for 16S RNA, 31 KDa (bcsp31</bold><bold> <italic>) </italic> </bold><bold>and buffalo β actin genes</bold></p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"> </th><th align="left"><bold>Forward primer</bold></th><th align="left"><bold>Reverse primer</bold></th><th align="left"><bold>Probe</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">16S RNA<hr/></td><td align="left" valign="bottom">5′- GCGCGTAAGGATGCAAACAT -3′<hr/></td><td align="left" valign="bottom">5′- CTTGCCTTTCAGGTCTGC-3′<hr/></td><td align="left" valign="bottom">5′- GGCTCATCCAGCGAAACG -3′<hr/></td></tr><tr><td align="left" valign="bottom">31 KDa<hr/></td><td align="left" valign="bottom">5′- AAACGGTAGGTTGCCTAGAG -3′<hr/></td><td align="left" valign="bottom">5′- AATGCCTTGTAGGTCTTT-3′<hr/></td><td align="left" valign="bottom">5′- TTATCATCCGGTGAAGAC -3′<hr/></td></tr><tr><td align="left">β actin</td><td align="left">5′-CTGGCACCACACCTTCTACAA -3′</td><td align="left">5′- GCCTCGGTCAGCAGCA -3′</td><td align="left">5′- CCACGCGCAGCTCG -3′</td></tr></tbody></table></table-wrap></sec><sec><title>Real time PCR</title><p>Real-time PCR was performed to amplify DNA and cDNA as previously described [<xref ref-type="bibr" rid="B31">31</xref>]. Serial 10-fold dilutions of a known amount (210<sup>9</sup> CFU) of <italic>B. abortus</italic> purified DNA were utilized for standard curve preparation. In each real time PCR run, standards, samples, and negative controls were analyzed in triplicate. For each sample, the cycle threshold (Ct) value was calculated by determining the point at which the fluorescence exceeded the threshold limit. The detection range for each set of primers and probe was from 2 × 10<sup>9</sup> to 2 × 10<sup>1</sup> CFU. The standard curve, calculated by independent experiments, was linear over an at least 6-log range of DNA or cDNA concentration points, with an average correlation coefficient of 0.988. The difference for each point of the curve was one log factor. The target DNA amount was divided by the housekeeping gene amount to obtain a normalized target value.</p></sec><sec><title>Statistical analysis</title><p>Statistical analysis of data was performed by ANOVA [<xref ref-type="bibr" rid="B32">32</xref>]. The mean quantity of colony forming unit recorded in samples of <italic>H. tuberculatus</italic> collected from seropositive (n = 36) animals were compared for both 16S rRNA and the <italic>Brucella</italic> spp. Cell Surface 31kDalton Protein (<italic>bcsp31</italic>) genes. The prevalence of positive samples among different groups was evaluated by the chi-square test [<xref ref-type="bibr" rid="B32">32</xref>].</p></sec></sec><sec sec-type="results"><title>Results</title><sec><title>DNA detection and quantification</title><p>The DNA of the <italic>Brucella</italic> spp. 16S rRNA gene was amplified in 55.6% of the adult and nymph samples, collected from seropositive water buffaloes, whereas it was never detected in any sample collected from seronegative animals. In particular, DNA was amplified in both adult and nymph specimens collected from 12 water buffaloes, in adult specimens collected from 12 different water buffaloes, and in nymph specimens from 4 different water buffaloes.</p><p>The <italic>Brucella</italic> spp. <italic>bcsp31</italic> gene was amplified in 44.4% of adult and nymph samples collected from seropositive water buffaloes. It was never detected in seronegative water buffaloes. The DNA was amplified in all the adult and nymph specimens collected from 3 water buffaloes, in adult specimens collected from 5 different water buffaloes, and in nymph specimens from 5 different water buffaloes. Interestingly, all the samples positive to <italic>bcsp31</italic> amplification were also positive to 16S rRNA (Table <xref ref-type="table" rid="T2">2</xref>).</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p><bold>Frequency of detection of </bold><bold> <italic>Brucella </italic> </bold><bold>spp. 16S rRNA and 31 Kd protein genes DNA and cDNA in different samples of </bold><bold> <italic>H. tuberculatus </italic> </bold><bold>collected from seropositive and seronegative buffaloes</bold></p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="left" valign="bottom"><bold>Sample</bold><hr/></th><th colspan="8" align="center" valign="bottom"><bold>Frequency (% positive)</bold><hr/></th></tr><tr><th align="left" valign="bottom"> <hr/></th><th colspan="4" align="center" valign="bottom"><bold>Seropositive buffaloes</bold><hr/></th><th colspan="4" align="center" valign="bottom"><bold>Seronegative buffaloes</bold><hr/></th></tr><tr><th align="left" valign="bottom"> <hr/></th><th colspan="2" align="center" valign="bottom"><bold>16S rRNA</bold><hr/></th><th colspan="2" align="center" valign="bottom"><bold>31 Kd protein</bold><hr/></th><th colspan="2" align="center" valign="bottom"><bold>16S rRNA</bold><hr/></th><th colspan="2" align="center" valign="bottom"><bold>31 Kd protein</bold><hr/></th></tr><tr><th align="left"> </th><th align="center"><bold>DNA</bold></th><th align="center"><bold>cDNA</bold></th><th align="center"><bold>DNA</bold></th><th align="center"><bold>cDNA</bold></th><th align="center"><bold>DNA</bold></th><th align="center"><bold>cDNA</bold></th><th align="center"><bold>DNA</bold></th><th align="center"><bold>cDNA</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom"><bold>Adults</bold><hr/></td><td align="center" valign="bottom">24/36 (66.7)<sup>AB</sup><hr/></td><td align="center" valign="bottom">20/24 (83.3)<hr/></td><td align="center" valign="bottom">8/36 (22.2)<hr/></td><td align="center" valign="bottom">4/8 (50.0)<hr/></td><td align="center" valign="bottom">0/36 (0.0)<hr/></td><td align="center" valign="bottom">NP<hr/></td><td align="center" valign="bottom">0/36 (0.0)<hr/></td><td align="center" valign="bottom">NP<hr/></td></tr><tr><td align="left" valign="bottom"><bold>Nymphs</bold><hr/></td><td align="center" valign="bottom">16/36 (44.4)<sup>A</sup><hr/></td><td align="center" valign="bottom">13/16 (81.3)<hr/></td><td align="center" valign="bottom">8/36 (22.2)<hr/></td><td align="center" valign="bottom">4/7 (57.1)<hr/></td><td align="center" valign="bottom">0/36 (0.0)<hr/></td><td align="center" valign="bottom">NP<hr/></td><td align="center" valign="bottom">0/36 (0.0)<hr/></td><td align="center" valign="bottom">NP<hr/></td></tr><tr><td align="left"><bold>Nits</bold></td><td align="center">16/18 (88.9)<sup>B</sup></td><td align="center">14/16 (87.5)</td><td align="center">8/18 (44.4)</td><td align="center">5/8 (62.5)</td><td align="center">0/36 (0.0)</td><td align="center">NP</td><td align="center">0/36 (0.0)</td><td align="center">NP</td></tr></tbody></table><table-wrap-foot><p>The amount of DNA and cDNA were normalized by using the housekeeping (buffalo β actin gene).</p><p>RNA was not detected in one sample.</p><p>NP = Not performed.</p><p>Values with different superscripts within columns are significantly different (<sup>A,B,</sup> P < 0.01).</p></table-wrap-foot></table-wrap><p>Regarding the stage of development (Table <xref ref-type="table" rid="T2">2</xref>), the DNA of <italic>Brucella</italic> spp. was amplified in 66.% and 22.2% of the adult specimens, by 16S rRNA and <italic>bcsp31</italic> genes, respectively, whereas it was detected in 44.4% and 22.2% of the nymphs specimens, by 16S rRNA and <italic>bcsp31</italic> genes, respectively. The mean quantity of colony forming units (CFU) per ml amplified by the 16S rRNA gene were similar for both adult and nymph specimens, whereas the CFU amplified by the <italic>bcsp31</italic> gene was higher (P < 0.05) for adults than for nymphs (Figure <xref ref-type="fig" rid="F3">3</xref>).</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Mean quantity of </bold><bold><italic>Brucella </italic></bold><bold>spp. DNA amplified in adult, nynphs and nits samples of </bold><bold><italic>Haematopinus tuberculatus </italic></bold><bold>by real time PCR using two different sets of primers and TaqMan probes specific for 16S rRNA and 31 K dalton genes.</bold> Data are expressed as colony forming units per ml. , **, indicate significant differences (P < 0.05).</p></caption><graphic xlink:href="1746-6148-9-236-3"/></fig><p>Although the nits were collected from all animals, only those belonging to 18 seropositive and 14 seronegative water buffaloes were analyzed by real time PCR assay, since some nits were already hatched, because of the biological cycle of the louse. Interestingly, all samples collected from seronegative animals were negative to real-time PCR, whereas the eggs laid by naturally infected female <italic>H. tuberculatus</italic> lice contained the genomic DNA of <italic>Brucella</italic> spp., as detected by both sets of primers and probes. In this case 88.9% of the samples were positive to 16S rRNA gene detection, whereas the <italic>bcsp31</italic> gene was amplified in only 44.4% of cases (Table <xref ref-type="table" rid="T2">2</xref>). In conclusion, the DNA was amplified in at least one stage of louse development in 28 water buffaloes (77.8%).</p><p>Real time PCR was unable to detect <italic>Brucella</italic> DNA in any blood samples.</p></sec><sec><title>cDNA detection</title><p>As shown in Table <xref ref-type="table" rid="T2">2</xref>, bacterial cDNA of 16S rRNA gene was amplified in 83.3, 81.3 and 87.5% of the samples positive to <italic>Brucella</italic> spp. DNA detection in adult, nymph and nit specimens, respectively. Similarly, bacterial <italic>bcsp31</italic> cDNA was amplified in 50.0, 57.1 and 62.5% of the samples positive to DNA detection in adult, nymph and nit specimens, respectively (Table <xref ref-type="table" rid="T2">2</xref>).</p></sec><sec><title>Molecular characterization and MLVA analysis</title><p>Results from molecular characterization highlighted the presence of <italic>B. abortus</italic> bv. 1 in all lymph-nodes and parasite pools samples. The MLVA assay provided complete genetic profiles from all the lymph-nodes samples and from 38% (19/50) of individual adult parasite samples, 32% (16/50) individual nymph samples and 16% (24/150) individual nit samples. The resulting <italic>Brucella</italic> genetic profiles indicated the prevalence of one main genotype both in water buffaloes and parasites samples (Table <xref ref-type="table" rid="T3">3</xref>). Two nymph samples collected from the same water buffalo exhibited 2 different genotypes, and one adult louse sample collected from a different water buffalo showed an additional different genotype (Table <xref ref-type="table" rid="T3">3</xref>).</p><table-wrap position="float" id="T3"><label>Table 3</label><caption><p><bold>MLVA genetic profiles originated from water buffaloes and </bold><bold> <italic>H. tuberculatus </italic> </bold><bold>samples</bold></p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Sample</bold></th><th align="left"><bold>MLVA genotype</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Water buffaloes<hr/></td><td align="left" valign="bottom">4 5 4 12 2 2 3 3 6 21 8 3 7 3 3 4<hr/></td></tr><tr><td align="left" valign="bottom"><italic>H. tuberculatus</italic> adults<sup><italic>a</italic></sup><hr/></td><td align="left" valign="bottom">4 5 4 12 2 2 3 3 6 21 8 3 7 3 3 4<hr/></td></tr><tr><td align="left" valign="bottom"><italic>H. tuberculatus</italic> adult A<hr/></td><td align="left" valign="bottom">4 5 4 12 2 2 3 3 6 21 8 3 <bold><underline>14</underline></bold> 3 3 4<hr/></td></tr><tr><td align="left" valign="bottom"><italic>H. tuberculatus</italic> nymphs<hr/></td><td align="left" valign="bottom">4 5 4 12 2 2 3 3 6 21 8 3 7 3 3 4<hr/></td></tr><tr><td align="left" valign="bottom"><italic>H. tuberculatus</italic> nymph A<hr/></td><td align="left" valign="bottom">4 5 4 12 2 2 3 3 6 21 8 3 7 3 3 <bold><underline>6</underline></bold><hr/></td></tr><tr><td align="left" valign="bottom"><italic>H. tuberculatus</italic> nymph B<hr/></td><td align="left" valign="bottom">4 5 4 12 2 2 3 3 6 21 8 <bold><underline>11 5</underline></bold> 3 3 <bold><underline>6</underline></bold><hr/></td></tr><tr><td align="left"><italic>H. tuberculatus</italic> nits</td><td align="left">4 5 4 12 2 2 3 3 6 21 8 3 7 3 3 4</td></tr></tbody></table><table-wrap-foot><p><sup><italic>a</italic></sup>All <italic>H. tuberculatus</italic> adult samples except for <italic>H. tuberculatus</italic> adult A.</p><p><sup><italic>b</italic></sup>All <italic>H. tuberculatus</italic> nymph samples except for <italic>H. tuberculatus</italic> nymphs A and B; nymphs A and B were collected from the same water buffalo, while the <italic>H. tuberculatus</italic> adult A was collected from a different water buffalo. Underlined and bold data highlight different loci among tested strains.</p></table-wrap-foot></table-wrap></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>In this study <italic>Brucella</italic> spp. DNA has been detected in all developmental stages of the sucking louse <italic>H. tuberculatus</italic>. The role of some lice to act as carriers for some diseases has been hypothesized for <italic>Haematopinus eurysternus</italic> and <italic>Haematopinus quadripertusus</italic>[<xref ref-type="bibr" rid="B9">9</xref>]. Although the DNA of <italic>Coxiella burnetii</italic> and some species of <italic>Bartonella</italic> were detected by PCR, the authors did not detect <italic>Brucella</italic> spp. DNA [<xref ref-type="bibr" rid="B9">9</xref>]. However, in this study lice were randomly collected and it is not specified if the bovines were infected by brucellosis. This aspect may be responsible for the different results recorded in our study.</p><p>It is known that both adults and nymphs are hematophagous, hence the blood feeding behavior may explain the presence of <italic>Brucella</italic> spp. DNA. However, two interesting aspects need to be considered. Firstly, real-time PCR assays performed on the blood collected from seropositive animals were negative for <italic>Brucella</italic> spp<italic>.</italic> DNA. This result is in agreement with previous studies [<xref ref-type="bibr" rid="B33">33</xref>], in which <italic>Brucella</italic> DNA was detected in milk and lymph tissue samples, rather than in blood. It has been demonstrated that blood is not an adequate substrate to detect the DNA of <italic>Brucella</italic> spp. by real-time PCR, since only a transient, short-lived bacteraemia is described during the infection [<xref ref-type="bibr" rid="B23">23</xref>]. Secondly as the bacteria are taken up by macrophages and non-professional phagocytes, only the white cell pellet may be a worthy template for use in PCR detection. In a recent study it was also observed that <italic>Trypanosoma cruzi</italic> is not detected by PCR in the blood of naturally infected wild rodents (<italic>Octodon degus</italic>), while the protozoon is found in the intestinal contents of two species of insect vector (<italic>Triatoma infestans</italic> and <italic>Mepraia spinolai</italic>) [<xref ref-type="bibr" rid="B34">34</xref>]. This result was explained by the high rate of parasite amplification of epimastigote forms in the intestines of the insects [34]. It cannot be ruled out that a similar phenomenon occurs for <italic>Brucella</italic> spp. in <italic>H. tuberculatus</italic>. The louse may represent a booster for the bacterium and the target DNA may be present in high copy numbers.</p><p>Some studies performed in ticks suggest that the traditional view that arthropods could only acquire infections by feeding on hosts that were parasitaemic, or through transovarial transmission, seems incorrect, since the co-feeding enables microparasite transmission between ticks in absence of a host parasitaemia [<xref ref-type="bibr" rid="B35">35</xref>]. Co-feeding alongside infected ticks increases the chances of transmission of microparasites in new-borne ticks, probably increasing also the transmission capability. This phenomenon has been reported for viruses, such as Thogoto virus and TBE group flaviviruses [<xref ref-type="bibr" rid="B36">36</xref>], and is probably one of the main routes of transmission for <italic>Borrelia burgdorferi</italic>[<xref ref-type="bibr" rid="B37">37</xref>] and <italic>Borrelia afzelii</italic>[<xref ref-type="bibr" rid="B38">38</xref>]. In the last case it was demonstrated that a direct passage of spirochetes between co-feeding vector ticks contributes to the likelihood that the Lyme disease spirochete <italic>B. afzelii</italic> perpetuates in nature. Interestingly, a typical scenario in <italic>H. tuberculatus</italic> infestation is the presence of nymphs and adults in clusters (up to 100 specimens in few cm<sup>2</sup>), especially in some specific regions of the animal [<xref ref-type="bibr" rid="B12">12</xref>]. This condition may explain the presence of <italic>Brucella</italic> spp. DNA in the lice rather than in blood.</p><p>The detection of <italic>Brucella</italic> spp. DNA in nit samples supports an hypothesis of a vertical transmission of bacteria between different phases of development (trans-stadial and trans-ovarial transmissions) of <italic>H. tuberculatus</italic>. Transovarial transmission is considered an important mechanism for maintaining and distributing tick-borne protozoa, bacteria and viruses in nature [<xref ref-type="bibr" rid="B36">36</xref>]<italic>: I</italic>n some cases (such as <italic>Rickettsia rickettsii</italic> infection) transovarial transmission, is probably more important in perpetuating infection in nature than the acquisition of the organism from rickettsaemic hosts, as rickettsaemia in mammalian hosts is generally short lived. Since a short-lived bacteraemia is described also during brucellosis, it is likely that engorged <italic>H. tuberculatus</italic> females are able to transfer the bacteria into the nits.</p><p>Although a similar CFU mean quantity was recorded in adult, nymph and nit samples, DNA of <italic>Brucella</italic> spp. 16S rRNA gene was detected with different prevalence. In particular, a higher prevalence was recorded in nits (around 90%) compared to adult and nymphs (66.7% and 44.4%, respectively). This may suggest that the rate of infection is relatively high in adult lice, which are able to lay a high rate of infected nits, but probably decreases during the hatching, as observed in <italic>Borrelia</italic> infected ticks [<xref ref-type="bibr" rid="B37">37</xref>]. The high resistance of the nits in the environment may also account for <italic>Brucella</italic> spp. survival., especially in some endemic areas.</p><p>However, the detection of <italic>Brucella</italic> spp. DNA in lice and nits does not necessarily demonstrate the presence of viable bacteria. As reported above, <italic>Brucella abortus</italic> was isolated also in the flies, but the bacteria were not able to replicate into the carrier [<xref ref-type="bibr" rid="B8">8</xref>]. The high stability of DNA molecules and the possibility of its persistence following bacterial death cannot be indicative of the presence of viable microbes. Because of its short half life and lability, RNA has been considered a plausible indicator of viability and a diagnostic target for several microbial infections [<xref ref-type="bibr" rid="B39">39</xref>]. The monitoring of bacterial gene expression can be used to characterize the transcriptome of intracellular pathogens and better understand the host:pathogen interaction during infection [<xref ref-type="bibr" rid="B40">40</xref>]. Little information is available about <italic>Brucella</italic> spp. gene expression during host:pathogen interaction, because of the difficulty in obtaining an adequate quantity of good quality eukaryotic RNA-free pathogen RNA for downstream applications [<xref ref-type="bibr" rid="B41">41</xref>]. The isolation of high-quality bacterial mRNA accurately reflected <italic>Brucella abortus</italic> gene expression and demonstrates the presence of whole and viable bacteria, with replication capability.</p><p>This interesting finding is also supported by the MLVA characterization data, since the strains present in animals and the relative parasites were not always identical. This data is indicative of bacterial replication within the parasite and it can not be ruled out that the lice may transmit <italic>B. abortus</italic> infection among the animals, similarly to what has been described for Ixodid ticks in a very old study [<xref ref-type="bibr" rid="B42">42</xref>].</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>This preliminary study gives a new perspective on the epidemiology of brucellosis and identifies <italic>H. tuberculatus</italic> as a new host of the bacterium. The presence of <italic>B. abortus</italic> DNA and RNA in the nits, confirms the presence of viable and whole bacteria and serves as evidence for bacterial transmission between different developmental stages (trans-stadial and trans-ovarian). Further studies are needed to elucidate the role of <italic>H. tuberculatus</italic> as a possible vector of <italic>Brucella abortus</italic>, by in vitro isolation of the bacterium and experimental infection of animals.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>GN, VV and LM designed the experiment. VV and MF performed the parasitological investigation. LM and GC designed primer and probes and carried out the real time PCR assay. GG EdC and GB performed the molecular characterization of bacteria by MLVA. GN, GC LZ were involved in drafting the manuscript. LM and VV performed the statistical analysis of the data. All authors contributed to the analysis of the data, discussion of results and implications and commented on the manuscript at all stages. All authors read and approved the final manuscript.</p></sec><sec><title>Authors’ information</title><p>GN, VV, and LM are aggregate professors at the department of Veterinary Medicine, Federico II University of Naples.</p><p>GC and LZ are full professors at the department of Veterinary Medicine, Federico II University of Naples.</p><p>MF is a Veterinary practitioner with long experience in buffalo breeding.</p><p>EDC is the Director of Istituto Zooprofilattico Sperimentale del Mezzogiorno (IZSM) – Salerno Section and the head of the National Centre for hygiene, breeding technologies and Buffalo production.</p><p>GG is Veterinary Manager at the Istituto Zooprofilattico Sperimentale del Mezzogiorno (IZSM) – Portici Section.</p><p>GB is researcher at the Istituto Zooprofilattico Sperimentale del Mezzogiorno (IZSM) – Portici Section.</p></sec>
The panorama of animal leptospirosis in Rio de Janeiro, Brazil, regarding the seroepidemiology of the infection in tropical regions	<sec><title>Background</title><p>Leptospirosis is an important disease caused by various serovars of <italic>Leptospira</italic> sp. It can affect humans as well as domestic and wild animals; therefore, it has importance for public health, animal production, and wild species. The aim of this paper is to discuss the epidemiology of animal leptospirosis in Rio de Janeiro, Brazil, as a possible model for other tropical regions. In several studies conducted in the last 20 years, a total of 47 rats, 120 dogs, 875 cows, 695 horses, 1,343 goats, 308 sheep and 351 pigs from all regions of the state, in addition to 107 wild mammals and 73 golden-lion tamarins were tested (MAT) for anti-<italic>Leptospira</italic> antibodies.</p></sec><sec><title>Results</title><p>Seroreactivity was frequent in all studied species, confirming that the infection is endemic in Rio de Janeiro. Serogroups Icterohaemorrhagiae and Sejroe were the most prevalent in urban and rural scenarios, respectively. This paper reviews the current knowledge on animal leptospirosis in Rio de Janeiro and describes important differences between urban versus rural cycles of the infection in various species.</p></sec><sec><title>Conclusion</title><p>Identification of the prevailing serogroups and their reservoirs is essential for understanding agent-host-environment interactions under tropical conditions.</p></sec>	<contrib contrib-type="author" id="A1"><name><surname>Martins</surname><given-names>Gabriel</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>gmartins@id.uff.br</email></contrib><contrib contrib-type="author" corresp="yes" id="A2"><name><surname>Lilenbaum</surname><given-names>Walter</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>mipwalt@vm.uff.br</email></contrib>	BMC Veterinary Research	<sec><title>Background</title><p>Leptospirosis is an important disease caused by various serovars of <italic>Leptospira</italic> sp. It can affect humans as well as wild and domestic animals; therefore, it has importance for both public health and animal production [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>].</p><p>Several syndromes have been identified in animal leptospirosis. Some species, e.g. dogs and less often horses, tend to present the classic acute disease, which includes the icteric-hemorrhagic syndrome, with fever, pulmonary involvement, and renal failure. Conversely, other species, mainly ruminants, but also swine, usually present the reproductive form of the disease, characterized by abortions and premature or weak offspring, stillbirths and fetal mummification, leading to substantial losses worldwide [<xref ref-type="bibr" rid="B3">3</xref>-<xref ref-type="bibr" rid="B6">6</xref>].</p><p>This infection has been classified into two major groups. The first is determined by strains adapted to and carried by the affected host, which are less dependent on the region or environmental conditions, as topography or rainfall (e.g. serovar Hardjo in cattle or Canicola in dogs), and usually leads to subclinical infection, becoming an important source of infection for humans or other animals. The other group consists of incidental infections caused by strains carried by other domestic and free-living animals, and are more dependent on environmental factors and management practices, which results in direct or indirect contact of the animal with the urine of the reservoirs of the bacterium. That last group constantly leads to an acute and severe leptospirosis syndrome, e.g. serovar Pomona in cattle or Icterohaemorrhagiae in humans or dogs [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B8">8</xref>]. It has been suggested that this second group could be relatively more important in tropical countries than in other regions [<xref ref-type="bibr" rid="B1">1</xref>], particularly under conditions of bad hygiene [<xref ref-type="bibr" rid="B9">9</xref>]. Though distinct variations in maintenance hosts and the serogroups they carry can occur throughout the world, a basic knowledge of serogroups and their maintenance hosts is required to understand the epidemiology of leptospirosis, either human or animal, in a particular region [<xref ref-type="bibr" rid="B6">6</xref>].</p><p>Human leptospirosis is usually due to serovars that are maintained by the animal populations of a region, which spread the bacterium on the environment [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. Recent studies conducted in many tropical countries reinforce the complex epidemiological relationship between human/animal leptospirosis. Human beings are not associated as maintenance hosts of any leptospiral serovar; therefore, they consistently present incidental infection [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B12">12</xref>].</p><p>Tropical regions have many particularities that affect the occurrence of the infection, as well as its routes and disease severity [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B14">14</xref>]. In addition to geographical conditions and aspects as climate or topography, management factors and husbandry practices, as well as frequency of veterinary assistance, may affect overall seroprevalence and also the serovar distribution [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B16">16</xref>].</p><p>The aim of this paper is to discuss in a broader perspective various results obtained during 20 years in several studies regarding animal leptospirosis, in order to contribute to a better understanding of the epidemiology of animal leptospirosis in Rio de Janeiro, Brazil, as a possible model for other tropical regions.</p></sec><sec sec-type="methods"><title>Methods</title><p>Two internet databases were consulted with the keywords “Leptospirosis”, “<italic>Leptospira</italic>” for the MEDLINE (Medical Literature Analysis and Retrieval System Online), PubMed and SciELO databases. In addition, the Google search engine was used to identify studies outside of the peer–reviewed journal literature. Relevant papers from each bibliographic database and the Google search engine were systematically searched using the following search terms, or derivatives of these, depending on the subject heading terms used by the databases: livestock OR cattle OR horse OR sheep OR goats OR dogs OR “<italic>Rattus norvegicus</italic>” OR pig OR “wild mammals” AND “Rio de Janeiro”. The filters applied to the whole research were Portuguese and English language, and the search interval was January 1990 to January 2013. For all obtained results, full texts were obtained and analyzed. After analysis of the full text of the obtained manuscripts, articles that did not meet the original criteria (animal leptospirosis in Rio de Janeiro, Brazil) of the review were removed.</p><p>A total of 15 articles met the inclusion criteria and were studied. They are related to the leptospirosis in rats (<italic>Rattus norvegicus</italic>) captured from an urban area [<xref ref-type="bibr" rid="B17">17</xref>] and dogs with clinical suspect of the disease [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B18">18</xref>]. Regarding livestock, there were studies about cows [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>], horses [<xref ref-type="bibr" rid="B21">21</xref>-<xref ref-type="bibr" rid="B23">23</xref>], goats [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B24">24</xref>], sheep [<xref ref-type="bibr" rid="B4">4</xref>] and pigs [<xref ref-type="bibr" rid="B25">25</xref>]. Wild mammals kept in the local zoo were also studied [<xref ref-type="bibr" rid="B26">26</xref>], as well as golden-lion tamarins from a research center [<xref ref-type="bibr" rid="B27">27</xref>] and wild felines [<xref ref-type="bibr" rid="B28">28</xref>]. The main limitation of that meta-analysis was that few groups are dedicated to the study of animal leptospirosis in Rio de Janeiro; therefore, the majority of cited articles were conducted by the same research group, reducing the possibility of comparing results and making a broader discussion.</p></sec><sec><title>Background</title><p>For the last 20 years, our research group has been dedicated to generating new knowledge regarding the seroepidemiology of animal leptospirosis in Rio de Janeiro, Brazil. Among other species and approaches, since the 1990’s it has been studied rats (<italic>Rattus norvegicus</italic>) captured from an urban area [<xref ref-type="bibr" rid="B17">17</xref>] and dogs with clinical indications of the disease [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B18">18</xref>]. In relation to livestock, it was studied cows [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>], horses [<xref ref-type="bibr" rid="B21">21</xref>-<xref ref-type="bibr" rid="B23">23</xref>], goats [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B24">24</xref>], sheep [<xref ref-type="bibr" rid="B4">4</xref>] and pigs [<xref ref-type="bibr" rid="B25">25</xref>]. Wild mammals kept in the local zoo were also studied [<xref ref-type="bibr" rid="B26">26</xref>], as well as golden-lion tamarins from a research center [<xref ref-type="bibr" rid="B27">27</xref>] and wild felines [<xref ref-type="bibr" rid="B28">28</xref>].</p><sec><title>Sampling</title><p>With the exception of the studies on urban rats and the wild animals, all of the studies were prospective and conducted in the last 20 years. For those, sampling was calculated based on the official population of each species on Rio de Janeiro state and statistical formulae <inline-formula><mml:math id="M1" name="1746-6148-9-237-i1" overflow="scroll"><mml:mrow><mml:mo stretchy="true">(</mml:mo><mml:mi mathvariant="italic">ss</mml:mi><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:msup><mml:mrow><mml:mi>z</mml:mi></mml:mrow><mml:mrow><mml:mn>2</mml:mn></mml:mrow></mml:msup><mml:mo></mml:mo><mml:mfenced open="(" close=")"><mml:mi>p</mml:mi></mml:mfenced><mml:mo></mml:mo><mml:mfenced open="(" close=")"><mml:mrow><mml:mi>p</mml:mi><mml:mo>-</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:mfenced></mml:mrow><mml:msup><mml:mrow><mml:mi>C</mml:mi></mml:mrow><mml:mrow><mml:mn>2</mml:mn></mml:mrow></mml:msup></mml:mfrac></mml:mrow></mml:math></inline-formula>, confidence level of 95%) determining the sampling as well as their distribution was applied, to represent the population of each species in various regions of the state. In the particular case of wild animals and urban rats, the totality of animals which fit the inclusion criteria for each study was tested. Therefore, a total of 47 urban rats, 120 dogs, 875 cows, 695 horses, 1,343 goats, 308 sheep and 351 pigs from all the regions of the state were tested, in addition to 77 mammals (except felines) and 30 wild felines from Rio de Janeiro Zoo and 73 golden-lion tamarins from the Primatology Center of Rio de Janeiro. All studies were previously approved by the Ethics Committee of Universidade Federal Fluminense. Blood samples were collected from the vena jugularis (for rats, direct cardiac puncture was used) into vacuum tubes and allowed to clot at room temperature. At the laboratory, sera was separated by centrifugation and stored at -20°C to be tested as a batch.</p></sec><sec><title>Serology</title><p>The microscopic agglutination test (MAT) with live antigens was employed, as recommended [<xref ref-type="bibr" rid="B29">29</xref>] (Table <xref ref-type="table" rid="T1">1</xref>). Briefly, serum samples were screened at a 1:100 dilution and a collection including 24 serovars from all serogroups was employed as antigens, in order to determine the adequate antigen battery for each animal in that particular region.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p><bold>Twenty-four </bold><bold> <italic>Leptospira </italic> </bold><bold>strains used in microscopic agglutination test (MAT) for serological diagnosis of various animal species in Rio de Janeiro, Brazil</bold></p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Species</bold></th><th align="left"><bold>Serovar</bold></th><th align="left"><bold>Serogroup</bold></th><th align="left"><bold>Reference strain</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom"><italic>L. borgpetersenii</italic><hr/></td><td align="left" valign="bottom">Ballum<hr/></td><td align="left" valign="bottom">Ballum<hr/></td><td align="left" valign="bottom">Mus 127<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. borgpetersenii</italic><hr/></td><td align="left" valign="bottom">Castellonis<hr/></td><td align="left" valign="bottom">Ballum<hr/></td><td align="left" valign="bottom">Castellon 3<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. borgpetersenii</italic><hr/></td><td align="left" valign="bottom">Javanica<hr/></td><td align="left" valign="bottom">Javanica<hr/></td><td align="left" valign="bottom">Veldrat Batavia 46<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. borgpetersenii</italic><hr/></td><td align="left" valign="bottom">Mini<hr/></td><td align="left" valign="bottom">Mini<hr/></td><td align="left" valign="bottom">Sari<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. borgpetersenii</italic><hr/></td><td align="left" valign="bottom">Tarassovi<hr/></td><td align="left" valign="bottom">Tarassovi<hr/></td><td align="left" valign="bottom">Perepelicin<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. borgpetersenii</italic><hr/></td><td align="left" valign="bottom">Whitcombi<hr/></td><td align="left" valign="bottom">Celledoni<hr/></td><td align="left" valign="bottom">Whitcomb<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. interrogans</italic><hr/></td><td align="left" valign="bottom">Australis<hr/></td><td align="left" valign="bottom">Australis<hr/></td><td align="left" valign="bottom">Ballico<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. interrogans</italic><hr/></td><td align="left" valign="bottom">Autumnalis<hr/></td><td align="left" valign="bottom">Autumnalis<hr/></td><td align="left" valign="bottom">Akiyami A<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. interrogans</italic><hr/></td><td align="left" valign="bottom">Bataviae<hr/></td><td align="left" valign="bottom">Bataviae<hr/></td><td align="left" valign="bottom">Van Tienen<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. interrogans</italic><hr/></td><td align="left" valign="bottom">Bratislava<hr/></td><td align="left" valign="bottom">Australis<hr/></td><td align="left" valign="bottom">Jez Bratislava<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. interrogans</italic><hr/></td><td align="left" valign="bottom">Canicola<hr/></td><td align="left" valign="bottom">Canicola<hr/></td><td align="left" valign="bottom">Hond Utrecht IV<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. interrogans</italic><hr/></td><td align="left" valign="bottom">Copenhageni<hr/></td><td align="left" valign="bottom">Icterohaemorrhagiae<hr/></td><td align="left" valign="bottom">M 20<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. interrogans</italic><hr/></td><td align="left" valign="bottom">Hardjo<hr/></td><td align="left" valign="bottom">Sejroe<hr/></td><td align="left" valign="bottom">Hardjoprajitno<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. interrogans</italic><hr/></td><td align="left" valign="bottom">Hebdomadis<hr/></td><td align="left" valign="bottom">Hebdomadis<hr/></td><td align="left" valign="bottom">Hebdomadis<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. interrogans</italic><hr/></td><td align="left" valign="bottom">Icterohaemorrhagiae<hr/></td><td align="left" valign="bottom">Icterohaemorrhagiae<hr/></td><td align="left" valign="bottom">RGA<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. interrogans</italic><hr/></td><td align="left" valign="bottom">Pomona<hr/></td><td align="left" valign="bottom">Pomona<hr/></td><td align="left" valign="bottom">Pomona<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. interrogans</italic><hr/></td><td align="left" valign="bottom">Pyrogenes<hr/></td><td align="left" valign="bottom">Pyrogenes<hr/></td><td align="left" valign="bottom">Salinem<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. interrogans</italic><hr/></td><td align="left" valign="bottom">Wolffi<hr/></td><td align="left" valign="bottom">Sejroe<hr/></td><td align="left" valign="bottom">3705<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. kirschneri</italic><hr/></td><td align="left" valign="bottom">Butembo<hr/></td><td align="left" valign="bottom">Autumnalis<hr/></td><td align="left" valign="bottom">Butembo<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. kirschneri</italic><hr/></td><td align="left" valign="bottom">Cynopteri<hr/></td><td align="left" valign="bottom">Cynopteri<hr/></td><td align="left" valign="bottom">3522 C<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. kirschneri</italic><hr/></td><td align="left" valign="bottom">Grippotyphosa<hr/></td><td align="left" valign="bottom">Grippotyphosa<hr/></td><td align="left" valign="bottom">Moskva V<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. noguchii</italic><hr/></td><td align="left" valign="bottom">Panama<hr/></td><td align="left" valign="bottom">Panama<hr/></td><td align="left" valign="bottom">CZ 214 K<hr/></td></tr><tr><td align="left" valign="bottom"><italic>L. santarosai</italic><hr/></td><td align="left" valign="bottom">Guaricurus<hr/></td><td align="left" valign="bottom">Sejroe<hr/></td><td align="left" valign="bottom">Bov. G<hr/></td></tr><tr><td align="left"><italic>L. santarosai</italic></td><td align="left">Shermani</td><td align="left">Shermani</td><td align="left">1342 K</td></tr></tbody></table></table-wrap><p>All strains were grown in liquid medium (EMJH for 7–10 days at 28-30°C), free of contamination or auto-agglutination. All samples with agglutinating activity at a 1:100 dilution were considered positive and subsequently titrated against reacting antigens, using serial two-fold dilutions of serum. The endpoint was the highest tube in which 50% agglutination was recorded, leaving 50% free cells compared to a control culture diluted 1/2 in phosphate buffered saline. The highest titre reached was used to suggest the infective serogroup. Since the reliability of MAT in discriminating among serovars of the same serogroup has been reported [<xref ref-type="bibr" rid="B30">30</xref>], in the present study we refer to the serological results by serogroup.</p></sec></sec><sec sec-type="results"><title>Results</title><p>In all species studied, seroreactivity to leptospirosis was very frequent (at varying levels). Therefore, we inferred that the infection is endemic and widely distributed in Rio de Janeiro, Brazil. The seroprevalence rates and the predominant serogroups for each species are shown (Table <xref ref-type="table" rid="T2">2</xref>).</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p><bold>Anti-</bold><bold> <italic>Leptospira </italic> </bold><bold>sp. antibodies prevalence rates and predominant serogroups for various animal species studied in Rio de Janeiro, Brazil</bold></p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Specie</bold></th><th align="left"><bold>N</bold></th><th align="left"><bold>Seroprevalence (%)</bold></th><th align="left"><bold>Predominant serogroup(s)</bold></th><th align="left"><bold> <italic>Reference</italic> </bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Rats (<italic>R. norvegicus</italic>)<hr/></td><td align="left" valign="bottom">47<hr/></td><td align="left" valign="bottom">17 (36.2)<hr/></td><td align="left" valign="bottom">Icterohaemorrhagiae<hr/></td><td align="left" valign="bottom">[<xref ref-type="bibr" rid="B17">17</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Dogs<hr/></td><td align="left" valign="bottom">120<hr/></td><td align="left" valign="bottom">88 (73.3)<hr/></td><td align="left" valign="bottom">Icterohaemorrhagiae<hr/></td><td align="left" valign="bottom">[<xref ref-type="bibr" rid="B18">18</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Cattle<hr/></td><td align="left" valign="bottom">875<hr/></td><td align="left" valign="bottom">335 (38.3)<hr/></td><td align="left" valign="bottom">Sejroe<hr/></td><td align="left" valign="bottom">[<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Horses<hr/></td><td align="left" valign="bottom">695<hr/></td><td align="left" valign="bottom">275 (39.6)<hr/></td><td align="left" valign="bottom">Icterohaemorrhagiae/Australis<hr/></td><td align="left" valign="bottom">[<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B23">23</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Goats<hr/></td><td align="left" valign="bottom">1,343<hr/></td><td align="left" valign="bottom">200 (14.9)<hr/></td><td align="left" valign="bottom">Sejroe<hr/></td><td align="left" valign="bottom">[<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B24">24</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Sheep<hr/></td><td align="left" valign="bottom">308<hr/></td><td align="left" valign="bottom">146 (47.4)<hr/></td><td align="left" valign="bottom">Sejroe<hr/></td><td align="left" valign="bottom">[<xref ref-type="bibr" rid="B4">4</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Pigs<hr/></td><td align="left" valign="bottom">351<hr/></td><td align="left" valign="bottom">232 (66.1)<hr/></td><td align="left" valign="bottom">Icterohaemorrhagiae<hr/></td><td align="left" valign="bottom">[<xref ref-type="bibr" rid="B25">25</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Wild mammals (except felines)<hr/></td><td align="left" valign="bottom">77<hr/></td><td align="left" valign="bottom">29 (37.7)<hr/></td><td align="left" valign="bottom">Icterohaemorrhagiae<hr/></td><td align="left" valign="bottom">[<xref ref-type="bibr" rid="B26">26</xref>]<hr/></td></tr><tr><td align="left" valign="bottom">Wild felines<hr/></td><td align="left" valign="bottom">30<hr/></td><td align="left" valign="bottom">4 (13.3)<hr/></td><td align="left" valign="bottom">Icterohaemorrhagiae/Pomona<hr/></td><td align="left" valign="bottom">[<xref ref-type="bibr" rid="B28">28</xref>]<hr/></td></tr><tr><td align="left">Golden-lion tamarins</td><td align="left">73</td><td align="left">11 (15.1)</td><td align="left">Icterohaemorrhagiae</td><td align="left">[<xref ref-type="bibr" rid="B27">27</xref>]</td></tr></tbody></table><table-wrap-foot><p>Average of the reported prevalence.</p></table-wrap-foot></table-wrap><p>The highest prevalence of seroreactivity among all studied species was for dogs (73.3%), followed by pigs (66.1%), both of them predominantly reactive to the serogroup Icterohaemorrhagiae. Horses had 27 and 42.9% (average 39.6%) seroreactivity, mainly to serogroups Australis and Icterohaemorrhagiae.</p><p>In ruminants, seroreactivity was highest in sheep (47.4%), followed by cows (23% 36.9 and 46.9%; average 38.3%) and goats (25.9 and 11.1%; average 14.9%). Most seropositive ruminants had reactions against the serogroup Sejroe, which contains the serovar Hardjo, usually reported as the most frequent leptospirosis agent for ruminants worldwide. Wild animals also had a high level of seroreactivity. In regards to the wild mammals from the zoo, antibodies were particularly more common in Canidae, Myrmecophagidae and Procyonidae, with 7 of 9, 5 of 9, and 5 of 9 being positive, respectively. The average of seroreactivity of wild mammals other than felines in the zoo was 37.7%, whereas for wild felines (genus <italic>Herpailurus, Leopardus, Panthera</italic> and <italic>Puma</italic>) it was 13.3%, with the majority of the reactions directed against the serogroups Pomona and Icterohaemorrhagiae. Despite the lower level of seroreactivity in golden-lion tamarins (15.0%), serogroup Icterohaemorrhagiae was the most common agent in these animals. Finally, in urban rats there was 36.2% of seroreactivity, predominantly against the serogroup Icterohaemorrhagiae.</p></sec><sec sec-type="discussion"><title>Discussion</title><p>Although serology has some limitations, not only on sensitivity but also specificity and cross reactions, it is still the most widely used tool for large-scale epidemiological studies [<xref ref-type="bibr" rid="B6">6</xref>]. Despite inherent limitations of serological testing, we inferred that leptospirosis is endemic and widespread in Rio de Janeiro, since all tested animal species had high seroreactivity. It was not an unexpected result, since studies from other regions of Brazil [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B31">31</xref>-<xref ref-type="bibr" rid="B33">33</xref>] as well as from other tropical areas reported leptospirosis as a widespread infection [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B11">11</xref>]. Despite this, animal leptospirosis is often neglected and many practitioners, even from endemic regions, are only familiar with the acute syndromes of the disease. Therefore, mainly in livestock, the silent reproductive syndrome is often not investigated, leading to failure to identify the disease. Furthermore, it is noteworthy that tropical regions offer excellent conditions for the survival and spread of leptospires, due to the climate and particularly the rainfall. Those conditions are often associated with poor management practices (rural areas) or poor sanitary conditions (urban cycle) and have been clearly demonstrated to be strongly associated with occurrence of the infection [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B12">12</xref>].</p><p>After the serogroup distribution was determined for each species, it was apparent that a reduced number of antigens would have detected more than 98% of the seroreactive samples from animals. Therefore, a simplified antigen battery including representants of the more frequent serogroups that occur in Rio de Janeiro for each species was recommended and employed at the laboratory, except for wild animals (Table <xref ref-type="table" rid="T3">3</xref>). Since it is well known that human leptospirosis is usually a reflection of serovars that are maintained by the animal population of a region [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B10">10</xref>], it is highly probable that also human leptospirosis can be reliably detected using that simplified antigen battery.</p><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>Recommended antigen battery (serogroups) for microscopic agglutination test for various animal species in Rio de Janeiro, Brazil</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Ruminants</bold></th><th align="left"><bold>Horses</bold></th><th align="left"><bold>Dogs</bold></th><th align="left"><bold>Pigs</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Australis<hr/></td><td align="left" valign="bottom">Australis<hr/></td><td align="left" valign="bottom">Canicola<hr/></td><td align="left" valign="bottom">Australis<hr/></td></tr><tr><td align="left" valign="bottom">Grippotyphosa<hr/></td><td align="left" valign="bottom">Canicola<hr/></td><td align="left" valign="bottom">Icterohaemorrhagiae<hr/></td><td align="left" valign="bottom">Icterohaemorrhagiae<hr/></td></tr><tr><td align="left" valign="bottom">Pomona<hr/></td><td align="left" valign="bottom">Icterohaemorrhagiae<hr/></td><td align="left" valign="bottom">Grippotyphosa<hr/></td><td align="left" valign="bottom">Pomona<hr/></td></tr><tr><td align="left" valign="bottom">Sejroe<hr/></td><td align="left" valign="bottom">Pomona<hr/></td><td align="left" valign="bottom">Pomona<hr/></td><td align="left" valign="bottom">Sejroe<hr/></td></tr><tr><td align="left">-</td><td align="left">Sejroe</td><td align="left">-</td><td align="left">Tarassovi</td></tr></tbody></table></table-wrap><p>Cut-off values varied between epidemiological studies (sampling animals randomly) and diagnostic investigations (testing animals or herds with clinical/reproductive symptoms). Overall, we concluded that, when paired serology is not available (which is particularly common for animal leptospirosis), a single sample reaction with titres ≥400 may be considered as acute leptospirosis [<xref ref-type="bibr" rid="B30">30</xref>]. Chronic and host-adapted infection can be accompanied by low titers, whereas acute infections are usually associated with high titers. In addition, the extent to which an infection is considered endemic in a specific region may also be considered for determining an appropriate cut-off point [<xref ref-type="bibr" rid="B34">34</xref>]. In that regard in tropical and endemic areas, many groups consider only reactions ≥800 for the diagnosis of acute leptospirosis using one single sample [<xref ref-type="bibr" rid="B35">35</xref>]. Our group has investigated the association of titres and clinical signs, together with complementary exams (hematology and biochemistry) and reproductive signs (in livestock). Many laboratories consider 100 as the recommended cut-off titre [<xref ref-type="bibr" rid="B29">29</xref>], and more recently, a cut-off of 160 has been suggested for MAT [<xref ref-type="bibr" rid="B30">30</xref>]. Nevertheless, that last cut-off (160) was related to laboratory case definition in humans, which are incidental hosts. Therefore, this suggestion cannot be simply extrapolated to animals, due to the chronic cases and to host-adapted infections. Additionally, it is known that infection by host-adapted serovars usually leads to low titres [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B32">32</xref>]. Therefore, considering that the disease is endemic in tropical areas, we suggest that, for animal leptospirosis in the tropics, a cut-off of 200 should be considered, although broader studies are required to increase the specificity and consequently the reliability of the diagnosis.</p><p>Besides observing that leptospirosis is widely distributed among all the studied species, it is possible to discriminate some epidemiological patterns that can be useful for the understanding of the agent-host-environment interactions under tropical conditions. Basically, there are two major conditions, with two cycles each: the urban/rural condition and the host-adapted/incidental cycle, which may or not be coincident.</p><p>The urban cycle is clearly dominated by members of the serogroup Icterohaemorrhagiae, which causes acute clinical disease. Members of this serogroup are maintained by rodents, mainly the Norwegian rat (<italic>Rattus norvegicus</italic>) [<xref ref-type="bibr" rid="B36">36</xref>]. They are extremely virulent [<xref ref-type="bibr" rid="B37">37</xref>] and cause severe acute illness in several species, including humans [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B13">13</xref>]. Rats captured in Rio de Janeiro were highly infected [<xref ref-type="bibr" rid="B17">17</xref>], consistent with findings from other regions [<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B39">39</xref>]. In addition to environmental conditions, this certainly contributes to the predominance of this serogroup in all the studied species in the urban cycle. The impact of this serogroup is so important in urban regions that, even with animal species that have their own host-adapted strains, as urban dogs (reservoirs of Canicola), seroreactivity against Icterohaemorrhagiae is much more common than that directed against Canicola, as should be expected. This outcome was not surprising, since urban dogs are constantly exposed to Icterohaemorrhagiae [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. In this scenario, dogs are susceptible to the same conditions as humans (flood risk areas and waste accumulation sites), which contribute to the presence of the agent in the environment, and therefore increase the zoonotic risk [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B40">40</xref>]. Serogroup distribution among wild mammals at the zoo was of particular interest. With the exception of wild felines, which are considered resistant to Icterohaemorrhagiae strains [<xref ref-type="bibr" rid="B28">28</xref>], several species, from different origins and with a great diversity of habits, the predominant seroreactivity was once more against Icterohaemorrhagiae [<xref ref-type="bibr" rid="B26">26</xref>]. Therefore, we hypothesize that, in the urban cycle of leptospirosis, environmental conditions that favor infection by Icterohaemorrhagiae are very important and have a greater impact on the epidemiology of the disease than differences among host species.</p><p>This scenario was not only applicable to Rio de Janeiro; it was recently reported that in the urban area of Salvador, Brazil [<xref ref-type="bibr" rid="B41">41</xref>], as well as in other tropical islands [<xref ref-type="bibr" rid="B38">38</xref>], where marmosets are predominantly seroreactive against that serogroup. Therefore, similar associations are expected in many tropical regions of the world. Indeed, dogs from a town in Amazon rain forest [<xref ref-type="bibr" rid="B42">42</xref>], as well as captive animals in Peruvian Amazon [<xref ref-type="bibr" rid="B43">43</xref>] were seroreactive against Icterohaemorrhagiae, and not to serogroups maintained by wild animals, as might be expected.</p><p>In relation to the rural cycle of the infection, differences between infections caused by incidental or adapted serogroups are clearer. Among livestock, the most studied species are ruminants (cattle, sheep and goats). In those species, serogroup Sejroe was highly predominant, consistent with studies conducted in other tropical [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B44">44</xref>] and even non-tropical areas [<xref ref-type="bibr" rid="B45">45</xref>,<xref ref-type="bibr" rid="B46">46</xref>], as well as in other regions of Brazil [<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B47">47</xref>,<xref ref-type="bibr" rid="B48">48</xref>]. Members of this serogroup, mainly Hardjo, are known to be adapted to ruminants [<xref ref-type="bibr" rid="B16">16</xref>] and its maintenance within a herd may also occur by animal-animal direct transmission. Although infection by Hardjo in cattle has been reported as less dependent on environmental conditions [<xref ref-type="bibr" rid="B44">44</xref>], it may not be absolutely valid for the tropics [<xref ref-type="bibr" rid="B19">19</xref>]. Therefore, we have proposed that under tropical conditions, a successful leptospirosis control program, in addition to vaccinations and antimicrobial treatment, should also include an investigation of environmental and herd management practices in order to identify factors likely to affect transmission and prevalence of the disease [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B19">19</xref>].</p><p>Incidental infection in ruminants are infrequent and usually leads to outbreaks and severe clinical illness, whereas infections by adapted strains are often chronic, with only mild symptoms, particularly with regards to reproductive function [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B49">49</xref>]. Pomona, the most commonly reported serogroup for incidental infections in cattle from 1970–1980, is currently only rarely detected [<xref ref-type="bibr" rid="B44">44</xref>]. We infer that this is due to improved husbandry practices, mainly the reduction of the habit of co-grazing cattle with other species, particularly pigs. In that regard, pigs act as reservoirs of that serogroup and their presence on cattle farms has been strongly associated with the occurrence of incidental leptospirosis [<xref ref-type="bibr" rid="B9">9</xref>].</p><p>Therefore, in contrast to in the urban cycle, where incidental acute infection by Icterohaemorrhagiae is predominant, in the rural cycle, chronic and mild infections determined by adapted strains, mainly from serogroup Sejroe, seem to be more common [<xref ref-type="bibr" rid="B20">20</xref>]. This fact may explain the misdiagnosis of the infection among practitioners that are not familiarized with these details, with significant impacts on animal production as well on public health. The study of animal leptospirosis is directly related to the understanding and prevention of the infection in humans, mainly in tropics [<xref ref-type="bibr" rid="B38">38</xref>]. Nevertheless, in the urban scenario, human leptospirosis is due to an interaction among the presence of reservoir (mainly urban rodents) of leptospires (e.g. Icterohaemorrhagiae serogroup members), a highly contaminated environment, with high rates of exposure to the agent (e.g. household clustering), as observed in slum communities. In contrast, in rural areas, serovars that most commonly infect livestock (e.g. Hardjo), only rarely cause human infections [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B15">15</xref>]. Furthermore, the role of rodents in the rural cycle has been neglected and more studies are needed.</p></sec><sec sec-type="conclusions"><title>Conclusion</title><p>In conclusion, this paper discusses the current knowledge on animal leptospirosis in Rio de Janeiro and demonstrates important differences between urban and rural cycles of the infection, depicting the serogroup distribution in both cycles and in various species. Identification of the prevailing serogroups and of their animal reservoirs is essential for understanding agent-host-environment interactions under tropical conditions.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>GM participated in the design, conducted the analysis and drafted the manuscript. WL carried out the coordination and critical and final review of the manuscript. Both authors read and approved the final manuscript.</p></sec>
Describing antimicrobial use and reported treatment efficacy in Ontario swine using the Ontario swine veterinary-based Surveillance program	<sec><title>Background</title><p>The objective of this work was to retrospectively assess records received through the Ontario Swine Veterinary-based Surveillance program July 2007 – July 2009 to describe and assess relationships between reported treatment failure, antimicrobial use, diagnosis and body system affected.</p></sec><sec><title>Results</title><p>Antimicrobial use occurred in 676 records, 80.4% of all records recording treatment (840). The most commonly used antimicrobials were penicillin (34.9%), tetracyclines (10.7%) and ceftiofur (7.8%), and the use of multiple antimicrobials occurred in 141/676 records (20.9%). A multi-level logistic regression model was built to describe the probability of reported treatment failure. The odds of reported treatment failure were significantly reduced if the record indicated that the gastro-intestinal (GI) system was affected, as compared to all other body systems (p < 0.05). In contrast, the odds of reported treatment failure increased by 1.98 times if two antimicrobials were used as compared to one antimicrobial (p = 0.009) and by 6.52 times if three or more antimicrobials were used as compared to one antimicrobial (p = 0.005). No significant increase in reported treatment failure was seen between the use of two antimicrobials and three or more antimicrobials. No other antimicrobials were significantly associated with reported treatment failure after controlling for body system and the number of antimicrobials used.</p></sec><sec><title>Conclusions</title><p>Failure of antimicrobial treatment is more likely to occur in non-GI conditions, as compared to GI conditions and the use of multiple antimicrobial products is also associated with an increased probability of antimicrobial treatment failure. The authors suggest that a more preventative approach to herd health should be taken in order to reduce antimicrobial inputs on-farm, including improved immunity via vaccination, management and biosecurity strategies. Furthermore, improved immunity may be viewed as a form of antimicrobial stewardship to the industry by reducing required antimicrobial inputs and consequently, reduced selection pressure for AMR.</p></sec>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Glass-Kaastra</surname><given-names>Shiona K</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>sglass@uoguelph.ca</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Pearl</surname><given-names>David L</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>dpearl@uoguelph.ca</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Reid-Smith</surname><given-names>Richard J</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><xref ref-type="aff" rid="I3">3</xref><email>richard.reid-smith@phac-aspc.gc.ca</email></contrib><contrib contrib-type="author" id="A4"><name><surname>McEwen</surname><given-names>Beverly</given-names></name><xref ref-type="aff" rid="I3">3</xref><xref ref-type="aff" rid="I4">4</xref><email>bmcewen@uoguelph.ca</email></contrib><contrib contrib-type="author" id="A5"><name><surname>McEwen</surname><given-names>Scott A</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>smcewen@uoguelph.ca</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Amezcua</surname><given-names>Rocio</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>mamezcua@uoguelph.ca</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Friendship</surname><given-names>Robert M</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>rfriends@uoguelph.ca</email></contrib>	BMC Veterinary Research	<sec><title>Background</title><p>Antimicrobial resistance (AMR) threatens the efficacy of antimicrobial drugs for treating infections in humans and animals alike. Antimicrobial resistance emerges in the presence of antimicrobial products and the transfer of resistance genes among bacteria may occur [<xref ref-type="bibr" rid="B1">1</xref>]. As clinical infection with resistant pathogens may lead to prolonged morbidity, increased costs and increased risk of mortality, AMR is a serious concern for food-animal production. Recently, it has been reported that multiple-class resistance within clinical pathogens of swine in Ontario is relatively common [<xref ref-type="bibr" rid="B2">2</xref>]. However, it is not known if this has led to increased antimicrobial treatment as there is currently no accurate measurement of the volume of antimicrobial use in Ontario swine. Furthermore, these resistance data do not include information regarding the treatment provided, or efficacy of the treatment. Given that in-vitro resistance does not necessarily predict failure of antimicrobial treatment in-vivo [<xref ref-type="bibr" rid="B3">3</xref>], industry stakeholders may also benefit from the knowledge of significant predictors for reporting treatment failure.</p><p>In July 2007, the Ontario Swine Veterinary-based Surveillance (OSVS) program was initiated to assess the potential for developing a practitioner-based syndromic health surveillance system for swine in Ontario [<xref ref-type="bibr" rid="B4">4</xref>]. Recruited practitioners were asked to record and report on all farm visits or calls related to swine. These reports requested information regarding suspected diagnosis, the body system affected, treatment(s) provided, and the efficacy of treatments. Consequently, these records may provide a valuable source of antimicrobial use and in-vivo efficacy data for Ontario; a complement to the available in-vitro resistance data from the Animal Health Laboratory [<xref ref-type="bibr" rid="B5">5</xref>]. Furthermore, these data allow for the examination of associations between treatment with certain antimicrobials and reported treatment failure. Therefore, the objective of this work was to retrospectively assess records received through the OSVS program from July 1, 2007 through June 30, 2009 to describe and assess relationships between reported treatment failure, antimicrobial use, diagnosis, and body system affected.</p></sec><sec sec-type="methods"><title>Methods</title><p>A full description of the OSVS pilot project is available elsewhere [<xref ref-type="bibr" rid="B4">4</xref>]. Briefly, the OSVS pilot program was funded by the Ontario Ministry of Agriculture Food & Rural Affairs (OMAFRA) and the Ontario Animal Health Strategic Investment (AHSI) fund. During the July 1 2007 to June 30 2009 time period, reports were received from up to ten practitioners representing five clinics known to service most of the swine industry in Ontario. During this period, practitioners recorded data for all daily swine-related farm visits and calls using either a paper or electronic submission via personal digital assistants (PDA) or an internet-based form. Data collected included whether the visit/call was a disease or routine visit, unique practitioner ID, unique farm ID, signs/symptoms displayed, diagnosis, body system affected, whether it was an incident or ongoing condition, farm history of the condition, diagnosis, veterinarian-prescribed treatment, and efficacy of treatment. As records were made at the visit level, a record could reflect treatment and efficacy at the individual, pen, or herd level.</p><p>A database was created through the electronic form submissions and manual input of the paper forms, using Microsoft Access (2003). Data cleaning, tabulations and multi-level logistic regression modeling were performed in Stata/MP 12.1 (Stata Corporation, College Station, Texas, USA). Manual mining of free-text fields was performed in order to determine the most commonly used antimicrobials and common diagnoses. Treatment failure was deemed to have occurred if practitioners recorded a treatment as being not efficacious, or “occasionally” efficacious. Due to small numbers of observations, the nervous, integument and reproductive body systems were combined into a single 'other’ category. The treatment variable was searched to create binary variables for each antimicrobial used and a variable describing the number of antimicrobials used was developed by adding across these binary variables. Therefore, multiple antimicrobial treatment was defined as any record with >1 antimicrobial listed within the treatment text field. Multiple antimicrobial treatments may not have been initiated at the time of the record, but were either 1) in use concurrently, or 2) recently used to treat/control the specific condition in the animals being seen at the time of the visit. As only a single record was found with more than 3 antimicrobials used, a “3 or more” category was created in the number of antimicrobials variable.</p><p>Multilevel logistic regression models were built to describe the probability of treatment failure, given antimicrobial treatment. Two- and three-level models were built using practitioner and farm as random intercepts, and the inclusion of a random slope for practitioner was also tested. Fixed effect predictors examined were diagnosis, body system affected, the number of antimicrobials used and each of the individual antimicrobial use variables. A manual backwards-selection process was used to build the model; all predictors were added to the model initially and removed one at a time based on the highest p-value. Categorical variables with > 2 levels were assessed for significance using the likelihood ratio test [<xref ref-type="bibr" rid="B6">6</xref>]. As predictors were removed, their impact on all other statistically significant coefficients was assessed to ensure that confounding variables remained in the model; a 30% change in any significant coefficient resulted in the removed variable being replaced in the model [<xref ref-type="bibr" rid="B6">6</xref>]. All two-way interaction terms were generated between all significant main-effects and tested within the model at p < 0.05. Where more than one model fit the data, the model with the most negative Akaike information criteria was chosen [<xref ref-type="bibr" rid="B6">6</xref>]. At the record level, Pearson and deviance residuals were visualized and any covariate patterns showing anomalous values were recorded. The normality of the best linear unbiased predictors (BLUPS) was assessed visually with normal quantile plots. Models were re-run with the exclusion of noted covariate patterns to assess any dramatic changes in the coefficients. Contrast statements were used to make comparisons between dummy variable categories within the body system and number of antimicrobials variables and the latent variable technique was employed to calculate variance components at each level [<xref ref-type="bibr" rid="B7">7</xref>].</p></sec><sec sec-type="results"><title>Results</title><p>In total, 3691 records were received by the OSVS program from July 2007 – July 2009. Antimicrobial use was reported in 676 of these records. These reflected reports from nine practitioners, submitting 7 to 255 records each and reflected data from 335 farms with 1 to 14 records each. A number of records indicated that an antimicrobial was used, without naming the product (182/676). When drug names were included, the most commonly used antimicrobials were penicillin (34.9%), tetracyclines (10.7%) and ceftiofur (7.8%) (Table <xref ref-type="table" rid="T1">1</xref>). Use of multiple antimicrobials in a single record was not uncommon; 141 (20.6%) records indicated that treatment included ≥ 2 antimicrobials (Table <xref ref-type="table" rid="T2">2</xref>). The most common combinations of antimicrobials used for treatment were penicillin with tetracyclines (24 records), penicillin with ceftiofur (16 records) and penicillin with a sulfonamide product (13 records) (Table <xref ref-type="table" rid="T3">3</xref>).</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Number of OSVS records where treatment with each antimicrobial was reported (July 1, 2007 – June 30, 2009)</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Antimicrobial</bold></th><th align="left"><bold>Frequency</bold></th><th align="left"><bold>Percent of records with antimicrobial treatment (N = 676)</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Penicillin<hr/></td><td align="left" valign="bottom">236<hr/></td><td align="left" valign="bottom">34.91<hr/></td></tr><tr><td align="left" valign="bottom">Specific drug not recorded<hr/></td><td align="left" valign="bottom">182<hr/></td><td align="left" valign="bottom">26.92<hr/></td></tr><tr><td align="left" valign="bottom">Tetracyclines<hr/></td><td align="left" valign="bottom">72<hr/></td><td align="left" valign="bottom">10.65<hr/></td></tr><tr><td align="left" valign="bottom">Ceftiofur<hr/></td><td align="left" valign="bottom">53<hr/></td><td align="left" valign="bottom">7.84<hr/></td></tr><tr><td align="left" valign="bottom">Lincomycin<hr/></td><td align="left" valign="bottom">41<hr/></td><td align="left" valign="bottom">6.07<hr/></td></tr><tr><td align="left" valign="bottom">Neomycin<hr/></td><td align="left" valign="bottom">41<hr/></td><td align="left" valign="bottom">6.07<hr/></td></tr><tr><td align="left" valign="bottom">Tylosin<hr/></td><td align="left" valign="bottom">36<hr/></td><td align="left" valign="bottom">5.33<hr/></td></tr><tr><td align="left" valign="bottom">Sulfonamides<hr/></td><td align="left" valign="bottom">28<hr/></td><td align="left" valign="bottom">4.14<hr/></td></tr><tr><td align="left" valign="bottom">Tulathromycin<hr/></td><td align="left" valign="bottom">26<hr/></td><td align="left" valign="bottom">3.85<hr/></td></tr><tr><td align="left" valign="bottom">Amoxicillin<hr/></td><td align="left" valign="bottom">22<hr/></td><td align="left" valign="bottom">3.25<hr/></td></tr><tr><td align="left" valign="bottom">Florfenicol<hr/></td><td align="left" valign="bottom">20<hr/></td><td align="left" valign="bottom">2.96<hr/></td></tr><tr><td align="left" valign="bottom">Trimethoprim-sulfamethoxazole<hr/></td><td align="left" valign="bottom">19<hr/></td><td align="left" valign="bottom">2.81<hr/></td></tr><tr><td align="left" valign="bottom">Tilmicosin<hr/></td><td align="left" valign="bottom">18<hr/></td><td align="left" valign="bottom">2.66<hr/></td></tr><tr><td align="left" valign="bottom">Gentamicin<hr/></td><td align="left" valign="bottom">14<hr/></td><td align="left" valign="bottom">2.07<hr/></td></tr><tr><td align="left" valign="bottom">Oxytetracycline<hr/></td><td align="left" valign="bottom">11<hr/></td><td align="left" valign="bottom">1.63<hr/></td></tr><tr><td align="left" valign="bottom">Apramycin<hr/></td><td align="left" valign="bottom">10<hr/></td><td align="left" valign="bottom">1.48<hr/></td></tr><tr><td align="left" valign="bottom">Trimethoprim<hr/></td><td align="left" valign="bottom">8<hr/></td><td align="left" valign="bottom">1.18<hr/></td></tr><tr><td align="left" valign="bottom">Spectinomycin<hr/></td><td align="left" valign="bottom">3<hr/></td><td align="left" valign="bottom">0.44<hr/></td></tr><tr><td align="left" valign="bottom">Bacitracin<hr/></td><td align="left" valign="bottom">3<hr/></td><td align="left" valign="bottom">0.44<hr/></td></tr><tr><td align="left" valign="bottom">Tiamulin<hr/></td><td align="left" valign="bottom">3<hr/></td><td align="left" valign="bottom">0.44<hr/></td></tr><tr><td align="left">Virginiamycin</td><td align="left">1</td><td align="left">0.15</td></tr></tbody></table></table-wrap><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Number of OSVS records where treatment efficacy or failure was reported by the number of antimicrobials used for treatment (July 1, 2007 – June 30, 2009)</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Number of antimicrobials used</bold></th><th align="left"><bold>1</bold></th><th align="left"><bold>2</bold></th><th align="left"><bold>3</bold></th><th align="left"><bold>4</bold></th><th align="left"><bold>Total</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Treatment failure<hr/></td><td align="left" valign="bottom">360<hr/></td><td align="left" valign="bottom">86<hr/></td><td align="left" valign="bottom">26<hr/></td><td align="left" valign="bottom">1<hr/></td><td align="left" valign="bottom">473<hr/></td></tr><tr><td align="left" valign="bottom">Efficacious treatment<hr/></td><td align="left" valign="bottom">175<hr/></td><td align="left" valign="bottom">26<hr/></td><td align="left" valign="bottom">2<hr/></td><td align="left" valign="bottom">0<hr/></td><td align="left" valign="bottom">203<hr/></td></tr><tr><td align="left">Total</td><td align="left">535</td><td align="left">112</td><td align="left">28</td><td align="left">1</td><td align="left"> </td></tr></tbody></table></table-wrap><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>Most frequent antimicrobial pairings as reported within OSVS records (July 1, 2007 – June 30, 2009)</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Combination</bold></th><th align="left"><bold>Number of records (N = 676)</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Penicillin + tetracycline<hr/></td><td align="left" valign="bottom">24<hr/></td></tr><tr><td align="left" valign="bottom">Penicillin + ceftiofur<hr/></td><td align="left" valign="bottom">16<hr/></td></tr><tr><td align="left" valign="bottom">Penicillin + sulfonamide<hr/></td><td align="left" valign="bottom">13<hr/></td></tr><tr><td align="left" valign="bottom">Sulfonamide + tetracycline<hr/></td><td align="left" valign="bottom">9<hr/></td></tr><tr><td align="left" valign="bottom">Neomycin + tetracycline<hr/></td><td align="left" valign="bottom">8<hr/></td></tr><tr><td align="left" valign="bottom">Tulathromycin + ceftiofur<hr/></td><td align="left" valign="bottom">8<hr/></td></tr><tr><td align="left" valign="bottom">Sulfonamide + trimethoprim<hr/></td><td align="left" valign="bottom">8<hr/></td></tr><tr><td align="left">Penicillin + sulfonamide + tetracycline</td><td align="left">7</td></tr></tbody></table></table-wrap><p>In records with antimicrobial treatments recorded, the recorded body systems affected were: respiratory, GI, musculoskeletal, multisystemic, or other. More than 27% (185/676) of records with antimicrobial use indicated that multiple systems were affected. Furthermore, 78.9% (146/185) of these records indicated treatment failure. The second and third most commonly noted body systems affected were respiratory and GI. Treatment failure was reported in 74.3 and 52.9% of these records, respectively. The most commonly noted diagnoses were <italic>Streptococcus suis</italic> infection (94 records, 13.9%) and porcine reproductive and respiratory syndrome (PRRS) (93; 13.8%) (Table <xref ref-type="table" rid="T4">4</xref>). Interestingly, antimicrobials were used in records where diagnoses included non-bacterial conditions (e.g. porcine circovirus infection and influenza) (Table <xref ref-type="table" rid="T4">4</xref>).</p><table-wrap position="float" id="T4"><label>Table 4</label><caption><p>Diagnoses recorded in OSVS records that indicated antimicrobial use, when antimicrobials were used in treatment (July 1, 2007 – June 30, 2009)</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Diagnosis</bold></th><th align="left"><bold>Number of records</bold></th><th align="left"><bold>Proportion of records</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom"><italic>Streptococcus suis</italic><hr/></td><td align="left" valign="bottom">94<hr/></td><td align="left" valign="bottom">13.91<hr/></td></tr><tr><td align="left" valign="bottom">PRRS<hr/></td><td align="left" valign="bottom">93<hr/></td><td align="left" valign="bottom">13.76<hr/></td></tr><tr><td align="left" valign="bottom">Scours<sup>a</sup><hr/></td><td align="left" valign="bottom">49<hr/></td><td align="left" valign="bottom">7.25<hr/></td></tr><tr><td align="left" valign="bottom"><italic>Escherichia coli</italic><hr/></td><td align="left" valign="bottom">45<hr/></td><td align="left" valign="bottom">6.66<hr/></td></tr><tr><td align="left" valign="bottom">Circovirus<hr/></td><td align="left" valign="bottom">43<hr/></td><td align="left" valign="bottom">6.36<hr/></td></tr><tr><td align="left" valign="bottom">Arthritis<hr/></td><td align="left" valign="bottom">40<hr/></td><td align="left" valign="bottom">5.92<hr/></td></tr><tr><td align="left" valign="bottom">Ileitis<hr/></td><td align="left" valign="bottom">31<hr/></td><td align="left" valign="bottom">4.59<hr/></td></tr><tr><td align="left" valign="bottom">Influenza<hr/></td><td align="left" valign="bottom">30<hr/></td><td align="left" valign="bottom">4.44<hr/></td></tr><tr><td align="left" valign="bottom">Erysipelas<hr/></td><td align="left" valign="bottom">29<hr/></td><td align="left" valign="bottom">4.29<hr/></td></tr><tr><td align="left" valign="bottom">Glassers<hr/></td><td align="left" valign="bottom">28<hr/></td><td align="left" valign="bottom">4.14<hr/></td></tr><tr><td align="left" valign="bottom">Greasy pig<sup>a</sup><hr/></td><td align="left" valign="bottom">19<hr/></td><td align="left" valign="bottom">2.81<hr/></td></tr><tr><td align="left" valign="bottom"><italic>Actinobacillus pleuropneumonia</italic><hr/></td><td align="left" valign="bottom">19<hr/></td><td align="left" valign="bottom">2.81<hr/></td></tr><tr><td align="left" valign="bottom"><italic>Salmonella</italic><hr/></td><td align="left" valign="bottom">18<hr/></td><td align="left" valign="bottom">2.66<hr/></td></tr><tr><td align="left" valign="bottom">Lameness<hr/></td><td align="left" valign="bottom">17<hr/></td><td align="left" valign="bottom">2.51<hr/></td></tr><tr><td align="left" valign="bottom">Mycoplasma<hr/></td><td align="left" valign="bottom">17<hr/></td><td align="left" valign="bottom">2.51<hr/></td></tr><tr><td align="left">Coccidiosis</td><td align="left">10</td><td align="left">1.48</td></tr></tbody></table><table-wrap-foot><p><sup>a</sup>Specific pathogen not indicated.</p></table-wrap-foot></table-wrap><p>Significant predictors within the final multi-level logistic regression model describing treatment failure included body system affected, the number of antimicrobials used and the use of neomycin (Table <xref ref-type="table" rid="T5">5</xref>). Practitioners and farm were included as random effects, accounting for the 3-level nested structure of the data (reports from farms within a veterinarian’s practice) (Table <xref ref-type="table" rid="T5">5</xref>). No significant interaction terms were found. Records indicating GI disease were at significantly decreased odds of treatment failure as compared to multisystemic, musculoskeletal and respiratory body systems (Table <xref ref-type="table" rid="T5">5</xref>). No other significant differences in treatment failure were present between body systems. As the number of antimicrobials used increased, so did the odds of treatment failure. The odds of failure increased by 2.29 times if two antimicrobials were used as compared to one antimicrobial (p <0.01) and by 7.56 times if three or more antimicrobials were used as compared to one antimicrobial (p = 0.01). No significant increase in treatment failure was seen between the use of two antimicrobials and three or more antimicrobials (OR = 1.19; p = 0.15; CI: -0.45 – 2.84). Finally, reduced odds of treatment failure was found in records where neomycin was used (OR = 0.34; p = 0.02). No other antimicrobials were significantly associated with treatment failure after controlling for body system affected.</p><table-wrap position="float" id="T5"><label>Table 5</label><caption><p>Odds ratios and p-values for the mixed logistic regression model describing the effect of body system treated, number of antimicrobials used and the use of neomycin upon treatment failure in 676 OSVS records, when antimicrobials were used in treatment (July 1, 2007 – July 30, 2009)</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="left"/><col align="left"/></colgroup><thead valign="top"><tr><th align="left"><bold>Fixed effects</bold></th><th align="left"><bold>Odds ratio</bold></th><th align="left"><bold>Standard error</bold></th><th align="left"><bold>P-value</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">System<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">0.01<hr/></td></tr><tr><td align="left" valign="bottom">GI<hr/></td><td align="left" valign="bottom">Referent<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td></tr><tr><td align="left" valign="bottom">Multisystemic<hr/></td><td align="left" valign="bottom">3.14<hr/></td><td align="left" valign="bottom">0.99<hr/></td><td align="left" valign="bottom"><0.01<hr/></td></tr><tr><td align="left" valign="bottom">Musculoskeletal<hr/></td><td align="left" valign="bottom">2.39<hr/></td><td align="left" valign="bottom">1.00<hr/></td><td align="left" valign="bottom">0.04<hr/></td></tr><tr><td align="left" valign="bottom">Other<hr/></td><td align="left" valign="bottom">1.96<hr/></td><td align="left" valign="bottom">0.76<hr/></td><td align="left" valign="bottom">0.08<hr/></td></tr><tr><td align="left" valign="bottom">Respiratory<hr/></td><td align="left" valign="bottom">2.40<hr/></td><td align="left" valign="bottom">0.75<hr/></td><td align="left" valign="bottom"><0.01<hr/></td></tr><tr><td align="left" valign="bottom">Not recorded<hr/></td><td align="left" valign="bottom">1.47<hr/></td><td align="left" valign="bottom">0.78<hr/></td><td align="left" valign="bottom">0.46<hr/></td></tr><tr><td align="left" valign="bottom">Number of antimicrobials used<hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom"> <hr/></td><td align="left" valign="bottom">0.01<sup>a</sup><hr/></td></tr><tr><td align="left" valign="bottom">1<hr/></td><td align="left" valign="bottom">Referent<hr/></td><td align="left" valign="bottom">-<hr/></td><td align="left" valign="bottom">-<hr/></td></tr><tr><td align="left" valign="bottom">2<hr/></td><td align="left" valign="bottom">2.29<hr/></td><td align="left" valign="bottom">0.72<hr/></td><td align="left" valign="bottom"><0.01<hr/></td></tr><tr><td align="left" valign="bottom">3 or more<hr/></td><td align="left" valign="bottom">7.56<hr/></td><td align="left" valign="bottom">6.10<hr/></td><td align="left" valign="bottom">0.01<hr/></td></tr><tr><td align="left" valign="bottom">Neomycin<hr/></td><td align="left" valign="bottom">0.34<hr/></td><td align="left" valign="bottom">0.15<hr/></td><td align="left" valign="bottom">0.02<hr/></td></tr><tr><td align="left" valign="bottom">Intercept<hr/></td><td align="left" valign="bottom">1.02<hr/></td><td align="left" valign="bottom">0.41<hr/></td><td align="left" valign="bottom">0.97<hr/></td></tr><tr><td align="left" valign="bottom">Random intercepts<hr/></td><td align="left" valign="bottom">Variance<hr/></td><td align="left" valign="bottom">Standard error<hr/></td><td align="left" valign="bottom">P-value<hr/></td></tr><tr><td align="left" valign="bottom">Practitioner<hr/></td><td align="left" valign="bottom">0.81<hr/></td><td align="left" valign="bottom">0.62<hr/></td><td align="left" valign="bottom"><0.01<sup>b</sup><hr/></td></tr><tr><td align="left">Farm</td><td align="left">0.60</td><td align="left">0.39</td><td align="left"> </td></tr></tbody></table><table-wrap-foot><p><sup>a</sup>P-value for the likelihood ratio test comparing the model with and without the system variable.</p><p><sup>b</sup>P-value for the likelihood ratio test comparing the random effects model to the fixed-effects model alone.</p></table-wrap-foot></table-wrap><p>The variance partition coefficient indicated that the majority of variation in reported treatment failure occurred at the report level after accounting for fixed effects (70.0%) as compared to the practitioner (17.2%) and farm levels (17.8%).</p><p>The BLUPS at the farm and practitioner levels were normally distributed by visual assessment. No anomalous values for the Pearson or deviance residuals were apparent at the report level.</p></sec><sec><title>Discussion and conclusion</title><p>This work presents an assessment of the use of antimicrobials in the Ontario swine industry and the frequency of treatment failure when antimicrobials were used for treating disease. Results here support other reports of common antimicrobial use within the swine industry [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B8">8</xref>]. Similarly, the most commonly used antimicrobials in the OSVS dataset were consistent with the three highest-use injectable antimicrobials reported by CIPARS [<xref ref-type="bibr" rid="B9">9</xref>]. Although the administration route for antimicrobials is available within the CIPARS data, the reasons for use are not available. However, sentinel practitioner systems such as the OSVS may provide insight on the proportion of antimicrobial use for disease treatment as compared to growth promotion or prophylaxis, as OSVS participants were requested to only record treatments (not routine use). Use of multiple antimicrobials was not uncommon within the reports assessed here; more than 20% of records indicated that ≥ 2 antimicrobials were used for treatment. Furthermore, these data displayed that antimicrobials were used in cases where the expected diagnosis was viral (e.g., porcine circovirus infection and influenza) or non-infectious (e.g., injury), either as a precautionary measure if a suspected viral infection is actually bacterial, or to prevent/treat secondary bacterial infections.</p><p>In reports indicating antimicrobial use, treatment failure was surprisingly high (70%). To assess predictors for treatment failure within these records, a multi-level logistic regression model was built with the assumption that the reported treatment failure reflected failure of the antimicrobial mentioned in the treatment field. Results of this model indicated that the odds of treatment failure was associated with the body system affected, the number of antimicrobials used in treatment and use of the antimicrobial neomycin. The odds of treatment failure was significantly lower when the GI system was affected as compared to respiratory, multisystemic, musculoskeletal or other body system conditions, and the odds of treatment failure with multisystemic conditions was significantly higher than in reports with no body system recorded.</p><p>Significant differences in treatment failure among body systems affected likely reflected the etiology of common swine conditions. Many respiratory and multi-systemic conditions in swine have a viral etiology (e.g., porcine circovirus infection, influenza, PRRS) and antimicrobial treatment is not expected to resolve the primary viral infection. This is not to say that antimicrobial treatment is always inappropriate however, as secondary bacterial infections or co-infections may also occur and antimicrobial treatment may prevent the exacerbation of clinical signs [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B11">11</xref>]. Accordingly, prudent use guidelines for antimicrobial use in swine encourage veterinarians to determine the causative agent of disease while recognizing the potential for secondary bacterial infections [<xref ref-type="bibr" rid="B12">12</xref>].</p><p>The differences in the probability of treatment failure among the body systems may also reflect the route of administration of antimicrobials in swine production. Due to large herd or group sizes, treatment of individual animals by injection can be difficult and impractical. Therefore, mass medication of herds through water or feed is a common practice [<xref ref-type="bibr" rid="B9">9</xref>]. Enteric bacterial infections are expected to be effectively treated with this method, given that no metabolism or uptake of the drug into the blood stream/tissue is required. As such, failure of orally administered treatment is expected to occur more often than when other body systems are affected. Furthermore, ill animals are likely to go off-feed, which can result in under-dosing of infections requiring drug metabolism or blood stream uptake. The observed sparing effect of neomycin use may also be explained by body system and route of administration; neomycin is supplied through feed or water in order to treat bacterial enteritis caused by <italic>Escherichia coli</italic> and <italic>Salmonella</italic> spp. [<xref ref-type="bibr" rid="B13">13</xref>].</p><p>The odds of treatment failure increased significantly with the use of multiple antimicrobials. Although practitioners were requested to record data only pertaining to the current visit, the data suggested that records listing multiple antimicrobials for treatment may have reflected either concurrent use of multiple products, or successive use following failure of the primary treatment. However, these results suggest that it may be prudent to explore non-bacterial etiologies and preventative approaches to swine health when the use of two or more antimicrobials is being considered.</p><p>Given that the variance in reported treatment failure was greater at the report level than the farm or practitioner level, it may be assumed that the disease in question has greater influence on the probability of treatment failure than farm- or practitioner-level factors. Therefore, the potential impact of prescription-only standards for accessing antimicrobials on-farm is great, as suggested by the Veterinary Drugs Directorate “Uses of antimicrobials in food animals in Canada” report [<xref ref-type="bibr" rid="B14">14</xref>]. These standards would require producers to obtain a prescription for all antimicrobial use, which is not a current practice in Ontario. Upon the adoption of this recommendation, a shift in the influence may occur towards practitioners. The impact of this shift upon the frequency of treatment failure presents an interesting topic for follow-up studies.</p><p>In instances of non-GI conditions or failure of first-line antimicrobial treatment, a review of the vaccination, biosecurity, artificial insemination, and air quality strategies used on-farm may provide a more effective means of improving and maintaining herd health. Vaccines are available and a topic of ongoing research for many common pathogens of swine, including <italic>S. suis</italic>[<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B16">16</xref>], PRRS [<xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B18">18</xref>], porcine circovirus [<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>], and <italic>E. coli</italic>[<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B22">22</xref>]. Given that the majority of antimicrobial use reported in the OSVS reports reflected the treatment of conditions caused by these pathogens, successful vaccination strategies are expected to lower the probability of antimicrobial use and treatment failure alike. Similarly, two manageable biosecurity measures, the presence of a shower on-site and limited access to main entrances by rendering trucks, have been shown to be associated with reduced probability of positive PRRS virus status on-farm [<xref ref-type="bibr" rid="B23">23</xref>]. Other management practices such as the use of semen from specific-pathogen free boars for artificial insemination [<xref ref-type="bibr" rid="B24">24</xref>], weaning at 28 days of age or later [<xref ref-type="bibr" rid="B25">25</xref>] improving ventilation [<xref ref-type="bibr" rid="B26">26</xref>,<xref ref-type="bibr" rid="B27">27</xref>], reducing group sizes to decrease density [<xref ref-type="bibr" rid="B28">28</xref>] and switching to all-in-all-out flow systems [<xref ref-type="bibr" rid="B28">28</xref>] have been shown to be associated with reduced probability of positive viral status. However, the practicality of some of these changes is questioned, as many may require large economic inputs by producers (e.g., changes requiring renovations to facilities). Therefore, it may be prudent for practitioners, producers, and policy makers to reassess the current guidelines around vaccinations, and the use and acquisition of antimicrobials in swine production. In order to reduce the volume of antimicrobial product being used to treat non-bacterial infections, priority should be given to research that focuses on assessing the health and economic impacts of vaccination, prescription-only standards for antimicrobial use, or increasing the frequency of contact between producers and practitioners.</p><p>Due to the nature of the data collection, it should be noted that there are some potential biases present in this dataset. There is some potential for over-estimation of the use of antimicrobials for treatment of disease, given that the reporting veterinarian(s) may have recorded antimicrobials used for growth promotion and/or prophylaxis in the treatment field. Furthermore, some diagnostic misclassification may have occurred between diseases that present similarly, as laboratory confirmation was not linked to these records.</p><p>This work suggests that failure of antimicrobial treatment is more likely to occur in non-GI conditions, as compared to GI conditions. Furthermore, the use of multiple antimicrobial products is also associated with an increased probability of antimicrobial treatment failure. Improved immunity via vaccination, management and biosecurity strategies may be viewed as a form of antimicrobial stewardship to the industry by reducing required antimicrobial inputs and consequently, reduced selection pressure for AMR. Furthermore, further research is suggested surrounding the economic and health impacts of changes to guidelines surrounding vaccination, antimicrobial acquisition and use, as well as increasing the frequency of contact between producers and practitioners.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>SGK: carried out analyses and drafted the manuscript. RA and RMF participated in the design of the study, data collection and interpretation of results. DLP was involved in the data analyses, manuscript revision, and interpretation. RRS BA and SAM were involved in the revision of the manuscript and the interpretation of data. All authors approved of the final manuscript.</p></sec>
A study on Ovine pneumonic pasteurellosis: Isolation and Identification of Pasteurellae and their antibiogram susceptibility pattern in Haramaya District, Eastern Hararghe, Ethiopia	<sec><title>Background</title><p>Sheep constitute the second major component of livestock in Ethiopia. However, efficient utilization of this potential resource is hampered by combination of health problems, poor management and feed shortage. Haramaya district is one of the remote settings in Ethiopia where information about the livestock disease is not well documented. Hence this study was conducted to determine the causative agents and their antimicrobial susceptibility pattern of bacterial <italic>Pasteurella</italic> isolates among pneumonic ovine in Haramaya district, Eastern Hararghe, Ethiopia.</p></sec><sec><title>Results</title><p>Out of 256 samples examined, <italic>Pasterurella</italic> was isolated in 64 (25%), of which 38 (59.4%) were from lungs and 26 (40.6%) were from nasal cavities. 87.5% of the isolates were <italic>Mannheimia haemolytica</italic> and 12.5% were <italic>Pasteurella multocida</italic>. All of the isolates from the lungs were <italic>Mannheimia haemolytica</italic> whereas 69% of the isolates from nasals cavities were <italic>Mannheimia haemolytica</italic>. Age and body temperature were significantly associated with <italic>Pasteurella</italic> isolates from clinic (P < 0.05). Despite diverse in the site of origins, the isolates exhibited uniformity in sensitivity to a majority of the antibacterial agents. The most effective drug was Cholramphenicol (100%) followed by Sulfamethoxazole (89.1%) and Tetracycline (84.4%). Both species were completely resistant to Gentamycin and Vancomycin.</p></sec><sec><title>Conclusion</title><p><italic>Mannheimia haemolytica</italic> is the most common cause of ovine pneumonic pasteurellosis in the study area. The isolates were susceptible to limited antimicrobial agents. Therefore, the antimicrobial susceptibility test should be conducted before treatment, except for critical cases.</p></sec>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Marru</surname><given-names>Haimanot D</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>haimadis2012@gmail.com</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Anijajo</surname><given-names>Takele T</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>takeleteklua@gmail.com</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Hassen</surname><given-names>Adem A</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>ademadab@gmail.com</email></contrib>	BMC Veterinary Research	<sec><title>Background</title><p>Pneumonia is extremely common in sheep and can be responsible for enormous financial losses worldwide. The condition usually appears when sheep are exposed to combinations of predisposing factors such as adverse physical condition, physiological stress, bacterial and viral infections. As the exact nature of these combinations is unknown, much remain to be understood about why the disease occurs in the way it does [<xref ref-type="bibr" rid="B1">1</xref>]. Sub-Saharan Africa is endowed with a large population of sheep. The population of sheep is estimated at 132.5 million head [<xref ref-type="bibr" rid="B2">2</xref>]. However, animal productivity per head is low.</p><p>Ethiopia lies within the tropical latitude of Africa and has an extremely diverse topography, a wide range of climatic features and a multitude of agro-ecological zone which makes the country suitable for different agricultural production system. This in turn has contributed to the existence of a large diversity of farm animal genetic resource in the country [<xref ref-type="bibr" rid="B3">3</xref>]. Sheep constitute the second major component of livestock in Ethiopia. Despite the large livestock population of Ethiopia the economic benefits remain marginal due to prevailing diseases, poor nutrition, poor animal production systems, reproductive inefficiency management constraint and general lack of Veterinary core [<xref ref-type="bibr" rid="B4">4</xref>].</p><p>Pneumonic pasteurellosis is one of the priority diseases that deserve control. However, control of pneumonic pasteurellosis is difficult task that requires integration of various techniques. Developed and developing countries practice various control mechanisms for primary diseases. While developing countries including Ethiopia could not apply the strategies used by developed countries because of economic reason. The most economic and feasible control method for developing nations is the use of vaccine (Fekadu and Girum<italic>:</italic> Study of the prevalent serotype of Ovine pasteurellosis in the high land of Eastern Amhara sub-region. Kombolcha Veterinary laboratory, unpublished<italic>:</italic> 2001).</p><p>In Ethiopia there are some works conducted on ovine pneumonic pasteurellosis: seroprevalence, bacteriological prevalence in pneumonic lungs and detection of most prevalent serotyping throughout the selected sites of Ethiopia. But, no work was done in Eastern Hararghe at Haramaya district on the <italic>Pasteurella</italic> isolation as well as the drugs or antimicrobial to which these isolates are susceptible for. Therefore, this study was focused on the isolation and identification of <italic>Pastuerella</italic> species that involved in ovine pneumonic pasteurellosis among apparently healthy sheep with pneumonic lungs slaughtered at Haramaya municipal abattoir and sheep with clinical manifestation presented to the Haramaya veterinary clinic during the study period. In addition, this study attempted to determine drug susceptibility pattern of these isolates.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Study area</title><p>The study was conducted at Haramaya district which is located in Eastern Hararghe Zone of Ethiopia, approximately 500 kms East of Addis Ababa. Its elevation is approximately 2000 m a.s.l. and the mean annual temperature and relative humidity are 18°C and 65%, respectively. An annual rainfall is approximately 900 mm, with a bimodal distribution pattern, peaking in mid April and mid August. There are four seasons, such as a short rainy season (from mid March to mid May), a short dry season (from end May to June), a long wet season (July to mid October) and a long dry season (end of October to February). Main pasture production is expected after the short rainy season, continuing until the end of the long wet season. Mixed type agriculture is the main occupation of the population of the area. According to the veterinary sector of Haramaya district agricultural office 2000 report, a total livestock population of this district was 193,334; of which 64,510 Cattle, 28,359 goats, 18,930 sheep, 15,277 donkeys, 5 mules, 530 camels and 65,723 poultry [<xref ref-type="bibr" rid="B5">5</xref>].</p></sec><sec><title>Study design</title><p>A cross-sectional study was undertaken from November 2007- April 2008 on ovine pneumonic pasteurellosis to isolate and identify the <italic>Pasteurella</italic> species from specimens of nasal swab of clinically suspected pneumonic sheep from Haramaya veterinary clinic and lung swab of pneumonic lungs of apparently healthy sheep slaughtered at Haramaya municipal abattoir.</p></sec><sec><title>Study animal</title><p>The animals used in this study were apparently healthy sheep slaughtered at Haramaya municipal abattoir and sheep with clinical manifestations presented to the Haramaya veterinary clinic during the study period. At the abattoir, the lungs of the slaughtered animals were visualized, palpated and inspected thoroughly and finally the pneumonic lungs were considered. According to the postmortem meat inspection principle the lungs of the slaughtered animal were visualized and palpated for haemorrhage, edema and pneumonia. Up on visualization, palpation and inspection the lungs with consolidated inflamed area, deep red and sharply demarcated lesion were considered as pneumonic lungs [<xref ref-type="bibr" rid="B6">6</xref>]. All sheep irrespective of age, sex, and color were examined at clinic for the evidence of pneumonic pasteurellosis. Up on clinical examination, all sheep manifesting; anorexia, coughing, dyspnea, lethargy, serous to muco-purulent ocular and nasal discharge, and fever were considered. In both cases the animals with pneumonic lung and the mentioned clinical features were considered to be study population. The sample size for each varied according to the availability of suspected cases of sheep with pneumonia in respective to study site. A total of 256 (83 from clinic and 173 from abattoir) suspected cases were collected.</p></sec><sec><title>Ethical consideration</title><p>Investigators treated animals with kindness and took proper care by minimizing discomfort, distress or pain. They assumed that all procedures which would cause pain in human beings may cause pain in study animals. The required procedures were conducted by qualified and experienced persons [<xref ref-type="bibr" rid="B7">7</xref>]. The ethical clearance was obtained from Haramaya University ethical review board.</p></sec><sec><title>Data collection</title><sec><title>Nasal swab</title><p>Each animal was individually identified and restrained by an assistant and kept fixed. After disinfection of external part of the nose with 70% alcohol, a sterile cotton-tipped swab was inserted in to the nostril and rotated against the wall of the nasal cavity [<xref ref-type="bibr" rid="B8">8</xref>]. The swab was placed in labeled sterile test tube that contains 3 ml of tryptose Soya broth, and then kept in an ice box for transport to Haramaya University, FVM. Laboratory [<xref ref-type="bibr" rid="B8">8</xref>].</p></sec><sec><title>Lung swab</title><p>Following the slaughter of the apparently healthy animal, all the lungs were inspected according to the standard postmortem meat inspection procedure. Up on inspection the surface of each suspected lung was incised using sterile scalpel blade and the inner surface of the incision was sampled with sterile swab. The swab was transported to Veterinary Microbiology Laboratory of Haramaya University in the same procedure with nasal swab [<xref ref-type="bibr" rid="B9">9</xref>].</p></sec></sec><sec><title>Isolation and Identification of Pasteurella</title><p>The isolation and identification of <italic>Pasteurella</italic> were performed at the Veterinary Microbiology Laboratory of Haramaya University using techniques recommended by Hardy Diagnostics, Santa Maria, CA, USA. The isolation and identification involves the following steps: first, the pre-enriched in tryptose Soya broth specimen was incubated for 24 hrs at 37°C. After 24 hrs incubation, a loop full of the broth cultures were taken and streaked over an identified Petri-plate containing blood agar base supplemented with 7% sheep blood and immediately incubated aerobically at 37°C for 24 hours [<xref ref-type="bibr" rid="B9">9</xref>]. Secondly, from culture positive plates, typical colonies were subjected to gram’s staining to study staining reactions and cellular morphology under light microscope, at 100x magnification. Mixed and gram-negative bacteria were further sub cultured with due care, on both blood and MacConkey agar plates [<xref ref-type="bibr" rid="B9">9</xref>] for further analysis. The growth of typical colonies on both blood and MacConkey agar was characterized using blood agar for the presence of haemolysis, the type of haemolysis, the general appearance of colonies (morphology, color, shape size and consistency) and the ability to ferment lactose [<xref ref-type="bibr" rid="B10">10</xref>]. Thirdly, pure cultures of single colony type from both Blood and MacConkey agars were transferred onto nutrient agar-slants for a series of primary biochemical tests: catalase (Hydrogen peroxide, Fisher Chemical, UK), oxidase (TM-pphenylenediamine dihydrocholoide, Merck Co., Germany) and fermentative/oxidative (OF Basal Medium, Titan Biotech Ltd, India). Final identification of the bacteria to the species level was aided by using the secondary tests which include: metabolism of sugars such as glucose and L-arabinose; and alcohols and tests for metabolic end products such as Indole (Peptone water, Merck Co., Germany) following standard procedures [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B12">12</xref>]. The assay for biochemical properties of the bacterial isolates were conducted according to MacFaddin’s method [<xref ref-type="bibr" rid="B13">13</xref>]. For reliable identification and comparison of results, the AIPE 20 system (Biomariux France) was used. If the organism is able to produce a narrow zone of haemolysis on Blood agar and grow on McConkey agar, but unable produce indole, interpreted as <italic>P.haemolytica</italic>. If the organism is unable to produce haemolysis on Blood agar and cannot grow on MacConkey, but able to produce indole, interpreted as P. <italic>multocida.</italic></p></sec><sec><title>Antimicrobial susceptibility test</title><p>The antimicrobial susceptibility test on the isolates was performed according to the National Committee for Clinical Laboratory Standards (NCCLS, 1990) [<xref ref-type="bibr" rid="B14">14</xref>] method using Kibry-Bauer disk diffusion test on Muller-Hinton agar (Oxoid CM0337 Basingstoke, England). <italic>Escherichia coli</italic> ATCC 25922 was used as a quality control organism for the antimicrobial susceptibility test [<xref ref-type="bibr" rid="B15">15</xref>]. The isolates were tested for the following antibiotics; Chloramphenicol (CAF) 30Ng, Tetracycline (TTC) 30Ng, Penicillin-G (P) 10 unit, Ampicilin (Am) 10Ng, Sulfamethoxazole(SxT) 5Ng, Streptomycin(S) 10Ng, Vancomycin (VA) 30Ng and Gentamicin (CN) 30Ng based on the procedure recommended by Carter [<xref ref-type="bibr" rid="B16">16</xref>] and Quinn <italic>et al.</italic>[<xref ref-type="bibr" rid="B17">17</xref>]. The zone of inhibition was interpreted based on the Performance Standards for Antimicrobial Susceptibility Testing; Sixteenth Informational Supplement [<xref ref-type="bibr" rid="B18">18</xref>] as detailed in Table  <xref ref-type="table" rid="T1">1</xref>.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Zone interpretive chart for antimicrobials (inhibition zone diameter in mm)</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="center" valign="bottom"><bold>Antimicrobial agent</bold><hr/></th><th align="center" valign="bottom"><bold>Disc potency</bold><hr/></th><th align="center" valign="bottom"><bold>Resistant</bold><hr/></th><th align="center" valign="bottom"><bold>Intermediate</bold><hr/></th><th align="center" valign="bottom"><bold>Susceptible</bold><hr/></th></tr><tr><th align="left"> </th><th align="center"><bold>(μg)</bold></th><th align="center"><bold>(≤)</bold></th><th align="center"> </th><th align="center"><bold>(≥)</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Tetracycline<hr/></td><td align="center" valign="bottom">30<hr/></td><td align="center" valign="bottom">14<hr/></td><td align="center" valign="bottom">15-18<hr/></td><td align="center" valign="bottom">19<hr/></td></tr><tr><td align="left" valign="bottom">Streptomycin<hr/></td><td align="center" valign="bottom">10<hr/></td><td align="center" valign="bottom">11<hr/></td><td align="center" valign="bottom">12-14<hr/></td><td align="center" valign="bottom">15<hr/></td></tr><tr><td align="left" valign="bottom">Chloramphenicol<hr/></td><td align="center" valign="bottom">30<hr/></td><td align="center" valign="bottom">12<hr/></td><td align="center" valign="bottom">13-17<hr/></td><td align="center" valign="bottom">18<hr/></td></tr><tr><td align="left" valign="bottom">Vancomycin<hr/></td><td align="center" valign="bottom">30<hr/></td><td align="center" valign="bottom">14<hr/></td><td align="center" valign="bottom">15-16<hr/></td><td align="center" valign="bottom">17<hr/></td></tr><tr><td align="left" valign="bottom">Penicillin-G<hr/></td><td align="center" valign="bottom">10u<hr/></td><td align="center" valign="bottom">28<hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom">29<hr/></td></tr><tr><td align="left" valign="bottom">Ampicillin<hr/></td><td align="center" valign="bottom">10<hr/></td><td align="center" valign="bottom">13<hr/></td><td align="center" valign="bottom">14-16<hr/></td><td align="center" valign="bottom">17<hr/></td></tr><tr><td align="left" valign="bottom">Sulphamethoxozole<hr/></td><td align="center" valign="bottom">25<hr/></td><td align="center" valign="bottom">10<hr/></td><td align="center" valign="bottom">11-15<hr/></td><td align="center" valign="bottom">16<hr/></td></tr><tr><td align="left">Gentamicine</td><td align="center">10</td><td align="center">12</td><td align="center">13-14</td><td align="center">15</td></tr></tbody></table><table-wrap-foot><p>Source: Clinical and Laboratory Standards Institute (CLSI) [<xref ref-type="bibr" rid="B18">18</xref>].</p></table-wrap-foot></table-wrap></sec><sec><title>Data entry and analysis</title><p>Data collected from both the clinic and abattoir were recorded in the format developed for this purpose and later on entered into the Microsoft excel 2000 program and analyzed using STATA 7.0 software. Association of host risk factors with swab culture positives was calculated. A statistically significant association between variables was considered to exist if the computed p-valve was less than 0.05.</p></sec></sec><sec sec-type="results"><title>Results</title><sec><title>Bacterial isolation</title><p>From 256 samples (83 nasal swabs and 173 lung swab) collected and cultured, <italic>Pasteurella</italic> was isolated successfully in 64 (25%) sheep. Out of 64, 26 (40.6%) were from nasal cavities and 38 (59.4%) from lungs. Eighty seven point five percent of the isolate was <italic>M. haemolytica</italic> and 12.5% was <italic>P.multocida</italic> and of the lung samples which were culture positive, 100% was <italic>M. haemolytica.</italic> In nasal cavities, of the samples which were culture positive, 69% of the isolate was <italic>M. haemolytica</italic> and 31% of the isolate was <italic>P.multocida</italic> (Table  <xref ref-type="table" rid="T2">2</xref>). On the basis of these results <italic>M. haemolytica</italic> was the most common cause of pneumonic pasteurellosis in sheep at Haramaya district.</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p><bold> <italic>Pasturella</italic> </bold><bold>Isolates from nasal cavity and lung samples of Ovine at clinic and abattoir at Haramaya district in Eastern Hararghe, Ethiopia</bold></p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="center" valign="bottom"><bold> <italic>Site</italic> </bold><hr/></th><th colspan="2" align="center" valign="bottom"><bold> <italic>Species</italic> </bold><hr/></th><th align="center" valign="bottom"><bold> <italic>Total</italic> </bold><hr/></th><th align="center" valign="bottom"><bold>Percentage of total isolates</bold><hr/></th><th align="center" valign="bottom"><bold>X2 (P-value)</bold><hr/></th></tr><tr><th align="left" valign="bottom"> <hr/></th><th align="center" valign="bottom"><bold> <italic>M.haemolytica</italic> </bold><hr/></th><th align="center" valign="bottom"><bold> <italic>P.multocida</italic> </bold><hr/></th><th align="center" valign="bottom"> <hr/></th><th align="center" valign="bottom"> <hr/></th><td> </td></tr><tr><th align="left"> </th><th align="center"><bold>Positive</bold></th><th align="center"><bold>Positive</bold></th><th align="center"> </th><th align="center"> </th><th align="center"> </th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Nasal cavity<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Positive<hr/></td><td align="center" valign="bottom">18(69%)<hr/></td><td align="center" valign="bottom">8(31%)<hr/></td><td align="center" valign="bottom">26(31.3%)<hr/></td><td align="center" valign="bottom">26(40.6%)<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Negative<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">57(68.7%)<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Total<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">83(100%)<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">2.6208 (0.105)<hr/></td></tr><tr><td align="left" valign="bottom">Lung sample<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Positive<hr/></td><td align="center" valign="bottom">38(100%)<hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom">38(21.96)<hr/></td><td align="center" valign="bottom">38(59.4%)<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Negative<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">135(78.04%)<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Total<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom">173(100%)<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left"><bold>Total</bold></td><td align="center">56(87.5%)</td><td align="center">8(12.5%)</td><td align="center">256(100%)</td><td align="center">64(25%)</td><td align="center"> </td></tr></tbody></table></table-wrap><p><italic>M. haemolytica</italic> was isolated in both clinical and abattoir, but <italic>P.multocida</italic> was exclusive to clinical cases. The magnitude of <italic>M. haemolytica</italic> was almost equal in both clinical (21.69%) and abattoir (21.96%) cases; however, the overall magnitude of the bacteriological confirmed cases is higher in clinical cases (31.3%) than abattoir cases (21.96%). The result indicated that, there is positive association between body temperatures of the animals at clinic and bacteriological confirmed cases of pneumonic pasteurellosis (P = 0.007), but age has negatively associated (P = 0.028) particularly with <italic>P.multocida</italic> (P = 0.004) infection and sex has no any association with both species (P > 0.05) (Table  <xref ref-type="table" rid="T3">3</xref>). No associations were observed between the risk factors and pneumonic pasteurellosis at abattoir (P > 0.05) (Table  <xref ref-type="table" rid="T4">4</xref>).</p><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>Association of culture positive results with the risk factors (age, sex and body temperature) from clinic cases at Haramaya district in Eastern Hararghe, Ethiopia</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="left"><bold> <italic>Age</italic> </bold></th><th align="center"><bold> <italic>Young</italic> </bold></th><th align="center"><bold> <italic>Adult</italic> </bold></th><th colspan="2" align="center"><bold> <italic>Total</italic> </bold></th><th align="center"><bold> <italic>Chi</italic> </bold><sup> <bold> <italic>2</italic> </bold> </sup></th><th align="center"><bold>P-value</bold></th><th align="center"><bold>Overall</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom"><italic>M. haemolytica</italic><hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td colspan="2" align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td rowspan="8" align="center" valign="bottom"><bold> <italic>Chi</italic> </bold><sup> <bold> <italic>2</italic> </bold> </sup><sup> <bold> <italic>=</italic> </bold> </sup><sup> <bold> <italic>4.8501 P = 0.028</italic> </bold> </sup><hr/></td></tr><tr><td align="left" valign="bottom">  Negative<hr/></td><td align="center" valign="bottom">17<hr/></td><td align="center" valign="bottom">48<hr/></td><td colspan="2" align="center" valign="bottom">65<hr/></td><td align="center" valign="bottom">2.2405<hr/></td><td align="center" valign="bottom">0.134<hr/></td></tr><tr><td align="left" valign="bottom">  Positive<hr/></td><td align="center" valign="bottom">8<hr/></td><td align="center" valign="bottom">10<hr/></td><td colspan="2" align="center" valign="bottom">18<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Total<hr/></td><td align="center" valign="bottom">25<hr/></td><td align="center" valign="bottom">58<hr/></td><td colspan="2" align="center" valign="bottom">83<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"><italic>P.multocida</italic><hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td colspan="2" align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Negative<hr/></td><td align="center" valign="bottom">19<hr/></td><td align="center" valign="bottom">56<hr/></td><td colspan="2" align="center" valign="bottom">75<hr/></td><td align="center" valign="bottom">8.472<hr/></td><td align="center" valign="bottom">0.004<hr/></td></tr><tr><td align="left" valign="bottom">  Positive<hr/></td><td align="center" valign="bottom">6<hr/></td><td align="center" valign="bottom">2<hr/></td><td colspan="2" align="center" valign="bottom">8<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Total<hr/></td><td align="center" valign="bottom">25<hr/></td><td align="center" valign="bottom">58<hr/></td><td colspan="2" align="center" valign="bottom">83<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"><bold>Sex</bold><hr/></td><td align="center" valign="bottom"><bold>Male</bold><hr/></td><td align="center" valign="bottom"><bold>Female</bold><hr/></td><td colspan="2" align="center" valign="bottom"><bold>Total</bold><hr/></td><td align="center" valign="bottom"><bold>Chi</bold><sup> <bold>2</bold> </sup><hr/></td><td align="center" valign="bottom"><bold>P-value</bold><hr/></td><td rowspan="9" align="center" valign="bottom">Chi2 = 2.6186 Pr = 0.106<hr/></td></tr><tr><td align="left" valign="bottom"><italic>M.haemolytica</italic><hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td colspan="2" align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Negative<hr/></td><td align="center" valign="bottom">40<hr/></td><td align="center" valign="bottom">25<hr/></td><td colspan="2" align="center" valign="bottom">65<hr/></td><td align="center" valign="bottom">2.9443<hr/></td><td align="center" valign="bottom">0.086<hr/></td></tr><tr><td align="left" valign="bottom">  Positive<hr/></td><td align="center" valign="bottom">7<hr/></td><td align="center" valign="bottom">11<hr/></td><td colspan="2" align="center" valign="bottom">18<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Total<hr/></td><td align="center" valign="bottom">47<hr/></td><td align="center" valign="bottom">36<hr/></td><td colspan="2" align="center" valign="bottom">83<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"><italic>P.multocida</italic><hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td colspan="2" align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Negative<hr/></td><td align="center" valign="bottom">44<hr/></td><td align="center" valign="bottom">31<hr/></td><td colspan="2" align="center" valign="bottom">75<hr/></td><td align="center" valign="bottom">1.3187<hr/></td><td align="center" valign="bottom">0.251<hr/></td></tr><tr><td align="left" valign="bottom">  Positive<hr/></td><td align="center" valign="bottom">3<hr/></td><td align="center" valign="bottom">5<hr/></td><td colspan="2" align="center" valign="bottom">8<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Total<hr/></td><td align="center" valign="bottom">47<hr/></td><td align="center" valign="bottom">36<hr/></td><td colspan="2" align="center" valign="bottom">83<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"><bold>Temprature</bold><hr/></td><td align="center" valign="bottom"><bold>38.5-39.5o</bold><sup> <bold>C</bold> </sup><hr/></td><td align="center" valign="bottom"><bold>39.6-40o</bold><sup> <bold>C</bold> </sup><hr/></td><td align="center" valign="bottom"><bold>>40o</bold><sup> <bold>C</bold> </sup><hr/></td><td align="center" valign="bottom"><bold>Total</bold><hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td rowspan="3" align="center" valign="bottom"><bold><italic>Chi</italic></bold><sup><bold><italic>2</italic></bold></sup> <bold><italic>=</italic></bold> 10.0614 p = 0.007<hr/></td></tr><tr><td align="left" valign="bottom">  Negative<hr/></td><td align="center" valign="bottom">21<hr/></td><td align="center" valign="bottom">23<hr/></td><td align="center" valign="bottom">13<hr/></td><td align="center" valign="bottom">57<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Positive<hr/></td><td align="center" valign="bottom">4<hr/></td><td align="center" valign="bottom">7<hr/></td><td align="center" valign="bottom">15<hr/></td><td align="center" valign="bottom">26<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left">  <bold>Total</bold></td><td align="center">25</td><td align="center">30</td><td align="center">28</td><td align="center">83</td><td align="center"> </td><td align="center"> </td><td align="center"> </td></tr></tbody></table></table-wrap><table-wrap position="float" id="T4"><label>Table 4</label><caption><p>Association of culture positive results from abattoir with risk factors (age and sex) in Abattoir cases at Haramaya District in Eastern Hararghe, Ethiopia</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="left"><bold> <italic>Age</italic> </bold></th><th align="center"><bold> <italic>Young</italic> </bold></th><th align="center"><bold> <italic>Adult</italic> </bold></th><th align="center"><bold> <italic>Total</italic> </bold></th><th align="center"><bold> <italic>Chi</italic> </bold><sup> <bold> <italic>2</italic> </bold> </sup></th><th align="center"><bold>P-value</bold></th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom"><italic>M.haemolytica</italic><hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Negative<hr/></td><td align="center" valign="bottom">12<hr/></td><td align="center" valign="bottom">123<hr/></td><td align="center" valign="bottom">135<hr/></td><td align="center" valign="bottom">2.7560<hr/></td><td rowspan="3" align="center" valign="bottom">0.097<hr/></td></tr><tr><td align="left" valign="bottom">  Positive<hr/></td><td align="center" valign="bottom">7<hr/></td><td align="center" valign="bottom">31<hr/></td><td align="center" valign="bottom">38<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Total<hr/></td><td align="center" valign="bottom">19<hr/></td><td align="center" valign="bottom">154<hr/></td><td align="center" valign="bottom">173<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"><italic>P.multocida</italic><hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Negative<hr/></td><td align="center" valign="bottom">19<hr/></td><td align="center" valign="bottom">154<hr/></td><td align="center" valign="bottom">173<hr/></td><td align="center" valign="bottom">-<hr/></td><td rowspan="3" align="center" valign="bottom">-<hr/></td></tr><tr><td align="left" valign="bottom">  Positive<hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Total<hr/></td><td align="center" valign="bottom">19<hr/></td><td align="center" valign="bottom">154<hr/></td><td align="center" valign="bottom">173<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"><bold>Sex</bold><hr/></td><td align="center" valign="bottom"><bold>Male</bold><hr/></td><td align="center" valign="bottom"><bold>Female</bold><hr/></td><td align="center" valign="bottom"><bold>Total</bold><hr/></td><td align="center" valign="bottom"><bold>Chi</bold><sup> <bold>2</bold> </sup><hr/></td><td align="center" valign="bottom"><bold>P-value</bold><hr/></td></tr><tr><td align="left" valign="bottom"><italic>M.haemolytica</italic><hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Negative<hr/></td><td align="center" valign="bottom">118<hr/></td><td align="center" valign="bottom">17<hr/></td><td align="center" valign="bottom">135<hr/></td><td align="center" valign="bottom">1.1348<hr/></td><td rowspan="3" align="center" valign="bottom">0.567<hr/></td></tr><tr><td align="left" valign="bottom">  Positive<hr/></td><td align="center" valign="bottom">31<hr/></td><td align="center" valign="bottom">7<hr/></td><td align="center" valign="bottom">38<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Total<hr/></td><td align="center" valign="bottom">149<hr/></td><td align="center" valign="bottom">24<hr/></td><td align="center" valign="bottom">173<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom"><italic>P.multocida</italic><hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Negative<hr/></td><td align="center" valign="bottom">149<hr/></td><td align="center" valign="bottom">24<hr/></td><td align="center" valign="bottom">173<hr/></td><td align="center" valign="bottom">-<hr/></td><td rowspan="2" align="center" valign="bottom">-<hr/></td></tr><tr><td align="left" valign="bottom">  Positive<hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left">  <bold>Total</bold></td><td align="center">149</td><td align="center">24</td><td align="center">173</td><td align="center"> </td><td align="center"> </td></tr></tbody></table></table-wrap></sec><sec><title>Antimicrobial susceptibility pattern</title><p>Sixty four isolates of <italic>Pasturella</italic> isolates from clinic and abattoir were subjected to a panel of eight antimicrobials. The antimicrobial susceptibility pattern of the isolates indicated that all isolates were 100%, 100%, 90.6%, and 87.5% resistant to gentamycin, vancomycin, streptomycin and penicillin-G respectively. On the other hand, the isolates were 100%, 89.1%, 84.4% and 53.1% sensitive to chloramphenicol, sulfamethoxazole, tetracycline and ampicilin respectively (Table  <xref ref-type="table" rid="T5">5</xref>). Despite diverse in the site of origins, the isolates exhibited uniformity in sensitivity to a majority of the antibacterial agents. <italic>M. haemolytica</italic> showed 100% resistance to gentamycin and vancomycin while they were 100% sensitive to chloramphenicol followed by 89.3% and 83.9% to sulfamethoxazole and tetracycline consequently. Similarly <italic>P.multocida</italic> showed 100% sensitivity to chloramphenicol followed by sulfamethoxazole and tetracycline. However, <italic>P. multocida</italic> is more susceptible to sulfametrioxazole and tetracycline, but not statically significant (P > 0.05).</p><table-wrap position="float" id="T5"><label>Table 5</label><caption><p><bold>Antimicrobial susceptibility pattern of </bold><bold> <italic>Pasturella </italic> </bold><bold>Isolates from nasal cavity and lung samples of Ovine at clinic and abattoir at Haramaya district in Eastern Hararghe, Ethiopia</bold></p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="center" valign="bottom"><bold>Antimicrobials tested</bold><hr/></th><th colspan="2" align="center" valign="bottom"><bold>Species of bacteria</bold><hr/></th><th align="center" valign="bottom"><bold>Total</bold><hr/></th></tr><tr><th align="left"> </th><th align="center"><bold> <italic>M.haemolytica</italic> </bold></th><th align="center"><bold> <italic>P.multocida</italic> </bold></th><th align="center"> </th></tr></thead><tbody valign="top"><tr><td align="left" valign="bottom">Ampicllin<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Sensitive<hr/></td><td align="center" valign="bottom">30(53.6%)<hr/></td><td align="center" valign="bottom">4(50%)<hr/></td><td align="center" valign="bottom">34(53.1%)<hr/></td></tr><tr><td align="left" valign="bottom">  Resistance<hr/></td><td align="center" valign="bottom">26(46.4%)<hr/></td><td align="center" valign="bottom">4(50%)<hr/></td><td align="center" valign="bottom">30(46.8%)<hr/></td></tr><tr><td align="left" valign="bottom">Chloramphenicol<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Sensitive<hr/></td><td align="center" valign="bottom">56(100%)<hr/></td><td align="center" valign="bottom">8(100%)<hr/></td><td align="center" valign="bottom">64(100%)<hr/></td></tr><tr><td align="left" valign="bottom">  Resistance<hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom">-<hr/></td></tr><tr><td align="left" valign="bottom">Gentamycin<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  <bold>S</bold>ensitive<hr/></td><td align="center" valign="bottom">0(0%)<hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom">-<hr/></td></tr><tr><td align="left" valign="bottom">  Resistance<hr/></td><td align="center" valign="bottom">56(100)<hr/></td><td align="center" valign="bottom">8(100%)<hr/></td><td align="center" valign="bottom">64(100%)<hr/></td></tr><tr><td align="left" valign="bottom">Tetracycline<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Sensitive<hr/></td><td align="center" valign="bottom">47(83.9%)<hr/></td><td align="center" valign="bottom">7(87.5%)<hr/></td><td align="center" valign="bottom">54(84.4%)<hr/></td></tr><tr><td align="left" valign="bottom">  Resistance<hr/></td><td align="center" valign="bottom">9(16.1%)<hr/></td><td align="center" valign="bottom">1(12.5%)<hr/></td><td align="center" valign="bottom">10(15.6%)<hr/></td></tr><tr><td align="left" valign="bottom">Penicillin-G<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Sensitive<hr/></td><td align="center" valign="bottom">6(10.7%)<hr/></td><td align="center" valign="bottom">2(25%)<hr/></td><td align="center" valign="bottom">8(12.5%)<hr/></td></tr><tr><td align="left" valign="bottom">  Resistance<hr/></td><td align="center" valign="bottom">50(89.3%)<hr/></td><td align="center" valign="bottom">6(75%)<hr/></td><td align="center" valign="bottom">56(87.5%)<hr/></td></tr><tr><td align="left" valign="bottom">Streptomycin<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Sensitive<hr/></td><td align="center" valign="bottom">5(8.9%)<hr/></td><td align="center" valign="bottom">1(12.5%)<hr/></td><td align="center" valign="bottom">6(9.4%)<hr/></td></tr><tr><td align="left" valign="bottom">  Resistance<hr/></td><td align="center" valign="bottom">51(91.1%)<hr/></td><td align="center" valign="bottom">7(87.5%)<hr/></td><td align="center" valign="bottom">58(90.6%)<hr/></td></tr><tr><td align="left" valign="bottom">Sulfamethoxazole<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Sensitive<hr/></td><td align="center" valign="bottom">50(89.3%)<hr/></td><td align="center" valign="bottom">7(87.5%)<hr/></td><td align="center" valign="bottom">57(89.1%)<hr/></td></tr><tr><td align="left" valign="bottom">  Resistance<hr/></td><td align="center" valign="bottom">6(10.7%)<hr/></td><td align="center" valign="bottom">1(12.5%)<hr/></td><td align="center" valign="bottom">7(10.9%)<hr/></td></tr><tr><td align="left" valign="bottom">Vancomycin<hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td align="left" valign="bottom">  Sensitive<hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom">-<hr/></td><td align="center" valign="bottom">-<hr/></td></tr><tr><td align="left">  Resistance</td><td align="center">56(100%)</td><td align="center">8(100)</td><td align="center">64(100%)</td></tr></tbody></table></table-wrap></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>In the present study, <italic>Pasteurella</italic> species were isolated in 25% of the sheep, of which 59.4% and 40.6% were contributed from abattoir and clinic respectively. <italic>M. haemolytica</italic> was accounted for 87.5%. The prevalence of <italic>M. haemolytica</italic> was almost similar both at abattoir (21.96%) and clinic (21.69%). <italic>P. multiocida</italic> was discovered only from nasal swabs. Pneumonic pasteurellesis was positively associated with body temperature and negatively associated with the age of suspected ovine particularly <italic>P. multiocida</italic>. Sex has no any association.</p><p>In the present study, <italic>M. haemolytica</italic> was isolated from 21.96% of lung samples examined. This result is greater than the findings of Eshetu [<xref ref-type="bibr" rid="B19">19</xref>] and Nurhusien [<xref ref-type="bibr" rid="B20">20</xref>], which were reported as 13% and 8.7% respectively, but lower, than the findings of Mohammed [<xref ref-type="bibr" rid="B21">21</xref>] and Aschalew [<xref ref-type="bibr" rid="B22">22</xref>] that were reported as 40.8% and 56% respectively in pneumonic lungs. In nasal swab <italic>M. haemolytica</italic> was discovered with the rate of 21.7%, which is similar with that of the lung swab. Nonetheless, the present result indicates high prevalence of the isolate than the work done by Eshetu [<xref ref-type="bibr" rid="B19">19</xref>] who reported 13% in the same area before sixteen years ago. This difference might be due to the type of sample taken from purely pneumonic lung in the present study. The other possible explanation may be improvement of the health facilities within the last ~ two decades.</p><p>Comparing the two <italic>Pasteurella</italic> spp, <italic>M. haemolytica</italic> constituted 87. 5% of the total indicated that, <italic>M. haemolytica</italic> was the major causative agent involved in ovine pneumonic pasteurellosis. This is consistent with previous reports of Aschalew [<xref ref-type="bibr" rid="B22">22</xref>], Eshetu [<xref ref-type="bibr" rid="B19">19</xref>], Mohammed [<xref ref-type="bibr" rid="B21">21</xref>], and Tesfaye [<xref ref-type="bibr" rid="B23">23</xref>]. <italic>M.haemolytica</italic>, which is a normal flora of the upper respiratory tract, may play a secondary role after the primary initiating agent suppressed the host defense mechanism, and favors the multiplication of <italic>Pasteurella</italic> species leading to bronchopneumonia in purely pneumonic animal [<xref ref-type="bibr" rid="B24">24</xref>]. Although the percentage isolation was relatively low (12.5%), the possible role of <italic>P. multocida</italic> in the a etiology and pathogenesis of ovine pneumonia should not be under estimated.</p><p>In our study <italic>Paseurella</italic> species were isolated in 31.3% of nasal swabs. Although they may be found occasionally as a normal inhabitant of the respiratory system, experimental evidence has shown that under certain conditions associated with debilitation, nutrition and climatic factors, these organisms may singly or in concert with other organisms flare up to cause severe infections with high morbidity and mortality. In this particular study the recruitment of animals was based on their definitive clinical sign and symptom to pneumonia and the identification of <italic>Pateurella</italic> species among these animals makes more soundable and interesting. The magnitude of <italic>M. haemolytica</italic> was almost equal in nasal swabs (21.69%) and lung swabs (21.96%). <italic>P. multiocida</italic> was not isolated in the lung samples in the present study, but it was discovered in nasal swab at the rate of 9.6%. This result, however, supports the idea of previous studies [<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B25">25</xref>] that reported <italic>P. multocida</italic> from respiratory tract of sheep and goats with very lower isolation rates from lung.</p><p>In the current study a significant association between pneumonic pasteurellesis and body temperature of suspected ovine irrespective of <italic>Pasteurella</italic> species was observed. This result coincides with the findings of (Fekadu and Girum: Study of the prevalent serotype of Ovine pasteurellosis in the high land of Eastern Amhara sub-region. Kombolcha Veterinary laboratory<italic>,</italic> unpublished:2001) that explained, pneumonic pasteurellosis is characterized by fever, driving or pushing of the body as result of forced respiration, dry coughing and muco-purulent nasal discharges. Similarly, the disease is significantly associated with the age of sheep. This result is also in agreement with findings of Gilmour and Gilmour [<xref ref-type="bibr" rid="B26">26</xref>], that elucidates pneumonic pasteurellosis occur in all ages of sheep and goats, with the most susceptible in lambs and kids during first life, and dams at lambing.</p><p>It is important to monitor the antimicrobial susceptibility of <italic>Pasteurella</italic> species to determine resistance development. Increase in resistance against antibiotics in both <italic>P. multocida</italic> and <italic>M. haemolytica</italic> isolates have been reported [<xref ref-type="bibr" rid="B27">27</xref>,<xref ref-type="bibr" rid="B28">28</xref>]. In our study, according to the antimicrobial susceptibility test results, chloramphenicol (100%), sulfamethoxazole (89.1%) and tetracycline (84.4%) were the most effective drugs, whereas ampicilin (53.1%) was the only intermediate drug. Penicillin-G (12.5%) and streptomycin (9.4%) were inefficient drugs; gentamycin and vancomycin were totally inactive against both isolates. This was in line with the literature which stated as chloramphenicol is highly effective and well-tolerated broad spectrum antibiotic to many genera of gram-positive and gram-negative bacteria [<xref ref-type="bibr" rid="B24">24</xref>]. Susceptibility to sulfamethoxazole over 80% was very close to the rate reported previously [<xref ref-type="bibr" rid="B28">28</xref>]. However, this result contradicts the findings of Aschalew [<xref ref-type="bibr" rid="B22">22</xref>] who reported that tetracycline as ineffective drug of choice. This difference strengthens the recommendation of Kaan <italic>et al.</italic>[<xref ref-type="bibr" rid="B29">29</xref>] which stated that “Antibiotic susceptibility profiles of <italic>P. multocida</italic> and <italic>M. haemolytica</italic> help veterinarians to choose appropriate antibiotic against bovine respiratory disease; however, antibiotic susceptibility studies should be renewed periodically”. In addition, pre-existing resistances in which the cellular mechanisms required for antimicrobial susceptibility are absent from the bacterial cell or acquired genetically because of chromosomal mutation and accusation of transferable genetic material. Treatment with a specific antimicrobial agent selects those micro-organisms that have pre-existing or acquired resistance [<xref ref-type="bibr" rid="B9">9</xref>].</p><p>One of the interesting findings of our study was the demonstration of the highest resistance of <italic>Pasteurella</italic> isolates against gentamycin (100%) and vancomycin (100%). In contrast, Esra <italic>et al.</italic>[<xref ref-type="bibr" rid="B30">30</xref>], reported that gentamycin (95.0%) as the most effective antibiotic against <italic>M. haemolytica</italic> isolates. Similarly Post <italic>et al.</italic>[<xref ref-type="bibr" rid="B31">31</xref>], reported 90.0% of <italic>M. haemolytica</italic> isolates revealed from cattle with bovine respiratory disease complex were markedly susceptible to gentamycin. This might be due to difference in the strain of the isolate that may cause pasteurellosis in different species of animals or due to the existence of host factors that may affect the action of drug in sheep. However this study strengthens the statement vancomycin is active against most gram-positive bacteria, but is not effective against gram-negative cells because of their large size and poor penetrability [<xref ref-type="bibr" rid="B28">28</xref>]. In this study <italic>P. multocida</italic> showed resistance to penicillin-G, this result is in contrary to literature [<xref ref-type="bibr" rid="B32">32</xref>] which indicates most strains of <italic>P. multocida</italic> are susceptible to penicillin-G.</p><p>Limitation of the study: this study was conducted in short period of time; hence we did not saw the seasonal variation of the isolates. In addition, even though identification of serotypes is so important, in our study it was not conducted due to lack of facilities in our laboratory. A part from these limitations our study has the following strengths: the sample was collected from both abattoir and clinic which shows the reliability of the data, the isolates were obtained from pneumonic ovine that made us compare the clinical case with ethological agents, all laboratory works were followed standard procedures and quality control was exhibited in each step of the work.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>In conclusion, Pneumonic Pasteurellosis was the major disease of sheep in the area and <italic>M. haemolytica</italic> is the most common cause. Being young animal was a risk factor for the disease. Both species were susceptible to limited antimicrobial agents. Chloramphenicol, sulfamethoxazole and tetracycline were effective drugs whereas gentamycin and vancomycin were totally inactive against the isolates. Measures such as, improving management practices by providing optimal sanitation and air quality in housing, minimizing transportation stress, providing good quality hay and water, and supplement as appropriate should be taken into account to reduce the prevalence. In this line bacterial isolation and antibiotic susceptibility test should be conducted before treating with antibiotics except for critical ones. Moreover, further serotypoing and molecular techniques are needed to identify the isolate to the strain level.</p></sec><sec><title>Competing interests</title><p>All authors have declared that no competing interests exist.</p></sec><sec><title>Authors’ contributions</title><p>HD: Conception of the research idea, designing and data collection, data analysis and interpretation, and manuscript reviewing. TT: Data collection, interpretation of the results, and drafting the manuscript. (AA): Data collection, interpretation of the results and drafting the manuscript with TT. All authors read and approved the final manuscript.</p></sec>
Modelling diseases with relapse and nonlinear incidence of infection: a multi-group epidemic model	<p>In this paper, we introduce a basic reproduction number for a multi-group SIR model with general relapse distribution and nonlinear incidence rate. We find that basic reproduction number plays the role of a key threshold in establishing the global dynamics of the model. By means of appropriate Lyapunov functionals, a subtle grouping technique in estimating the derivatives of Lyapunov functionals guided by graph-theoretical approach and LaSalle invariance principle, it is proven that if it is less than or equal to one, the disease-free equilibrium is globally stable and the disease dies out; whereas if it is larger than one, some sufficient condition is obtained in ensuring that there is a unique endemic equilibrium which is globally stable and thus the disease persists in the population. Furthermore, our results suggest that general relapse distribution are not the reason of sustained oscillations. Biologically, our model might be realistic for sexually transmitted diseases, such as Herpes, Condyloma acuminatum, etc.</p>	<contrib contrib-type="author"><name><surname>Wang</surname><given-names>Jinliang</given-names></name><xref ref-type="aff" rid="AF1"> <sup>a</sup> </xref><xref ref-type="aff" rid="AF2"> <sup>b</sup> </xref><xref ref-type="corresp" rid="cor1"> <sup>*</sup> </xref></contrib><contrib contrib-type="author"><name><surname>Pang</surname><given-names>Jingmei</given-names></name><xref ref-type="aff" rid="AF2"> <sup>b</sup> </xref><xref rid="AN1" ref-type="author-notes"/></contrib><contrib contrib-type="author"><name><surname>Liu</surname><given-names>Xianning</given-names></name><xref ref-type="aff" rid="AF1"> <sup>a</sup> </xref><xref rid="AN2" ref-type="author-notes"/></contrib><aff id="AF1"><label><sup>a</sup></label><institution><named-content content-type="department">School of Mathematics and Statistics</named-content>, <named-content content-type="institution-name">Southwest University</named-content></institution>, <named-content content-type="city">Chongqing</named-content><named-content content-type="postal-code">400715</named-content>, <country>People's Republic of China</country></aff><aff id="AF2"><label><sup>b</sup></label><institution><named-content content-type="department">School of Mathematical Science</named-content>, <named-content content-type="institution-name">Heilongjiang University</named-content></institution>, <named-content content-type="city">Harin</named-content><named-content content-type="postal-code">150080</named-content>, <country>People's Republic of China</country></aff>	Journal of Biological Dynamics	<sec sec-type="intro" id="S001"><label>1. </label><title>Introduction</title><p>In recent years, great attention has been paid in analysing the multi-group epidemic models, which have been proposed to describe the disease transmission dynamics of many infectious diseases in heterogeneous host populations, such as measles, mumps and gonorrhoea and vector-borne diseases such as West-Nile virus and Malaria. For more and detailed justifications for multi-group disease models and many different types of heterogeneity epidemic models (see e.g. [<xref rid="CIT0003" ref-type="bibr">3</xref>,<xref rid="CIT0013" ref-type="bibr">13</xref>,<xref rid="CIT0014" ref-type="bibr">14</xref>,<xref rid="CIT0016" ref-type="bibr">16</xref>,<xref rid="CIT0020" ref-type="bibr">20</xref>,<xref rid="CIT0023" ref-type="bibr">23</xref>,<xref rid="CIT0032" ref-type="bibr">32</xref>] and references cited therein). It is well known that long-time behaviours of multi-group models with higher dimensions, especially the global asymptotical stability of the endemic equilibrium (EE), is a very challenging topic. The question of uniqueness and global stability of the EE, when the basic reproduction number is more than one, has largely been open. In [<xref rid="CIT0013" ref-type="bibr">13</xref>], a graph-theoretic approach was developed, which is known to be an effective tool for the global stability analysis of multi-group epidemic models. Much research has been done with this approach [<xref rid="CIT0003" ref-type="bibr">3</xref>,<xref rid="CIT0013" ref-type="bibr">13</xref>,<xref rid="CIT0014" ref-type="bibr">14</xref>,<xref rid="CIT0022" ref-type="bibr">22</xref>,<xref rid="CIT0023" ref-type="bibr">23</xref>,<xref rid="CIT0035" ref-type="bibr">35</xref>], focusing on understanding the transmission mechanism and the global behaviour of multi-group epidemic models. So the study of relapse distribution and different nonlinear incidence of infection have been the subject of intense theoretical analysis on the heterogeneous epidemic models in the literature [<xref rid="CIT0026" ref-type="bibr">26</xref>,<xref rid="CIT0034" ref-type="bibr">34</xref>,<xref rid="CIT0035" ref-type="bibr">35</xref>].</p><p>In this paper, motivated by the works of van den Driessche and Zou [<xref rid="CIT0010" ref-type="bibr">10</xref>], Wang <italic>et al.</italic> [<xref rid="CIT0034" ref-type="bibr">34</xref>] and Sun and Shi [<xref rid="CIT0031" ref-type="bibr">31</xref>], we shall investigate the global dynamics of a general multi-group epidemic model with general relapse distribution and nonlinear incidence rate, in particular to investigate the impacts of heterogeneity and nonlinear incidence rate on the dynamics of the basic SIR epidemic model. The population is divided into <italic>n</italic> distinct groups (<italic>n</italic>≥1). For 1≤<italic>i</italic>≤<italic>n</italic>, the <italic>i</italic>th group is further partitioned into three compartments:</p><list list-type="simple"><list-item><p> <italic>S</italic> <sub><italic>i</italic></sub>: susceptible individuals in the <italic>i</italic>th group;</p></list-item><list-item><p> <italic>I</italic> <sub><italic>i</italic></sub>: infectious individuals in the <italic>i</italic>th group;</p></list-item><list-item><p> <italic>R</italic> <sub><italic>i</italic></sub>: recovered individuals in the <italic>i</italic>th group,</p></list-item></list><p>and we denote the populations of individuals at time <italic>t</italic> in each compartment by <italic>S</italic> <sub><italic>i</italic></sub>(<italic>t</italic>), <italic>I</italic> <sub><italic>i</italic></sub>(<italic>t</italic>) and <italic>R</italic> <sub><italic>i</italic></sub>(<italic>t</italic>), respectively. Within the <italic>i</italic>th group, let <inline-formula><inline-graphic xlink:href="tjbd-8-099-m001.jpg"/></inline-formula> represent the growth rate of <italic>S</italic> <sub><italic>i</italic></sub>, which includes both the production and the natural death of susceptible individuals. Typical assumptions on <inline-formula><inline-graphic xlink:href="tjbd-8-099-m002.jpg"/></inline-formula> are the following:</p><list list-type="simple"><list-item><p>(<inline-formula><inline-graphic xlink:href="tjbd-8-099-m003.jpg"/></inline-formula>) ϕ<sub><italic>i</italic></sub> are <italic>C</italic> <sup>1</sup> non-increasing function on [0, ∞) with <inline-formula><inline-graphic xlink:href="tjbd-8-099-m004.jpg"/></inline-formula>, and there is a unique positive solution <inline-formula><inline-graphic xlink:href="tjbd-8-099-m005.jpg"/></inline-formula> for the equation <inline-formula><inline-graphic xlink:href="tjbd-8-099-m006.jpg"/></inline-formula>. <inline-formula><inline-graphic xlink:href="tjbd-8-099-m007.jpg"/></inline-formula> for <inline-formula><inline-graphic xlink:href="tjbd-8-099-m008.jpg"/></inline-formula>, and <inline-formula><inline-graphic xlink:href="tjbd-8-099-m009.jpg"/></inline-formula> for <inline-formula><inline-graphic xlink:href="tjbd-8-099-m010.jpg"/></inline-formula>, that is, <disp-formula id="UM0001"><graphic xlink:href="tjbd-8-099-u001.jpg" position="float" orientation="portrait"/></disp-formula> </p></list-item></list><p>The class of <inline-formula><inline-graphic xlink:href="tjbd-8-099-m011.jpg"/></inline-formula> that satisfy (<inline-formula><inline-graphic xlink:href="tjbd-8-099-m012.jpg"/></inline-formula>) include both <inline-formula><inline-graphic xlink:href="tjbd-8-099-m013.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-099-m014.jpg"/></inline-formula>, which have been widely used in the literature of population dynamics [<xref rid="CIT0001" ref-type="bibr">1</xref>,<xref rid="CIT0013" ref-type="bibr">13</xref>].</p><p>Since nonlinear incidence of infection has been observed in disease transmission dynamics, it has been suggested that the standard bilinear incidence rate shall be modified into a nonlinear incidence rate in many research [<xref rid="CIT0018" ref-type="bibr">18</xref>,<xref rid="CIT0028" ref-type="bibr">28</xref>,<xref rid="CIT0029" ref-type="bibr">29</xref>]. In this paper, we replace the incidence rate <italic>f</italic>(<italic>S</italic>)<italic>I</italic> in [<xref rid="CIT0034" ref-type="bibr">34</xref>] by a general form <italic>f</italic>(<italic>S, I</italic>). We assume that the disease incidence in the <italic>i</italic>th group can be calculated as <disp-formula id="UM0002"><graphic xlink:href="tjbd-8-099-u002.jpg" position="float" orientation="portrait"/></disp-formula> where the sum takes into account cross-infections from all groups, and β<sub><italic>ij</italic></sub> represents the transmission coefficient between compartments <italic>S</italic> <sub><italic>i</italic></sub> and <italic>I</italic> <sub><italic>j</italic></sub>. Throughout the paper, β<sub><italic>ij</italic></sub> is non-negative for all <inline-formula><inline-graphic xlink:href="tjbd-8-099-m015.jpg"/></inline-formula>, and <italic>n</italic>-square matrix <inline-formula><inline-graphic xlink:href="tjbd-8-099-m016.jpg"/></inline-formula> is irreducible [<xref rid="CIT0004" ref-type="bibr">4</xref>]. Biologically, this is the same as assuming that any two groups <italic>i</italic> and <italic>j</italic> have a direct or indirect route of transmission. More specifically, individuals in <italic>I</italic> <sub><italic>j</italic></sub> can infect ones in <italic>S</italic> <sub><italic>i</italic></sub> directly or indirectly.</p><p>Denote <italic>P</italic> <sub><italic>i</italic></sub>(<italic>t</italic>) by the fraction of recovered individuals remaining in the recovered class <italic>t</italic> time units after recovery in each group. It was assumed in [<xref rid="CIT0010" ref-type="bibr">10</xref>] that <italic>P</italic> <sub><italic>i</italic></sub>(<italic>t</italic>) satisfies the following reasonable properties:</p><list list-type="simple"><list-item><p>(<inline-formula><inline-graphic xlink:href="tjbd-8-099-m017.jpg"/></inline-formula>) <inline-formula><inline-graphic xlink:href="tjbd-8-099-m018.jpg"/></inline-formula> is non-increasing, piecewise continuous with possibly finitely many jumps and satisfies <italic>P</italic> <sub><italic>i</italic></sub>(0<sup>+</sup>)=1; <inline-formula><inline-graphic xlink:href="tjbd-8-099-m019.jpg"/></inline-formula> with <inline-formula><inline-graphic xlink:href="tjbd-8-099-m020.jpg"/></inline-formula> is positive and finite.</p></list-item></list><p>The proportion of recovered individuals in the <italic>i</italic>th group can be expressed by the integral <disp-formula id="UM0003"><graphic xlink:href="tjbd-8-099-u003.jpg" position="float" orientation="portrait"/></disp-formula> where γ<sub><italic>i</italic></sub>>0 is the recovery rate constant assuming that the infective period is exponentially distributed in the <italic>i</italic>th group. The term <inline-formula><inline-graphic xlink:href="tjbd-8-099-m021.jpg"/></inline-formula> in the above integral accounts for the death of infectives in <italic>i</italic>th group. It is assumed that no individuals are initially in the recovered class, i.e. <italic>R</italic> <sub><italic>i</italic></sub>(0)=0. Differentiating <italic>R</italic> <sub><italic>i</italic></sub>(<italic>t</italic>) gives <disp-formula id="UM0004"><graphic xlink:href="tjbd-8-099-u004.jpg" position="float" orientation="portrait"/></disp-formula> Here, the integral is in the Riemann–Stieltjes sense to allow for possible jump discontinuities of <italic>P</italic> <sub><italic>i</italic></sub>(<italic>t</italic>).</p><p>Hence, the new multi-group epidemic model with relapse distribution and nonlinear incidence rates can be written as the following 3<italic>n</italic>-dimensional system of differential and integral equations: <disp-formula-group id="M0001"><disp-formula><graphic xlink:href="tjbd-8-099-e001.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> which is an improvement of our earlier model studied in [<xref rid="CIT0034" ref-type="bibr">34</xref>]. It also covers many research works in the literature, for example, the ones in [<xref rid="CIT0010" ref-type="bibr">10</xref>,<xref rid="CIT0026" ref-type="bibr">26</xref>]. The disease transmission diagram is depicted in <xref rid="F0001" ref-type="fig">Figure 1</xref>. And the parameters in the model are summarized in the following list: <fig id="F0001" orientation="portrait" position="float"><label>Fig. 1. </label><caption><p>Transfer diagram for model (1).</p></caption><graphic xlink:href="tjbd-8-099-g001"/></fig> </p><list list-type="simple"><list-item><p>β<sub><italic>ij</italic></sub>: coefficient of transmission between compartments <italic>S</italic> <sub><italic>i</italic></sub> and <italic>I</italic> <sub><italic>j</italic></sub>;</p></list-item><list-item><p>d<sub><italic>i</italic></sub>: natural death rates of all compartments in the <italic>i</italic>th group;</p></list-item><list-item><p>γ<sub><italic>i</italic></sub>: rate of recovery of infectious individuals in the <italic>i</italic>th group.</p></list-item></list><p>In the present paper, our goal is to carry out a complete mathematical analysis of system (1) and establish its global dynamics. By virtue of both the relapse distribution and multi-group structure, model (1) is thought to be realistic for sexually transmitted diseases such as herpes simplex virus type 2 (Herpes) [<xref rid="CIT0006" ref-type="bibr">6</xref>,<xref rid="CIT0007" ref-type="bibr">7</xref>,<xref rid="CIT0033" ref-type="bibr">33</xref>], for which disease is transmitted by close physical or sexual contact, recovered individuals may revert back to the infective class because of reactivation of the latent infection or incomplete treatment [<xref rid="CIT0007" ref-type="bibr">7</xref>,<xref rid="CIT0033" ref-type="bibr">33</xref>]. One important feature of herpes is that an individual once infected remains infected for life, and the virus reactivates regularly with reactivation producing a relapse period of infectiousness (see e.g. [<xref rid="CIT0006" ref-type="bibr">6</xref>] and references cited therein). Therefore, the detailed mathematical analysis for model (1) is an important task from not only mathematical but also biological points. It is thus of interest to investigate whether sustained oscillations are the result of general relapse distribution. This provides us with one motivation to conduct our work.</p><p>The paper is organized as follows. In the next section, we give some preliminaries on system (1). Our main results are stated in Section 3. One case for <italic>n</italic>=1 is investigated in Section 4. Finally, we give a brief summary and discussion.</p></sec><sec id="S002"><label>2. </label><title>Preliminaries</title><p>We first consider system (1) in the phase space ℝ<sup>3<italic>n</italic></sup> and with the initial conditions <disp-formula-group id="M0002"><disp-formula><graphic xlink:href="tjbd-8-099-e002.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> On the basis of biological considerations, we make the following assumptions for <inline-formula><inline-graphic xlink:href="tjbd-8-099-m022.jpg"/></inline-formula> [<xref rid="CIT0031" ref-type="bibr">31</xref>]: <list list-type="simple"><list-item><p>(<inline-formula><inline-graphic xlink:href="tjbd-8-099-m023.jpg"/></inline-formula>): <inline-formula><inline-graphic xlink:href="tjbd-8-099-m024.jpg"/></inline-formula> for <inline-formula><inline-graphic xlink:href="tjbd-8-099-m025.jpg"/></inline-formula>.</p></list-item><list-item><p>(<bold>A</bold> <sub>3</sub>) <inline-formula><inline-graphic xlink:href="tjbd-8-099-m026.jpg"/></inline-formula> for all <italic>I</italic> <sub><italic>j</italic></sub>>0.</p></list-item><list-item><p>(<bold>A</bold> <sub>4</sub>) <inline-formula><inline-graphic xlink:href="tjbd-8-099-m027.jpg"/></inline-formula>, for <inline-formula><inline-graphic xlink:href="tjbd-8-099-m028.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-099-m029.jpg"/></inline-formula>.</p></list-item></list> </p><p>Typical examples of <inline-formula><inline-graphic xlink:href="tjbd-8-099-m030.jpg"/></inline-formula> satisfying <inline-formula><inline-graphic xlink:href="tjbd-8-099-m031.jpg"/></inline-formula>–<inline-formula><inline-graphic xlink:href="tjbd-8-099-m032.jpg"/></inline-formula> include common incidence functions such as <inline-formula><inline-graphic xlink:href="tjbd-8-099-m033.jpg"/></inline-formula> [<xref rid="CIT0012" ref-type="bibr">12</xref>,<xref rid="CIT0013" ref-type="bibr">13</xref>,<xref rid="CIT0019" ref-type="bibr">19</xref>]; <inline-formula><inline-graphic xlink:href="tjbd-8-099-m034.jpg"/></inline-formula> [<xref rid="CIT0001" ref-type="bibr">1</xref>,<xref rid="CIT0011" ref-type="bibr">11</xref>]; <inline-formula><inline-graphic xlink:href="tjbd-8-099-m035.jpg"/></inline-formula> [<xref rid="CIT0024" ref-type="bibr">24</xref>,<xref rid="CIT0025" ref-type="bibr">25</xref>,<xref rid="CIT0037" ref-type="bibr">37</xref>].</p><statement><label>lemma 2.1 </label><p>For initial conditions in Equation (2) with <inline-formula><inline-graphic xlink:href="tjbd-8-099-m036.jpg"/></inline-formula> <inline-formula><inline-graphic xlink:href="tjbd-8-099-m037.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-099-m038.jpg"/></inline-formula> the solutions of system (1) are ultimately uniformly bounded in ℝ<sup>3<italic>n</italic></sup>.</p></statement><p> <italic>Proof</italic> It follows from <inline-formula><inline-graphic xlink:href="tjbd-8-099-m039.jpg"/></inline-formula> and the first equation of Equation (5) that <inline-formula><inline-graphic xlink:href="tjbd-8-099-m040.jpg"/></inline-formula> for all <italic>i</italic>=1, 2, … , <italic>n</italic>. Next we show that the solution of system (1) is ultimately bounded in <inline-formula><inline-graphic xlink:href="tjbd-8-099-m041.jpg"/></inline-formula>. For each <italic>i</italic>, adding the three equations in Equation (1) gives <disp-formula-group id="M0003"><disp-formula><graphic xlink:href="tjbd-8-099-e003.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula id="UM0005"><graphic xlink:href="tjbd-8-099-u005.jpg" position="float" orientation="portrait"/></disp-formula> for <italic>i</italic>=1, 2, … , <italic>n</italic>. Hence, <disp-formula id="UM0006"><graphic xlink:href="tjbd-8-099-u006.jpg" position="float" orientation="portrait"/></disp-formula> It follows from <inline-formula><inline-graphic xlink:href="tjbd-8-099-m042.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-099-m043.jpg"/></inline-formula> that if <inline-formula><inline-graphic xlink:href="tjbd-8-099-m044.jpg"/></inline-formula> is a solution satisfying <inline-formula><inline-graphic xlink:href="tjbd-8-099-m045.jpg"/></inline-formula> for some <italic>t</italic> <sub>0</sub>>0, then <inline-formula><inline-graphic xlink:href="tjbd-8-099-m046.jpg"/></inline-formula> for all <italic>t</italic>≥<italic>t</italic> <sub>0</sub>. By Equation (3), we can also obtain that, for any <italic>i</italic>=1, 2, … , <italic>n</italic>, if <inline-formula><inline-graphic xlink:href="tjbd-8-099-m047.jpg"/></inline-formula> for some <italic>t</italic> <sub>1</sub>>0, then <inline-formula><inline-graphic xlink:href="tjbd-8-099-m048.jpg"/></inline-formula> for all <italic>t</italic>≥<italic>t</italic> <sub>1</sub>. Therefore, the set <disp-formula-group id="M0004"><disp-formula><graphic xlink:href="tjbd-8-099-e004.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> is a forward invariant compact absorbing set with respect to the system (1). Let <disp-formula id="UM0007"><graphic xlink:href="tjbd-8-099-u007.jpg" position="float" orientation="portrait"/></disp-formula> It can be shown that Γ<sub>0</sub> is the interior of Γ. Furthermore, all positive semi-orbits in Γ are precompact ℝ<sup>3<italic>n</italic></sup> [<xref rid="CIT0002" ref-type="bibr">2</xref>] and thus have non-empty ω-limit sets.</p><p>Note that the first two equations of system (1) are independent from <italic>R</italic> <sub><italic>i</italic></sub>, and therefore, the dynamics is governed by the reduced system <disp-formula-group id="M0005"><disp-formula><graphic xlink:href="tjbd-8-099-e005.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> The initial condition of system (5) is assumed to be given as <disp-formula-group id="M0006"><disp-formula><graphic xlink:href="tjbd-8-099-e006.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> For system (5) with properties (<inline-formula><inline-graphic xlink:href="tjbd-8-099-m049.jpg"/></inline-formula>), the existence, uniqueness and continuity of solutions follow from the theory for integro-differential equations in [<xref rid="CIT0030" ref-type="bibr">30</xref>]. Moreover, it can be verified that every solution of Equation (5) with non-negative initial data remains non-negative. It follows from Equation (4) that the set <disp-formula-group id="M0007"><disp-formula><graphic xlink:href="tjbd-8-099-e007.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> is the subset of Ω. Thus, <disp-formula id="UM0008"><graphic xlink:href="tjbd-8-099-u008.jpg" position="float" orientation="portrait"/></disp-formula> is also the subset of Ω<sub>0</sub>. Furthermore, it is a forward invariant compact absorbing set with respect to the system (5). All positive semi-orbits in Γ<sub>0</sub> are precompact ℝ<sup>2<italic>n</italic></sup> and thus have non-empty ω-limit sets.</p><p>System (5) always admits a disease-free equilibrium (DFE), <inline-formula><inline-graphic xlink:href="tjbd-8-099-m050.jpg"/></inline-formula> in Γ<sub>0</sub>. Let <disp-formula id="UM0009"><graphic xlink:href="tjbd-8-099-u009.jpg" position="float" orientation="portrait"/></disp-formula> Clearly, <italic>P˜</italic> <sub><italic>i</italic></sub> are the average time that recovered individuals remain in the recovered class before relapsing. By the properties of <italic>P</italic> <sub><italic>i</italic></sub>, one knows that <disp-formula id="UM0010"><graphic xlink:href="tjbd-8-099-u010.jpg" position="float" orientation="portrait"/></disp-formula> Actually, <inline-formula><inline-graphic xlink:href="tjbd-8-099-m051.jpg"/></inline-formula> are the probability that recovered individuals will die during the recovery period. Hence, <italic>Q</italic> <sub><italic>i</italic></sub> represent the proportion of the recovered individuals who could survive the recovery period, where <disp-formula id="UM0011"><graphic xlink:href="tjbd-8-099-u011.jpg" position="float" orientation="portrait"/></disp-formula> Define <disp-formula id="UM0012"><graphic xlink:href="tjbd-8-099-u012.jpg" position="float" orientation="portrait"/></disp-formula> then <italic>J</italic> <sub><italic>i</italic></sub>(<italic>t</italic>)≥0, ∀ <italic>t</italic>>0, and <inline-formula><inline-graphic xlink:href="tjbd-8-099-m052.jpg"/></inline-formula>.</p><p>The basic reproduction number <inline-formula><inline-graphic xlink:href="tjbd-8-099-m053.jpg"/></inline-formula> is defined as the expected number of secondary cases produced in an entirely susceptible population by a typical infected individual during their entire infectious period [<xref rid="CIT0008" ref-type="bibr">8</xref>,<xref rid="CIT0009" ref-type="bibr">9</xref>]. For system (5), we can calculate it as the spectral radius of a matrix called the next generation matrix. Let <disp-formula id="UM0013"><graphic xlink:href="tjbd-8-099-u013.jpg" position="float" orientation="portrait"/></disp-formula> then the next generation matrix is <disp-formula id="UM0014"><graphic xlink:href="tjbd-8-099-u014.jpg" position="float" orientation="portrait"/></disp-formula> and hence the basic reproduction number of model (5) is calculated by the spectral radius of the next generation matrix <disp-formula id="UM0015"><graphic xlink:href="tjbd-8-099-u015.jpg" position="float" orientation="portrait"/></disp-formula> where ρ(·) denotes the spectral radius of matrix. It is well known that <inline-formula><inline-graphic xlink:href="tjbd-8-099-m054.jpg"/></inline-formula>. Thus <disp-formula-group id="M0008"><disp-formula><graphic xlink:href="tjbd-8-099-e008.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula id="UM0016"><graphic xlink:href="tjbd-8-099-u016.jpg" position="float" orientation="portrait"/></disp-formula> Since it can be verified that system (5) satisfies conditions <inline-formula><inline-graphic xlink:href="tjbd-8-099-m055.jpg"/></inline-formula>–<inline-formula><inline-graphic xlink:href="tjbd-8-099-m056.jpg"/></inline-formula> of [9, Theorem 2], we have the following proposition.</p><statement><label>Proposition 1 </label><p>For system (5), the DFE <italic>P</italic> <sub>0</sub> is locally asymptotically stable if <inline-formula><inline-graphic xlink:href="tjbd-8-099-m057.jpg"/></inline-formula> while it is unstable if <inline-formula><inline-graphic xlink:href="tjbd-8-099-m058.jpg"/></inline-formula>.</p></statement><p>Note that Equation (5) may not have an EE for finite time <italic>t</italic>. It follows from [<xref rid="CIT0030" ref-type="bibr">30</xref>] that if Equation (5) has an EE, then the EE must satisfy the limiting system given by <disp-formula-group id="M0009"><disp-formula><graphic xlink:href="tjbd-8-099-e009.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Since the limiting system (9) contains an infinite delay, its associated initial condition needs to be restricted in an appropriate fading memory space. For any <inline-formula><inline-graphic xlink:href="tjbd-8-099-m059.jpg"/></inline-formula>, define the following Banach space of fading memory type (see e.g. [<xref rid="CIT0015" ref-type="bibr">15</xref>,<xref rid="CIT0017" ref-type="bibr">17</xref>] and references therein) <disp-formula-group id="M0010"><disp-formula><graphic xlink:href="tjbd-8-099-e010.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> with norm <inline-formula><inline-graphic xlink:href="tjbd-8-099-m060.jpg"/></inline-formula>. Let φ∈<italic>C</italic> <sub><italic>i</italic></sub> be such that <inline-formula><inline-graphic xlink:href="tjbd-8-099-m061.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-099-m062.jpg"/></inline-formula>. Let <inline-formula><inline-graphic xlink:href="tjbd-8-099-m063.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-099-m064.jpg"/></inline-formula> such that <inline-formula><inline-graphic xlink:href="tjbd-8-099-m065.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-099-m066.jpg"/></inline-formula>. It turns to consider solutions of system (9), <inline-formula><inline-graphic xlink:href="tjbd-8-099-m067.jpg"/></inline-formula>, with initial conditions <disp-formula-group id="M0011"><disp-formula><graphic xlink:href="tjbd-8-099-e011.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Standard theory of functional differential equations [<xref rid="CIT0017" ref-type="bibr">17</xref>] implies <italic>I</italic> <sub><italic>it</italic></sub>∈<italic>C</italic> <sub><italic>i</italic></sub> for <italic>t</italic>>0. Thus, we will consider system (9) in the phase space <disp-formula id="UM0017"><graphic xlink:href="tjbd-8-099-u017.jpg" position="float" orientation="portrait"/></disp-formula> It can be verified that solutions of Equation (9) in Δ with initial conditions (11) remain non-negative.</p><p>An equilibrium <inline-formula><inline-graphic xlink:href="tjbd-8-099-m068.jpg"/></inline-formula> in the interior of Γ<sub>0</sub> is called an EE, where <inline-formula><inline-graphic xlink:href="tjbd-8-099-m069.jpg"/></inline-formula> satisfy the equilibrium equations <disp-formula-group id="M0012"><disp-formula><graphic xlink:href="tjbd-8-099-e012.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0013"><disp-formula><graphic xlink:href="tjbd-8-099-e013.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p></sec><sec id="S003"><label>3. </label><title>Global stability results</title><p>The global dynamical behaviour of system (5) and (9) is completely established in the following results. Denote <disp-formula-group id="M0014"><disp-formula><graphic xlink:href="tjbd-8-099-e014.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Obviously, <inline-formula><inline-graphic xlink:href="tjbd-8-099-m070.jpg"/></inline-formula> attains its strict global minimum at <italic>z</italic>=1 and <italic>H</italic>(1)=0.</p><sec id="S003-S2001"><label>3.1. </label><title>Global dynamics of DFE</title><statement><label>Theorem 3.1 </label><p>Assume that the functions <italic>f</italic> <sub><italic>ij</italic></sub> satisfy <inline-formula><inline-graphic xlink:href="tjbd-8-099-m071.jpg"/></inline-formula>–<inline-formula><inline-graphic xlink:href="tjbd-8-099-m072.jpg"/></inline-formula>, and the matrix <inline-formula><inline-graphic xlink:href="tjbd-8-099-m073.jpg"/></inline-formula> is irreducible. The following results hold for system (5) with <inline-formula><inline-graphic xlink:href="tjbd-8-099-m074.jpg"/></inline-formula> given in Equation (8). <list list-type="simple"><list-item><p>(i) If <inline-formula><inline-graphic xlink:href="tjbd-8-099-m075.jpg"/></inline-formula> then the DFE of system (5) is globally asymptotically stable in Γ and there does not exist any EE.</p></list-item><list-item><p>(ii) If <inline-formula><inline-graphic xlink:href="tjbd-8-099-m076.jpg"/></inline-formula> then the DFE is unstable and system (5) is uniformly persistent in Γ.</p></list-item></list> </p></statement><p> <italic>Proof</italic> Let us first define matrix value function <disp-formula id="UM0018"><graphic xlink:href="tjbd-8-099-u018.jpg" position="float" orientation="portrait"/></disp-formula> where <inline-formula><inline-graphic xlink:href="tjbd-8-099-m077.jpg"/></inline-formula>. Note that <italic>M</italic>(<italic>S</italic> <sub>0</sub>)=<italic>M</italic> <sup>0</sup>. Since <inline-formula><inline-graphic xlink:href="tjbd-8-099-m078.jpg"/></inline-formula> is irreducible, the matrix <italic>M</italic> <sup>0</sup> is also irreducible.</p><p>We first claim that there does not exist any EE <italic>P</italic>* in Γ. If we assume that <italic>S</italic>≠<italic>S</italic> <sub>0</sub>, it follows that 0<<italic>M</italic>(<italic>S</italic>)<<italic>M</italic> <sup>0</sup>. Since non-negative matrix <italic>M</italic>(<italic>S</italic>)+<italic>M</italic> <sup>0</sup> is irreducible, it follows from the Perron–Frobenius theorem [4, Corollary 2.1.5] that <inline-formula><inline-graphic xlink:href="tjbd-8-099-m079.jpg"/></inline-formula>. This implies that equation <italic>M</italic>(<italic>S</italic>)<italic>I</italic>=<italic>I</italic> has only the trivial solution <italic>I</italic>=0, where <inline-formula><inline-graphic xlink:href="tjbd-8-099-m080.jpg"/></inline-formula>. Hence the claim is true. Next we claim that the DFE <italic>P</italic> <sub>0</sub> is globally asymptotically stable in Γ. From the Perron–Frobenius theorem [4, Theorem 2.1.4], we have that the non-negative irreducible matrix <italic>M</italic> <sup>0</sup> has a strictly positive left eigenvector <inline-formula><inline-graphic xlink:href="tjbd-8-099-m081.jpg"/></inline-formula> associated with the eigenvalue ρ(<italic>M</italic> <sup>0</sup>) such that <disp-formula id="UM0019"><graphic xlink:href="tjbd-8-099-u019.jpg" position="float" orientation="portrait"/></disp-formula> Let <disp-formula id="UM0020"><graphic xlink:href="tjbd-8-099-u020.jpg" position="float" orientation="portrait"/></disp-formula> Consider a Lyapunov functional <disp-formula id="UM0021"><graphic xlink:href="tjbd-8-099-u021.jpg" position="float" orientation="portrait"/></disp-formula> where <italic>U</italic> <sub>+</sub> is given as <inline-formula><inline-graphic xlink:href="tjbd-8-099-m082.jpg"/></inline-formula>. It is easy to know that <inline-formula><inline-graphic xlink:href="tjbd-8-099-m083.jpg"/></inline-formula> with equality if and only if <italic>I</italic> <sub><italic>i</italic></sub>(<italic>t</italic>)=0 and <inline-formula><inline-graphic xlink:href="tjbd-8-099-m084.jpg"/></inline-formula> for almost all ξ≥0. Differentiating <italic>U</italic> <sub>+</sub> along the solutions of system (5) and using integration by parts, we obtain <disp-formula id="UM0022"><graphic xlink:href="tjbd-8-099-u022.jpg" position="float" orientation="portrait"/></disp-formula> Thus, the derivative along the trajectories of system (5) is <disp-formula-group id="M0015"><disp-formula><graphic xlink:href="tjbd-8-099-e015.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Where <italic>E</italic> <sub><italic>n</italic></sub> denote the <italic>n</italic>×<italic>n</italic> identity matrix, <inline-formula><inline-graphic xlink:href="tjbd-8-099-m085.jpg"/></inline-formula>. Let <disp-formula id="UM0023"><graphic xlink:href="tjbd-8-099-u023.jpg" position="float" orientation="portrait"/></disp-formula> and <italic>Z</italic> be the largest compact invariant set in <italic>Y</italic>. We will show <inline-formula><inline-graphic xlink:href="tjbd-8-099-m086.jpg"/></inline-formula>. From inequality (15) and <italic>c</italic> <sub><italic>i</italic></sub>>0, <inline-formula><inline-graphic xlink:href="tjbd-8-099-m087.jpg"/></inline-formula> if and only if <italic>I</italic>=0. Then, by irreducibility of <italic>B</italic>, for each <italic>j</italic>, there exists <italic>i</italic>≠<italic>j</italic> such that <inline-formula><inline-graphic xlink:href="tjbd-8-099-m088.jpg"/></inline-formula>, thus <italic>I</italic> <sub><italic>j</italic></sub>(<italic>t</italic>)=0, <italic>j</italic>=1, 2, … , <italic>n</italic>. Therefore, <inline-formula><inline-graphic xlink:href="tjbd-8-099-m089.jpg"/></inline-formula>, which implies that the compact invariant subset of the set where <inline-formula><inline-graphic xlink:href="tjbd-8-099-m090.jpg"/></inline-formula> is only the singleton <italic>P</italic> <sub>0</sub>. Using Lemma 2.1 of [<xref rid="CIT0034" ref-type="bibr">34</xref>] and the LaSalle–Lyapunov theorem (see [21, Theorem 3.4.7] or [15, Theorem 5.3.1]), we conclude that <inline-formula><inline-graphic xlink:href="tjbd-8-099-m091.jpg"/></inline-formula> globally attracts all the solutions of model (5) if <italic>R</italic> <sub>0</sub>≤1. If <italic>R</italic> <sub>0</sub>>1 and <italic>I</italic>(<italic>t</italic>)≠0, it follows that <disp-formula id="UM0024"><graphic xlink:href="tjbd-8-099-u024.jpg" position="float" orientation="portrait"/></disp-formula> which implies that, in a sufficiently small enough neighbourhood of <inline-formula><inline-graphic xlink:href="tjbd-8-099-m092.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-099-m093.jpg"/></inline-formula>. Therefore, <inline-formula><inline-graphic xlink:href="tjbd-8-099-m094.jpg"/></inline-formula> is unstable when <italic>R</italic> <sub>0</sub>>1. Using a uniform persistence result from [<xref rid="CIT0026" ref-type="bibr">26</xref>,<xref rid="CIT0036" ref-type="bibr">36</xref>] and a similar argument as in [<xref rid="CIT0027" ref-type="bibr">27</xref>] and the proof of Theorem 3.2 of [<xref rid="CIT0034" ref-type="bibr">34</xref>], we can show that the instability of <italic>P</italic> <sub>0</sub> implies the uniform persistence of system (5) when <italic>R</italic> <sub>0</sub>>1. This completes the proof of Theorem 3.1.</p><p>Next we show that the EE <italic>P</italic>* of system (9) is unique and globally asymptotically stable when <inline-formula><inline-graphic xlink:href="tjbd-8-099-m095.jpg"/></inline-formula>. Summarizing the statements, uniform persistence of Equation (9) from Theorem 3.1, together with uniform boundedness of solutions in the interior of Γ, implies system (9) admits at least one EE [5, Theorem 2.8.6].</p><p>Let <inline-formula><inline-graphic xlink:href="tjbd-8-099-m096.jpg"/></inline-formula> be a solution of Equation (9). By Theorem 3.1 and using similar arguments to [<xref rid="CIT0027" ref-type="bibr">27</xref>], it follows that the ω-limit set <italic>W</italic> of <italic>X</italic> is non-empty, compact and invariant and that <italic>W</italic> is the union of orbits of Equation (9).</p></sec><sec id="S003-S2002"><label>3.2. </label><title>Global dynamics of EE</title><p>To get the global stability of <italic>P</italic>, we make the following assumptions [<xref rid="CIT0031" ref-type="bibr">31</xref>]: <list list-type="simple"><list-item><p>(<inline-formula><inline-graphic xlink:href="tjbd-8-099-m097.jpg"/></inline-formula>): <inline-formula><inline-graphic xlink:href="tjbd-8-099-m098.jpg"/></inline-formula>, for <italic>S</italic> <sub><italic>i</italic></sub>≥0.</p></list-item><list-item><p>(<bold>A</bold> <sub>6</sub>) For <inline-formula><inline-graphic xlink:href="tjbd-8-099-m099.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-099-m100.jpg"/></inline-formula> </p></list-item><list-item><p>(<bold>A</bold> <sub>7</sub>) For <italic>S</italic> <sub><italic>i</italic></sub>, <italic>I</italic> <sub><italic>j</italic></sub>>0, <disp-formula id="UM0025"><graphic xlink:href="tjbd-8-099-u025.jpg" position="float" orientation="portrait"/></disp-formula> </p></list-item></list> </p><p>For convenience of notations, set <disp-formula id="UM0026"><graphic xlink:href="tjbd-8-099-u026.jpg" position="float" orientation="portrait"/></disp-formula> and <disp-formula id="UM0027"><graphic xlink:href="tjbd-8-099-u027.jpg" position="float" orientation="portrait"/></disp-formula> Then, <inline-formula><inline-graphic xlink:href="tjbd-8-099-m101.jpg"/></inline-formula> is also irreducible. One knows that the solution space of the linear system <disp-formula id="UM0028"><graphic xlink:href="tjbd-8-099-u028.jpg" position="float" orientation="portrait"/></disp-formula> has dimension 1 and <disp-formula id="UM0029"><graphic xlink:href="tjbd-8-099-u029.jpg" position="float" orientation="portrait"/></disp-formula> gives a base of this space where <inline-formula><inline-graphic xlink:href="tjbd-8-099-m102.jpg"/></inline-formula>, is the co-factor of the <italic>i</italic>th diagonal entry of <inline-formula><inline-graphic xlink:href="tjbd-8-099-m103.jpg"/></inline-formula>.</p><statement><label>Theorem 3.2 </label><p>Consider system (9). Suppose that <inline-formula><inline-graphic xlink:href="tjbd-8-099-m104.jpg"/></inline-formula> and functions <italic>f</italic> <sub><italic>ij</italic></sub> and ϕ<sub><italic>i</italic></sub> satisfy <inline-formula><inline-graphic xlink:href="tjbd-8-099-m105.jpg"/></inline-formula>–<inline-formula><inline-graphic xlink:href="tjbd-8-099-m106.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-099-m107.jpg"/></inline-formula>–<inline-formula><inline-graphic xlink:href="tjbd-8-099-m108.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-099-m109.jpg"/></inline-formula> is a solution to Equation (9) that lies in Γ<sub>0</sub>, then <disp-formula id="UM0030"><graphic xlink:href="tjbd-8-099-u030.jpg" position="float" orientation="portrait"/></disp-formula> </p></statement><p> <italic>Proof</italic> Let <inline-formula><inline-graphic xlink:href="tjbd-8-099-m110.jpg"/></inline-formula> denote the unique EE of system (9). Define a Lyapunov functional as <disp-formula id="UM0031"><graphic xlink:href="tjbd-8-099-u031.jpg" position="float" orientation="portrait"/></disp-formula> where <disp-formula id="UM0032"><graphic xlink:href="tjbd-8-099-u032.jpg" position="float" orientation="portrait"/></disp-formula> and <disp-formula id="UM0033"><graphic xlink:href="tjbd-8-099-u033.jpg" position="float" orientation="portrait"/></disp-formula> The definition of the fading memory space implies that <italic>L</italic> <sub>EE</sub> is well-defined, that is, <italic>L</italic> <sub>EE</sub> is bounded for all <italic>t</italic>≥0. It follows from Lemma 2.1 of [<xref rid="CIT0034" ref-type="bibr">34</xref>] and assumptions (<bold>A</bold> <sub>5</sub>)–(<bold>A</bold> <sub>6</sub>) that <inline-formula><inline-graphic xlink:href="tjbd-8-099-m111.jpg"/></inline-formula> with equality if and only if <inline-formula><inline-graphic xlink:href="tjbd-8-099-m112.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-099-m113.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-099-m114.jpg"/></inline-formula> for almost all ξ≥0.</p><p>Differentiating <italic>L</italic> <sub><italic>S</italic></sub> along the solution of system (9) and using equilibrium equations (12) and (13), we obtain <disp-formula-group id="M0016"><disp-formula><graphic xlink:href="tjbd-8-099-e016.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Differentiating <italic>L</italic> <sub><italic>I</italic></sub> along the solution of system (9), we obtain <disp-formula-group id="M0017"><disp-formula><graphic xlink:href="tjbd-8-099-e017.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Differentiating <italic>L</italic> <sub>−</sub> along the solution of system (9) and using integration by parts, we obtain <disp-formula-group id="M0018"><disp-formula><graphic xlink:href="tjbd-8-099-e018.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Combining Equations (16)–(18) yields <disp-formula id="UM0034"><graphic xlink:href="tjbd-8-099-u034.jpg" position="float" orientation="portrait"/></disp-formula> Let <inline-formula><inline-graphic xlink:href="tjbd-8-099-m115.jpg"/></inline-formula> and <disp-formula id="UM0035"><graphic xlink:href="tjbd-8-099-u035.jpg" position="float" orientation="portrait"/></disp-formula> Then, by assumption <inline-formula><inline-graphic xlink:href="tjbd-8-099-m116.jpg"/></inline-formula>, <disp-formula-group id="M0019"><disp-formula><graphic xlink:href="tjbd-8-099-e019.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Furthermore, under <inline-formula><inline-graphic xlink:href="tjbd-8-099-m117.jpg"/></inline-formula>, we have <disp-formula-group id="M0020"><disp-formula><graphic xlink:href="tjbd-8-099-e020.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <inline-formula><inline-graphic xlink:href="tjbd-8-099-m118.jpg"/></inline-formula> Obviously, the equalities in Equations (19) and (20) hold if and only if <disp-formula id="UM0036"><graphic xlink:href="tjbd-8-099-u036.jpg" position="float" orientation="portrait"/></disp-formula> and <disp-formula id="UM0037"><graphic xlink:href="tjbd-8-099-u037.jpg" position="float" orientation="portrait"/></disp-formula> i.e. <inline-formula><inline-graphic xlink:href="tjbd-8-099-m119.jpg"/></inline-formula>. We can show that <italic>F</italic> <sub><italic>ij</italic></sub> and <inline-formula><inline-graphic xlink:href="tjbd-8-099-m120.jpg"/></inline-formula> satisfy the assumptions of Theorem 3.1 and Corollary 3.3 in [<xref rid="CIT0022" ref-type="bibr">22</xref>]. Therefore, the function <inline-formula><inline-graphic xlink:href="tjbd-8-099-m121.jpg"/></inline-formula> is a Lyapunov function for system (9), namely, <inline-formula><inline-graphic xlink:href="tjbd-8-099-m122.jpg"/></inline-formula> for all <inline-formula><inline-graphic xlink:href="tjbd-8-099-m123.jpg"/></inline-formula>. One can only show that the largest invariant subset where <inline-formula><inline-graphic xlink:href="tjbd-8-099-m124.jpg"/></inline-formula> is the singleton <italic>P</italic> using the same argument as in [<xref rid="CIT0022" ref-type="bibr">22</xref>]. By LaSalle's invariance principle, <italic>P</italic>* is globally asymptotically stable in Γ<sub>0</sub>. This completes the proof of Theorem 3.2.</p></sec></sec><sec id="S004"><label>4. </label><title>Special case for <italic>n</italic>=1</title><p>When <italic>n</italic>=1, system (5) reduces to a single-group SIR model with general relapse distribution and nonlinear incidence rate: <disp-formula-group id="M0021"><disp-formula><graphic xlink:href="tjbd-8-099-e021.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <inline-formula><inline-graphic xlink:href="tjbd-8-099-m125.jpg"/></inline-formula> will reduce to <inline-formula><inline-graphic xlink:href="tjbd-8-099-m126.jpg"/></inline-formula>, where <inline-formula><inline-graphic xlink:href="tjbd-8-099-m127.jpg"/></inline-formula>. Following the method of constructing Lyapunov functionals, <disp-formula-group id="M0022"><disp-formula><graphic xlink:href="tjbd-8-099-e022.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <italic>H</italic> is given in Equation (14), and <disp-formula id="UM0038"><graphic xlink:href="tjbd-8-099-u038.jpg" position="float" orientation="portrait"/></disp-formula> One can determine the global dynamics of single-group SIR model (21). Applying Theorems 3.1 and 3.2, we obtain the following global dynamics but omit the proof.</p><statement><label>Theorem 4.1 </label><p>Under the simpler version of assumptions <inline-formula><inline-graphic xlink:href="tjbd-8-099-m128.jpg"/></inline-formula>–<inline-formula><inline-graphic xlink:href="tjbd-8-099-m129.jpg"/></inline-formula> on functions ϕ and <italic>f</italic>, let (<italic>S</italic>(<italic>t</italic>), <italic>I</italic>(<italic>t</italic>)) be a solution to Equation (21). If <italic>R</italic> <sub>0</sub>>1, then <inline-formula><inline-graphic xlink:href="tjbd-8-099-m130.jpg"/></inline-formula> if <italic>R</italic> <sub>0</sub>≤1, then <inline-formula><inline-graphic xlink:href="tjbd-8-099-m131.jpg"/></inline-formula>.</p></statement><p> <italic>Remark 1</italic> The results in Theorem 4.1 improve the corresponding results in [<xref rid="CIT0026" ref-type="bibr">26</xref>], which gives part of the proof for this problem when <inline-formula><inline-graphic xlink:href="tjbd-8-099-m132.jpg"/></inline-formula> in system (9).</p></sec><sec id="S005"><label>5. </label><title>Numerical simulation</title><p>In this section, we carry out numerical simulation to illustrate and support our analytical results. We consider a simpler case in which all groups share the same natural death rate: <italic>d</italic> <sub><italic>i</italic></sub>=<italic>d</italic> for <italic>i</italic>=1, 2, … , <italic>n</italic>. Further, we denote <inline-formula><inline-graphic xlink:href="tjbd-8-099-m133.jpg"/></inline-formula> and assume that the functions <italic>g</italic> <sub><italic>i</italic></sub>(ξ) are disease-specific only, implying that <inline-formula><inline-graphic xlink:href="tjbd-8-099-m134.jpg"/></inline-formula> for <italic>i</italic>=1, 2, … , <italic>n</italic>. We choose the gamma distribution: <disp-formula id="UM0039"><graphic xlink:href="tjbd-8-099-u039.jpg" position="float" orientation="portrait"/></disp-formula> where <italic>b</italic>>0 is a real number and <italic>n</italic>>1 is an integer. This is widely used and can approximate several frequently used distributions. For example, when <italic>b</italic>→0<sup>+</sup>, <italic>h</italic> <sub><italic>n, b</italic></sub>(<italic>s</italic>) will approach the Dirac delta function, and when <italic>n</italic>=1, <italic>h</italic> <sub><italic>n, b</italic></sub>(<italic>s</italic>) is an exponentially decaying function. Following the technique and method in [<xref rid="CIT0035" ref-type="bibr">35</xref>], define <disp-formula id="UM0040"><graphic xlink:href="tjbd-8-099-u040.jpg" position="float" orientation="portrait"/></disp-formula> which can absorb the exponential term <inline-formula><inline-graphic xlink:href="tjbd-8-099-m135.jpg"/></inline-formula> into the delay kernel. The second equation in Equation (5) can be rewritten as <disp-formula-group id="M0023"><disp-formula><graphic xlink:href="tjbd-8-099-e023.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> For <italic>l</italic>=1, … , <italic>n</italic>, let <disp-formula id="UM0041"><graphic xlink:href="tjbd-8-099-u041.jpg" position="float" orientation="portrait"/></disp-formula> Thus, for <inline-formula><inline-graphic xlink:href="tjbd-8-099-m136.jpg"/></inline-formula>, we obtain <disp-formula id="UM0042"><graphic xlink:href="tjbd-8-099-u042.jpg" position="float" orientation="portrait"/></disp-formula> For <italic>l</italic>=1, we have <disp-formula id="UM0043"><graphic xlink:href="tjbd-8-099-u043.jpg" position="float" orientation="portrait"/></disp-formula> It follows that <disp-formula id="UM0044"><graphic xlink:href="tjbd-8-099-u044.jpg" position="float" orientation="portrait"/></disp-formula> Thus, the integro-differential system (5) is equivalent to the ordinary differential equations <disp-formula-group id="M0024"><disp-formula><graphic xlink:href="tjbd-8-099-e024.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Consider the system (24) for the case <disp-formula id="UM0045"><graphic xlink:href="tjbd-8-099-u045.jpg" position="float" orientation="portrait"/></disp-formula> One then has a two-group model as follows: <disp-formula-group id="M0025"><disp-formula><graphic xlink:href="tjbd-8-099-e025.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> We adopt the following values for the parameters <disp-formula id="UM0046"><graphic xlink:href="tjbd-8-099-u046.jpg" position="float" orientation="portrait"/></disp-formula> taken from the model parameters for herpes simplex virus type 2 presented by Blower <italic>et al.</italic> [<xref rid="CIT0006" ref-type="bibr">6</xref>] and van den Driessche and Zou [<xref rid="CIT0010" ref-type="bibr">10</xref>]. Moreover, we choose parameters from [<xref rid="CIT0006" ref-type="bibr">6</xref>] as <inline-formula><inline-graphic xlink:href="tjbd-8-099-m137.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-099-m138.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-099-m139.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-099-m140.jpg"/></inline-formula>. We can compute <inline-formula><inline-graphic xlink:href="tjbd-8-099-m141.jpg"/></inline-formula>, hence <italic>P</italic> <sub>0</sub>=(3, 0, 0, 0, 3, 0, 0, 0) is the unique equilibrium of system (25) and it is globally stable from Theorem 3.1 (<xref rid="F0002" ref-type="fig">Figure 2</xref>). For Theorem 3.2, we take the following parameters from [<xref rid="CIT0006" ref-type="bibr">6</xref>,<xref rid="CIT0010" ref-type="bibr">10</xref>]: 1/<italic>d</italic>=20, <inline-formula><inline-graphic xlink:href="tjbd-8-099-m142.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-099-m143.jpg"/></inline-formula>. We can compute <inline-formula><inline-graphic xlink:href="tjbd-8-099-m144.jpg"/></inline-formula>, hence <disp-formula id="UM0047"><graphic xlink:href="tjbd-8-099-u047.jpg" position="float" orientation="portrait"/></disp-formula> is the unique equilibrium of system (25) and is globally stable from Theorem 3.2 (<xref rid="F0003" ref-type="fig">Figure 3</xref>). <fig id="F0002" orientation="portrait" position="float"><label>Fig. 2. </label><caption><p>Numerical simulation of Equation (25) with <italic>R</italic> <sub>0</sub>=0.144355<1, hence <italic>P</italic> <sub>0</sub>=(3, 0, 0, 0, 3, 0, 0, 0) is globally stable. Graphs (a) and (b) illustrate that <italic>S</italic> <sub>1</sub>(<italic>t</italic>) and <italic>S</italic> <sub>2</sub>(<italic>t</italic>) will eventually lead to steady state. Graphs (c) and (d) illustrate that <italic>I</italic> <sub>1</sub>(<italic>t</italic>) and <italic>I</italic> <sub>2</sub>(<italic>t</italic>) will eventually lead to zero. Initial value is <italic>S</italic> <sub>1</sub>(0)=6, <italic>S</italic> <sub>2</sub>(0)=4, <italic>y</italic> <sub>1, 1</sub>(0)=0.1, <italic>y</italic> <sub>1, 2</sub>(0)=0.1, <italic>y</italic> <sub>2, 1</sub>(0)=0.1, <italic>y</italic> <sub>2, 2</sub>(0)=0.1, <italic>I</italic> <sub>1</sub>(0)=2, <italic>I</italic> <sub>2</sub>(0)=1.</p></caption><graphic xlink:href="tjbd-8-099-g002"/></fig> <fig id="F0003" orientation="portrait" position="float"><label>Fig. 3. </label><caption><p>Numerical simulation of Equation (25) with <italic>R</italic> <sub>0</sub>=1.13422>1, hence <italic>P</italic>=(0.621928, 3.82371, 0.377123, 0.377123, 0.621928, 3.82371, 0.377123, 0.377123) is globally stable. Graphs (a) and (b) illustrate that <italic>S</italic> <sub>1</sub>(<italic>t</italic>) and <italic>S</italic> <sub>2</sub>(<italic>t</italic>) will eventually lead to steady state. Graphs (c) and (d) illustrate that <italic>I</italic> <sub>1</sub>(<italic>t</italic>) and <italic>I</italic> <sub>2</sub>(<italic>t</italic>) will eventually lead to steady state. Initial value is <italic>S</italic> <sub>1</sub>(0)=6, <italic>S</italic> <sub>2</sub>(0)=4, <italic>y</italic> <sub>1, 1</sub>(0)=0.1, <italic>y</italic> <sub>1, 2</sub>(0)=0.1, <italic>y</italic> <sub>2, 1</sub>(0)=0.1, <italic>y</italic> <sub>2, 2</sub>(0)=0.1, <italic>I</italic> <sub>1</sub>(0)=2, <italic>I</italic> <sub>2</sub>(0)=1.</p></caption><graphic xlink:href="tjbd-8-099-g003"/></fig> </p></sec><sec id="S006"><label>6. </label><title>Summary and discussion</title><p>We present a complete mathematical analysis for global asymptotic stability of unique equilibrium <italic>P</italic> of system (9); this complements our earlier work [<xref rid="CIT0034" ref-type="bibr">34</xref>], where we assumed <inline-formula><inline-graphic xlink:href="tjbd-8-099-m145.jpg"/></inline-formula>.</p><p>Theorems 3.1 and 3.2 are also applicable to system (5) with other special nonlinear incidence rates appearing in the literature including <inline-formula><inline-graphic xlink:href="tjbd-8-099-m146.jpg"/></inline-formula> [<xref rid="CIT0012" ref-type="bibr">12</xref>,<xref rid="CIT0013" ref-type="bibr">13</xref>,<xref rid="CIT0019" ref-type="bibr">19</xref>]; <inline-formula><inline-graphic xlink:href="tjbd-8-099-m147.jpg"/></inline-formula> [<xref rid="CIT0001" ref-type="bibr">1</xref>,<xref rid="CIT0011" ref-type="bibr">11</xref>]; and <inline-formula><inline-graphic xlink:href="tjbd-8-099-m148.jpg"/></inline-formula> [<xref rid="CIT0024" ref-type="bibr">24</xref>,<xref rid="CIT0025" ref-type="bibr">25</xref>,<xref rid="CIT0037" ref-type="bibr">37</xref>].</p><p>The new model includes many existing ones as special cases. For <italic>n</italic>=1, system (5) reduces to a single-group SIR model with general relapse distribution and nonlinear incidence rate. Biologically, Theorems 3.1 and 3.2 imply that, if the basic reproduction number <inline-formula><inline-graphic xlink:href="tjbd-8-099-m149.jpg"/></inline-formula>, then the disease always dies out from all groups; if <inline-formula><inline-graphic xlink:href="tjbd-8-099-m150.jpg"/></inline-formula>, then the disease always persists in all groups at the unique EE level, irrespective of the initial conditions. On the other hand, Theorems 3.1, 3.2 and 4.1 demonstrate that heterogeneity and nonlinear incidence rate do not alter the dynamical behaviour of the SIR model with general relapse distribution and nonlinear incidence rate. Compared to results in [<xref rid="CIT0013" ref-type="bibr">13</xref>,<xref rid="CIT0014" ref-type="bibr">14</xref>], the group structure in system (9) greatly increases the complexity exhibited in the derivatives of the Lyapunov functionals. The key to our analysis is a complete description of the patterns exhibited in the derivative of the Lyapunov functionals using graph theory. Our approach may provide a frame work for dynamics of sexually transmitted diseases with relapse distribution and multi-group structure. Heterogeneity in the host population can result from different behaviours (e.g. numbers of sexual partners for some sexually transmitted infections). The global dynamics exclude the existence of Hopf bifurcation leading to sustained oscillatory solutions.</p><p>We should point here that this work is motivated by Yuan <italic>et al.</italic> [<xref rid="CIT0035" ref-type="bibr">35</xref>] and Sun <italic>et al.</italic> [<xref rid="CIT0031" ref-type="bibr">31</xref>] where disease with latency spreading in a heterogeneous host population and nonlinear incidence of infection and nonlinear removal functions between compartments was considered. In the proof, we utilize a graph-theoretical approach to the method of global Lyapunov functions that is motivated by the works in [<xref rid="CIT0013" ref-type="bibr">13</xref>,<xref rid="CIT0014" ref-type="bibr">14</xref>,<xref rid="CIT0018" ref-type="bibr">18</xref>,<xref rid="CIT0019" ref-type="bibr">19</xref>,<xref rid="CIT0022" ref-type="bibr">22</xref>,<xref rid="CIT0023" ref-type="bibr">23</xref>,<xref rid="CIT0026" ref-type="bibr">26</xref>,<xref rid="CIT0031" ref-type="bibr">31</xref>,<xref rid="CIT0034" ref-type="bibr">34</xref>].</p></sec><sec id="S007"><title>Funding</title><p>J. Wang is supported by National Natural Science Foundation of China (no. 11201128), the Science and Technology Research Project of the Department of Education of Heilongjiang Province (no. 12531495), the Natural Science Foundation of Heilongjiang Province (no. A201211), and the Science and Technology Innovation Team in Higher Education Institutions of Heilongjiang Province. X. Liu is supported by the National Natural Science Foundation of China (11271303).</p></sec>
Expression of inflammation-related genes is associated with adipose tissue location in horses	<sec><title>Background</title><p>In humans, adipose tissue (AT) originating from different depots shows varying gene expression profiles. In horses, the risk of certain metabolic disorders may also be influenced by the impact of specific AT depots. Macrophage infiltration in human and rat AT is considered to be a source of inflammatory changes. In horses, this relationship has not been extensively studied yet. Reverse transcription quantitative real-time polymerase chain reaction (RT-qPCR), a useful method to evaluate differences in mRNA expression across different tissues, can be used to evaluate differences between equine AT depots. For a correct interpretation of the RT-qPCR results, expression data have to be normalized by the use of validated reference genes. The main objectives of this study were to compare mRNA expression of inflammation-related genes, as well as adipocyte morphology and number between different equine AT depots; and in addition, to investigate the presence of antigen presenting cells in equine AT and any potential relationship with adipokine mRNA expression.</p></sec><sec><title>Results</title><p>In this study, the mRNA expression of inflammation-related genes (leptin, chemokine ligand 5, interleukin 1β, interleukin 6, interleukin 10, adiponectin, matrix metalloproteinase 2, and superoxide dismutase 2) and candidate reference gene stability was investigated in 8 different AT depots collected from the nuchal, abdominal (mesenteric, retroperitoneal, and peri-renal) and subcutaneous (tail head and loin) AT region. By using GeNorm analysis, <italic>HPRT1, RPL32,</italic> and <italic>GAPDH</italic> were found to be the most stable genes in equine AT. The mRNA expression of leptin, chemokine ligand 5, interleukin 10, interleukin 1β, adiponectin, and matrix metalloproteinase 2 significantly differed across AT depots (P < 0.05). No significant AT depot effect was found for interleukin 6 and superoxide dismutase 2 (P > 0.05). Adipocyte area and number of antigen presenting cells per adipocyte significantly differed between AT depots (P < 0.05).</p></sec><sec><title>Conclusions</title><p>Adipose tissue location was associated with differences in mRNA expression of inflammation-related genes. This depot-specific difference in mRNA expression suggests that the overall inflammatory status of horses could be partially determined by the relative proportion of the different AT depots.</p></sec>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Bruynsteen</surname><given-names>Lien</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>lien.bruynsteen@ugent.be</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Erkens</surname><given-names>Tim</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>tim.erkens@ugent.be</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Peelman</surname><given-names>Luc J</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>luc.peelman@ugent.be</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Ducatelle</surname><given-names>Richard</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>richard.ducatelle@ugent.be</email></contrib><contrib contrib-type="author" id="A5"><name><surname>Janssens</surname><given-names>Geert PJ</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>geert.janssens@ugent.be</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Harris</surname><given-names>Patricia A</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>pat.harris@effem.com</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Hesta</surname><given-names>Myriam</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>myriam.hesta@ugent.be</email></contrib>	BMC Veterinary Research	<sec><title>Background</title><p>Adipose tissue (AT) can be divided into brown and white AT [<xref ref-type="bibr" rid="B1">1</xref>]. The latter is now recognized as being more than an energy storage site. It is accepted as a highly active metabolic and endocrine organ [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B3">3</xref>] comprising different cell types (adipocytes, pre-adipocytes, endothelial cells, fibroblasts and macrophages) [<xref ref-type="bibr" rid="B4">4</xref>] that actively secrete proteins involved in the regulation of energy, as well as neuroendocrine, autonomic, and immune functions [<xref ref-type="bibr" rid="B5">5</xref>]. These different cell types may contribute to the secretion of the pro-inflammatory cytokines tumor necrosis factor alpha (TNF-α), interleukin 1 (IL-1), interleukin 6 (IL-6), chemokine ligand 5 (CCL5), and anti-inflammatory cytokine interleukin 10 (IL-10), as well as hormones such as resistin, leptin, and adiponectin that are involved in the inflammatory response and insulin sensitivity [<xref ref-type="bibr" rid="B5">5</xref>-<xref ref-type="bibr" rid="B7">7</xref>]<italic>.</italic></p><p>The AT also secretes matrix metalloproteinases (MMPs, e.g. MMP-2 or gelatinase A, MMP-9 or gelatinase B [<xref ref-type="bibr" rid="B8">8</xref>], and MMP-1,3,7, [<xref ref-type="bibr" rid="B9">9</xref>]) which have a functional role in the development of the AT [<xref ref-type="bibr" rid="B10">10</xref>] and are important for the extracellular matrix remodelling, which occurs during obesity-mediated AT formation, at least in mice [<xref ref-type="bibr" rid="B11">11</xref>].</p><p>A study by Fain and colleagues [<xref ref-type="bibr" rid="B12">12</xref>] in obese women revealed that over 90% of the adipokine release by AT, except for adiponectin and leptin, could be attributed to non-fat cells. When excessive amounts of AT are deposited, inflammatory markers in the circulation can rise as result of the adipokine secreting ability of AT. Cinti and colleagues [<xref ref-type="bibr" rid="B13">13</xref>] demonstrated that > 90% of all macrophages in white AT of obese mice and humans were localized around dead adipocytes, forming crown-like structures (CLS). Vick and colleagues [<xref ref-type="bibr" rid="B14">14</xref>] first demonstrated an association between obesity and increased inflammatory markers (TNF-α and IL-1) in horses, although age was also an important and possible confounding factor. Currently, there is some controversy whether obesity in horses is or is not associated with low grade inflammation [<xref ref-type="bibr" rid="B14">14</xref>-<xref ref-type="bibr" rid="B16">16</xref>] and there is no evidence whether CLS do or do not form in the obese horse.</p><p>Adipocyte size is positively correlated with frequency of adipocyte death, macrophage numbers, as well as CLS in visceral and subcutaneous (SC) depots in mice [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B17">17</xref>], and leptin mRNA expression in humans and cattle [<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B19">19</xref>]. To the authors’ knowledge, size of adipocytes originating from different horse AT regions has not been previously reported.</p><p>The specific site of AT deposition is clinically very important. Humans with a higher accumulation of visceral fat are at a higher risk for the development of obesity-related metabolic disorders [<xref ref-type="bibr" rid="B20">20</xref>]. Similarly in horses, it has recently been demonstrated that expression of glucose transporters was influenced by AT location in insulin sensitive and insulin resistant individuals [<xref ref-type="bibr" rid="B21">21</xref>]. It has also been suggested in equidae that AT distributed specifically on the crest of the neck could indicate or contribute to hyperinsulinemia, insulin resistance (IR), and/or an increased risk for laminitis [<xref ref-type="bibr" rid="B22">22</xref>,<xref ref-type="bibr" rid="B23">23</xref>]. Therefore, clinical interest on AT in this region is increased in horses [<xref ref-type="bibr" rid="B22">22</xref>,<xref ref-type="bibr" rid="B24">24</xref>]. Burns and coworkers [<xref ref-type="bibr" rid="B24">24</xref>] found no differences in pro-inflammatory cytokine IL-1β and IL-6 mRNA expression between insulin resistant and insulin sensitive horses. Higher mRNA concentrations of these two cytokines, however, were found in the nuchal ligament AT compared to the other AT depots sampled in that study.</p><p>Our hypothesis was that mRNA expression of inflammation-related genes varied across AT depots. Therefore, the first aim of this study was to compare adipocyte size and mRNA expression between different equine AT depots with special interest in the nuchal AT region. The second aim was to investigate the presence of antigen presenting cells in equine AT and any potential relationship with adipokine mRNA expression.</p></sec><sec><title>Results and discussion</title><sec><title>Animals</title><p>A variety of different breeds was chosen for this study (Table <xref ref-type="table" rid="T1">1</xref>). Average age was 14 ± 7 years.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Information on the horses involved in this study</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="center"><bold>Horse</bold></th><th align="center"><bold>Breed</bold></th><th align="center"><bold>Age (years)</bold></th><th align="center"><bold>Nutritional status</bold></th></tr></thead><tbody><tr><td align="center" valign="bottom">1<hr/></td><td align="center" valign="bottom">German riding horse<hr/></td><td align="center" valign="bottom">16<hr/></td><td align="center" valign="bottom">overweight to obese<hr/></td></tr><tr><td align="center" valign="bottom">2<hr/></td><td align="center" valign="bottom">Dutch riding horse<hr/></td><td align="center" valign="bottom">20<hr/></td><td align="center" valign="bottom">normal<hr/></td></tr><tr><td align="center" valign="bottom">3<hr/></td><td align="center" valign="bottom">Belgian riding horse<hr/></td><td align="center" valign="bottom">1<hr/></td><td align="center" valign="bottom">normal<hr/></td></tr><tr><td align="center" valign="bottom">4<hr/></td><td align="center" valign="bottom">French riding pony, breed Haflinger<hr/></td><td align="center" valign="bottom">17<hr/></td><td align="center" valign="bottom">obese<hr/></td></tr><tr><td align="center" valign="bottom">5<hr/></td><td align="center" valign="bottom">Selle Français<hr/></td><td align="center" valign="bottom">10<hr/></td><td align="center" valign="bottom">normal<hr/></td></tr><tr><td align="center" valign="bottom">6<hr/></td><td align="center" valign="bottom">French Thouroughbred<hr/></td><td align="center" valign="bottom">12<hr/></td><td align="center" valign="bottom">normal<hr/></td></tr><tr><td align="center" valign="bottom">7<hr/></td><td align="center" valign="bottom">Royal Dutch Sport Horse (KWPN)<hr/></td><td align="center" valign="bottom">11<hr/></td><td align="center" valign="bottom">normal<hr/></td></tr><tr><td align="center" valign="bottom">8<hr/></td><td align="center" valign="bottom">Belgian trotter<hr/></td><td align="center" valign="bottom">3<hr/></td><td align="center" valign="bottom">normal<hr/></td></tr><tr><td align="center" valign="bottom">9<hr/></td><td align="center" valign="bottom">Belgian Warmblood (BWP)<hr/></td><td align="center" valign="bottom">21<hr/></td><td align="center" valign="bottom">overweight to obese<hr/></td></tr><tr><td align="center" valign="bottom">10<hr/></td><td align="center" valign="bottom">Dutch riding horse<hr/></td><td align="center" valign="bottom">25<hr/></td><td align="center" valign="bottom">overweight to obese<hr/></td></tr><tr><td align="center" valign="bottom">11<hr/></td><td align="center" valign="bottom">Dutch riding horse<hr/></td><td align="center" valign="bottom">11<hr/></td><td align="center" valign="bottom">overweight to obese<hr/></td></tr><tr><td align="center">12</td><td align="center">French trotter</td><td align="center">15</td><td align="center">overweight to obese</td></tr></tbody></table></table-wrap></sec><sec><title>Blood analysis</title><p>Average glucose, insulin, and leptin levels were 106 ± 16 mg/dl, 5.9 ± 0.9 mU/l, and 3.3 ± 1.4 ng/ml respectively.</p></sec><sec><title>Candidate reference gene selection and GeNorm analysis</title><p>To study depot-related variation in mRNA expression in AT, a very sensitive and specific technique is required, such as RT-qPCR [<xref ref-type="bibr" rid="B25">25</xref>], because of the low yield of mRNA isolated from AT [<xref ref-type="bibr" rid="B26">26</xref>]. However, before comparing mRNA expression profiles across samples, correction for variables such as quality and quantification of the starting material and enzymatic efficiencies must be carried out [<xref ref-type="bibr" rid="B26">26</xref>-<xref ref-type="bibr" rid="B28">28</xref>]. Consequently, the need for accurate data normalization is crucial [<xref ref-type="bibr" rid="B29">29</xref>]. In human AT, obesity and type 2 diabetes can exert a detectable influence on reference gene expression in SC and visceral fat depots [<xref ref-type="bibr" rid="B30">30</xref>]. This demonstrates that the expression level of reference genes is influenced by body region and health status of the test subject.</p><p>The efficiency of each RT-qPCR run was calculated from a relative standard curve based on a 5-point 5-fold cDNA dilution series, and ranged between 93 and 102.5%. Linear correlation coefficients varied between 0.996 and 0.999.</p><p>One sample from the neck region, two from the SC region, and four from the abdominal region showed consistently higher transcription levels (Cq value: the fractional PCR cycle at which the fluorescent signal significantly rises above the background signal [<xref ref-type="bibr" rid="B31">31</xref>]) compared to the other samples. Amplification problems were considered to be the cause as the RNA quantity and quality was comparable to the other samples of the same region (Experion analysis). These 7 AT samples were therefore excluded from further analysis. Transcription levels across all AT studied were almost similar for <italic>ACTB, GAPDH,</italic> and <italic>RPL32</italic>, which had higher Cq values than <italic>HPRT1, SDHA,</italic> and <italic>TUBA4A</italic>. The raw gene expression data from the genes of interest were normalised using the geometric mean of the most stable candidate reference genes <italic>GAPDH, HPRT1,</italic> and <italic>RPL32</italic>.</p></sec><sec><title>Depot-specific mRNA expression</title><p>The present study investigated the AT depot related mRNA expression of inflammation-related genes in horses of different breeds, different ages, and with varying body condition or nutritional status. Adipokine expression was primarily studied at the transcription level, which does not necessarily reflect the protein level and/or its activity. It should be mentioned that AT is made up of multiple cell types [<xref ref-type="bibr" rid="B4">4</xref>]. The aim of mRNA expression was not to examine the individual cell populations, but to consider AT as a whole [<xref ref-type="bibr" rid="B12">12</xref>].</p><p>Leptin mRNA expression was significantly higher in the three neck samples compared to the mesenteric AT samples (Neck (N) ¼, N ½, N ¾: P = 0.034, 0.008, and 0.015 respectively). In contrast, CCL5 and IL-10 showed significantly lower mRNA expression in the nuchal AT compared to the mesenteric AT (N ¼, N ½, N ¾; P = 0.009, 0.009, 0.019 for CCL5, and 0.032, 0.008, 0.009 for IL-10 respectively). A significant lower expression of adiponectin mRNA was found in the tail head AT region compared to the nuchal AT (N ¼, N ½, N ¾: P = 0.010, 0.014, and 0.004 respectively), retroperitoneal (P = 0.003), and peri-renal AT region (P = 0.009). Pro-inflammatory cytokine IL-1β mRNA expression was significantly lower in the loin AT compared to the mesenteric and peri-renal AT (P = 0.004). A trend for lower mRNA expression was found in the retroperitoneal AT (P = 0.074). The MMP2 mRNA expression was significantly lower in the peri-renal region compared to AT originating from N ¼ (P = 0.019), N ½ (P = 0.006), tail head (P = 0.004) and loin region (P = 0.008). Mesenteric AT had a significantly lower MMP2 mRNA expression compared to N ½ (P < 0.001), tail head (P = 0.004) and loin AT (P = 0.001). Retroperitoneal AT had a significantly lower MMP2 mRNA expression compared to the loin AT (P = 0.014). Interleukin 6 tended to have a higher mRNA expression in the N ½ AT compared to mesenteric AT (P = 0.073). No significant AT depot effect was found for superoxide dismutase (SOD) 2.</p><p>A correlation was found between plasma leptin and insulin concentrations (P = 0.035; r = 0.610). There was also a correlation between IL-6 and IL-1β in the nuchal AT region (N ¼, N ½, and N ¾: P = 0.004, 0.001, 0.003; r = 0.756, 0.827, 0.782 respectively) and the tail head region (P = 0.007; r = 0.734). Higher leptin mRNA expression in the nuchal AT region compared to the mesenteric AT suggests that nuchal AT may contribute proportionally more to the overall leptin concentration in the horse. The strong correlation between leptin concentration and degree of IR [<xref ref-type="bibr" rid="B32">32</xref>] supports the hypothesis that enlarged nuchal AT indeed is an important risk factor for IR [<xref ref-type="bibr" rid="B22">22</xref>,<xref ref-type="bibr" rid="B23">23</xref>]. In a study of Liburt and coworkers, decreased IL-6 mRNA in nuchal AT was associated with increased insulin sensitivity [<xref ref-type="bibr" rid="B33">33</xref>]. Higher nuchal AT IL-6 mRNA expression together with a significant correlation between the expression of IL-6 and IL-1β in the nuchal AT depot, could indicate that in horses, the nuchal AT depot is an important contributor for gene expression of these pro-inflammatory markers whereas in humans, the visceral AT is responsible for this [<xref ref-type="bibr" rid="B34">34</xref>,<xref ref-type="bibr" rid="B35">35</xref>]. If such expression leads to increased protein formation, then an increase in the size of nuchal AT depot could potentially contribute more to the total body inflammatory status. In humans, elevated inflammatory cytokines such as TNF-α, IL-6, and IL-1 play important roles in the development of obesity-associated IR [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B36">36</xref>]. If this is also the case in horses, it would further confirm the link between a high cresty neck score and IR.</p><p>Lower adiponectin mRNA expression in the SC region compared to the abdominal and nuchal region suggests that abdominal AT and nuchal AT may be more important for circulating adiponectin concentrations. Differences between gene expression in the different AT depots and the protein levels in the blood can be caused by differences at the translation level, which can be influenced by cytokines. Bruun and colleagues [<xref ref-type="bibr" rid="B37">37</xref>] showed that TNF-α and IL-6 significantly decreased the human adiponectin mRNA levels in vitro suggesting that endogenous cytokines may affect adiponectin. In the present study, however, no correlations between adiponectin gene expression and cytokine mRNA expression were found.</p><p>Chemotactic cytokine CCL5 mediates chemotaxis of different leukocytes, depending on the tissue protein levels. High levels of CCL5 can trigger cytokine release. In humans, CCL5 production is upregulated by inflammatory cytokines, such as IL-1 [<xref ref-type="bibr" rid="B38">38</xref>]. In the present study, CCL5, IL-10, and IL-1β mRNA expression was higher in the abdominal region, although no correlations between these cytokines were found. This could indicate that in horses, other cytokines regulate CCL5 production. Higher mRNA expression of CCL5, IL-10, and IL-1β in the abdominal AT suggests that this AT depot may be more important for the circulating levels of these cytokines. This is in contrast with the findings from Burns and coworkers [<xref ref-type="bibr" rid="B24">24</xref>] who found higher IL-1β mRNA expression in the nuchal ligament AT compared with the other depots sampled in that study. It has been suggested that different reference genes should be tested in each study setup to find the most suitable one not influenced by the experimental treatment [<xref ref-type="bibr" rid="B39">39</xref>]. As candidate reference gene selection was different in the present study (<italic>HPRT1, RPL32,</italic> and <italic>GAPDH)</italic> and the Burns’ study (β<italic>-actin</italic> and β<sub><italic>2</italic></sub><italic>-microglobulin</italic>), it may mean that results from both studies cannot be simply compared.</p><p>In mice, many MMPs are expressed by AT and stromal vascular cells in a depot-specific manner [<xref ref-type="bibr" rid="B40">40</xref>]. Higher MMP-2 mRNA expression in the SC AT suggests that in horses, this AT depot is more stimulated to differentiate pre-adipocytes into adipocytes [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B41">41</xref>] and extracellular matrix remodeling [<xref ref-type="bibr" rid="B11">11</xref>].</p><p>In conclusion, different mRNA expression of inflammation-related genes in different AT depots suggests that AT depots may be a driving force for total body inflammation. It is possible that if more fat is deposited in an AT depot with high mRNA expression levels of pro-inflammatory cytokines, such as nuchal AT with high leptin and pro-inflammatory IL-6 mRNA expression, this may contribute to a greater overall inflammation in that individual horse than if the fat had been deposited in an AT depot with a low mRNA expression of pro-inflammatory cytokines. Further research into the final translation of mRNA expression of inflammation-related genes into adipokines will be necessary to correctly evaluate the impact of fat deposition at specific places in the horse body.</p></sec><sec><title>Histology and immunohistochemistry</title><p>To the authors’ knowledge, adipocyte size and area in cross-section in different AT regions in the horse body have not been previously reported.</p><p>Adipocyte area and number of antigen presenting cells (APC)/adipocyte could not be determined in 10 samples from different AT depots in 6 different horses due to technical cutting and staining difficulties (N ¼ AT for horse number 4; N ¾ AT for horse number 8; tail head AT for horse number 3; mesenteric AT for horse number 3,4,7,8 and right kidney AT for horse number 2,7,9). Negative controls using isotype-matched nonsense antibody showed no staining.</p><p>Average adipocyte size was 70 ± 7 μm, with the largest average adipocyte diameter being found in peri-renal AT (82 ± 14 μm). Average adipocyte area in cross-section was 3980 ± 1355 μm<sup>2</sup>. Peri-renal adipocyte area (5370 ± 1919 μm<sup>2</sup>) was significantly higher compared to N ½ (3116 ± 556 μm<sup>2</sup>; P < 0.001), N ¼ (3195 ± 831 μm<sup>2</sup>, P = 0.003) and tail head adipocyte area (3537 ± 1375 μm<sup>2</sup>, P = 0.022) (Figure <xref ref-type="fig" rid="F1">1</xref>). Retroperitoneal adipocyte area (4795 ± 1610 μm<sup>2</sup>) was significantly higher than N ½ (3116 ± 556 μm<sup>2</sup>; P = 0.020) adipocyte area. A significant lower number of APC/adipocyte was found in the N ½ AT compared to the loin AT (P = 0.024) (Figure <xref ref-type="fig" rid="F2">2</xref>).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Mean adipocyte area of equine adipocytes.</bold> Data are reported as means ± SD. Peri-renal adipocyte area is higher compared to N ½ (P < 0.001), N ¼ (P = 0.003) and loin adipocyte area (P = 0.022). Retroperitoneal adipocyte area is higher than N ½ (P = 0.020) adipocyte area. Superscripts (abc) indicate differences between adipose tissue location (P < 0.05). Missing values and outliers were excluded from this analysis (N3/4 8; N1/4 4; TH 3; M 3,4,7,8; RK 2,7,9; L 11, RP 3). Abbreviations: N ¼, neck ¼; TH, tail head; L, loin; M, mesenteric; RP, retroperitoneal; RK, right kidney.</p></caption><graphic xlink:href="1746-6148-9-240-1"/></fig><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Antigen presenting cells (APC) per adipocyte in equine adipose tissue.</bold> Major Histocompatibility Complex II (MHC II) stain was performed to enable the identification of antigen presenting cells/HPF (APC/HPF) in the different AT depots. Data are reported as means ± SD. A lower amount of APC/adipocyte was found in the N ½ AT compared to the loin AT (0.024). Superscripts (ab) indicate differences between adipose tissue location (P < 0.05). Missing values and outliers were excluded from this analysis (N3/4 8; N1/4 4; TH 3; M 3,4,7,8; RK 2,7,9; L 11, RP 3). Abbreviations: N ¼, neck ¼; L, loin; TH, tail head; M, mesenteric; RP, retroperitoneal; RK, right kidney.</p></caption><graphic xlink:href="1746-6148-9-240-2"/></fig><p>Capping structures (dead adipocytes engulfed by APC) (Figure <xref ref-type="fig" rid="F3">3</xref>), similar to CLS (dead adipocytes surrounded by macrophages [<xref ref-type="bibr" rid="B17">17</xref>]) in mice), were found in 6 horses in 4 different AT depots (N ¼, N ½, loin, and peri-renal). In 2 horses (number 3 and 4) CLS were found in multiple AT depots.</p><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Capping structures in horse adipose tissue.</bold> Major Histocompatibility Complex II (MHC II) stain was performed to enable the identification of antigen presenting cells in the different AT depots. Capping structures in the loin adipose tissue <bold>(a)</bold> and Neck ¼ <bold>(b)</bold> are indicated with black arrows.</p></caption><graphic xlink:href="1746-6148-9-240-3"/></fig><p>Clear differences in adipocyte area between AT depots were found, which is in accordance with the findings in other species [<xref ref-type="bibr" rid="B42">42</xref>-<xref ref-type="bibr" rid="B44">44</xref>]. Differences in adipocyte area can influence mRNA expression [<xref ref-type="bibr" rid="B45">45</xref>-<xref ref-type="bibr" rid="B47">47</xref>]. Significant correlations between mRNA expression of inflammation-related genes and adipocyte area were also found in the present study between multiple AT depots (N ¼, N ¾, tail head, mesenteric, retroperitoneal, and perirenal) and multiple genes (adiponectin, CCL5, SOD, IL-10, leptin, and MMP2), although no clear pattern could be determined.</p><p>In the subcutaneous region, APC/adipocyte was high compared to the nuchal and abdominal region. In humans, this is related to high adipocyte death and CLS formation [<xref ref-type="bibr" rid="B13">13</xref>], but in the current study, this relationship was not found. Macrophage inflammation in AT is also correlated with inflammation in humans [<xref ref-type="bibr" rid="B48">48</xref>,<xref ref-type="bibr" rid="B49">49</xref>].</p><p>This is the first study describing capping structures (adipocytes surrounded by APC) in the horse (Figure <xref ref-type="fig" rid="F3">3</xref>), probably similar to CLS in mice and humans [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B17">17</xref>]. One limitation of the present study was the staining method used, as this labeled MHC II molecules, which are not exclusively expressed on macrophages, but also on other APC such as monocytes, B-cells, and dentritic cells [<xref ref-type="bibr" rid="B50">50</xref>,<xref ref-type="bibr" rid="B51">51</xref>].</p></sec></sec><sec sec-type="conclusions"><title>Conclusions</title><p>Despite the fact that mRNA levels of inflammation-related genes were studied instead of protein levels, still interesting conclusions concerning the deposition of fat in various depots in horses could be drawn. The inflammatory profile in AT clearly varies with its location in the horse’s body in horses of different breeds varying in age and nutritional status, suggesting that the total inflammatory status of the horse may be at least partly a reflection of the relative contribution of each AT. The factors driving the interindividual differences in AT distribution thus warrant further investigation.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Study animals</title><p>Twelve horses, due to be euthanized for non-research purposes, were selected at a local abattoir. Horses were chosen so that : 1) they were all geldings to exclude potential gender-related gene expression [<xref ref-type="bibr" rid="B52">52</xref>]; 2) they represented a variety of different breeds presented at the abattoir (Table <xref ref-type="table" rid="T1">1</xref>); 3) they showed no obvious lameness or overt laminitic rings; 4) they all had healthy appearance; 5) they ranged in their nutritional status (as scored visually by 2 experienced veterinarians) from normal to obese (Table <xref ref-type="table" rid="T1">1</xref>); and 6) they ranged in age from 1–25 years (Table <xref ref-type="table" rid="T1">1</xref>). Horses were euthanized according to the procedure Royal Decree of January 16, 1988 concerning the protection of animals at killing.</p></sec><sec><title>Sample collection</title><p>Blood samples were taken immediately post mortem from the vena jugularis for the analysis of glucose (Vacuette® tube, FX Sodium Fluoride/Potassium oxalate, 2 ml), as well as insulin and leptin (Vacuette® tube, Z Serum Clot Activator, 9 ml). Within 15 minutes after euthanasia, AT samples were collected from the different sites: three samples in the nuchal region taken at ¼, ½, and ¾ of the distance between the poll and the withers; three abdominal samples taken at the right kidney, retroperitoneally 10 cm lateral to the linea alba, and at the mesenterium; and two samples from SC AT taken at the level of the loin and around the tail head. All samples were taken in duplicate. Gene expression samples (thickness 0.4 – 0.5 cm) were immediately submerged in RNA<italic>later</italic> (Sigma-Aldrich, AMBION, Inc., Austin, Texas, USA) for RNA preservation and stored at 4°C for 24 hours and then stored at -20°C until RNA extraction. Histology AT samples were stored in formalin until further processing.</p></sec><sec><title>Blood sample analysis</title><p>Plasma glucose analysis was performed using a spectrophotometric method based on glucose hexokinase [<xref ref-type="bibr" rid="B53">53</xref>] (Architect C16000; Abbott, Abbott Laboratories, Abbott Park, Illinois, USA). Serum insulin concentrations were measured with an immunoradiometric assay test kit [<xref ref-type="bibr" rid="B32">32</xref>] (insulin IRMA Ref 5251, Diasource Europe S.A., Nivelles, Belgium). An implementation validation has been carried out before use in horses. A dilution curve has been designed (100–80 – 60 – 40 – 20 – 0% sample). Theoretical and measured values were compared to evaluate possible matrix-influences. Inter-assay variance was < 4%, intra-assay variance in the high sample% was 9.2%, in the low sample% 1.9%. Leptin was measured using a multispecies RIA kit (Merck Millipore., Billerica, MA 01821, USA), previously validated for use in equine plasma [<xref ref-type="bibr" rid="B54">54</xref>].</p></sec><sec><title>RNA isolation and cDNA synthesis</title><p>Total RNA was isolated using the RNeasy Lipid Tissue Mini Kit (Qiagen®, AMBION, Inc., USA) and the TissueRuptor (Qiagen) for complete sample disruption/homogenization, as described in the manufacturer’s protocol. An on-column DNase digestion (RNase-Free DNase Set, Qiagen) was included and was empirically verified by a minus reverse transcription (RT) control reaction. RNA quantity and purity (OD 260/280 ratio 1.9-2.1) were measured with the ND-1000 spectrophotometer (NanoDrop, NanoDrop Products, Wilmington, USA). The RNA quality was verified on an agarose gel and was assessed with the Ex-on RNA StdSens Analysis Kit (Bio-Rad, Bio-Rad Laboratories N.V., Hercules, USA) on an Experion Automated Electrophoresis System (Bio-Rad). The RNA quality indicator (RQI) for the AT ranged between 7–8.5 and for liver between 9–9.5. Subsequently, the iScript cDNA synthesis kit (Bio-Rad) was used to convert approximately 0.6 μg of total RNA into cDNA, which was verified by a control PCR.</p></sec><sec><title>Quantitative real-time PCR</title><p>All PCR reactions were performed in a 15 μl reaction volume on an iCycler iQ Real-Time PCR Detection System (Bio-Rad) using 7.5 μl of Kapa SYBR Fast Bio-Rad qPCR Master Mix (Sopachem, Kapabiosystems, Woburn, USA) supplemented with 2.5 μl of diluted cDNA. The addition of RNAse free water and primer concentration varied according to the primer used. The qRT-PCR measurements for all samples were performed in duplicate and every run included a no-template control.</p><p>The PCR program started with an initial denaturation at 95°C for 3 minutes to activate the <italic>Taq</italic> polymerase, followed by 40 cycles of denaturation at 95°C for 10 seconds and a combined primer annealing/extension at the primer specific annealing temperature for 30 seconds during which fluorescence was measured. A melting curve was constructed to verify the presence of a single gene-specific amplicon and the absence of any primer dimers by heating the samples from 70 to 95°C in 0.5°C increments with a dwell time at each temperature of 10 seconds while continuously monitoring the fluorescence. The efficiency of each RT-qPCR run was calculated from a relative standard curve based on a 5-point 5-fold cDNA dilution series using pooled cDNA obtained from AT in the neck, loin and tail head region and liver. The RT-qPCR data from all genes were converted to raw data as described in Erkens <italic>et al.</italic>[<xref ref-type="bibr" rid="B31">31</xref>]. Six candidate reference genes were selected based on previous gene expression studies in human [<xref ref-type="bibr" rid="B30">30</xref>,<xref ref-type="bibr" rid="B55">55</xref>-<xref ref-type="bibr" rid="B57">57</xref>] and bovine AT [<xref ref-type="bibr" rid="B58">58</xref>], as well as equine tissues (skin, blastocysts, and lymphocytes) [<xref ref-type="bibr" rid="B59">59</xref>-<xref ref-type="bibr" rid="B61">61</xref>]. Candidate reference gene primers for <italic>ACTB</italic>, <italic>HPRT1</italic>, <italic>RPL32,</italic> and <italic>TUBA4A</italic> were used from Bogaert <italic>et al</italic><italic>.</italic> (2006) [<xref ref-type="bibr" rid="B59">59</xref>]; <italic>GAPDH</italic> and <italic>SDHA</italic> were used from Smits <italic>et al</italic><italic>.</italic>[<xref ref-type="bibr" rid="B60">60</xref>].</p></sec><sec><title>Determination of candidate reference gene expression stability</title><p>Candidate reference gene expression stability was evaluated with the M value of the geNorm algorithm [<xref ref-type="bibr" rid="B62">62</xref>]. The most stable control genes (lowest variation in mRNA expression) have the lowest M value. The raw gene expression data from the genes of interest were then normalised using the geometric mean of the best performing candidate reference genes.</p></sec><sec><title>Primers for the genes of interest</title><p>Primers for the genes of interest <italic>CCL5</italic>, <italic>IL-10</italic>, <italic>IL-1</italic>β<italic>, IL-6,</italic> and <italic>SOD2</italic> were used from Figueiredo <italic>et al.</italic>[<xref ref-type="bibr" rid="B63">63</xref>], and for <italic>MMP2</italic> from Loftus <italic>et al.</italic>[<xref ref-type="bibr" rid="B64">64</xref>]. For <italic>ADIPOQ</italic> and <italic>LEP</italic>, primers were designed using Primer3 [<xref ref-type="bibr" rid="B65">65</xref>]<italic>,</italic> while taking primer specificity (Blast, [<xref ref-type="bibr" rid="B66">66</xref>]) and possible secondary structures (Mfold, [<xref ref-type="bibr" rid="B67">67</xref>]) into account. As for the candidate reference genes, all primer amplicons were sequenced using the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems) on an Applied Biosystems 3730xl DNA Analyser. In addition, gel electrophoresis was performed to check the formation of 1 amplicon of the expected size, and to control the absence of primer dimers.</p></sec><sec><title>Histology</title><p>Adipose tissue samples were fixated by immersion in 4% paraformaldehyde, embedded in paraffin and sectioned. Two five μm thick serial sections were obtained, the first stained by hematoxylin and eosin (HE) to assess morphology (adipocyte area) and the rest processed for immunohistochemistry (see below). Adipocyte numbers were counted in 10 high power fields (HPF) and the average number of adipocytes was calculated per HPF. The surface of 1 HPF (πr<sup>2</sup> = π*250 μm<sup>2</sup> = 196250 μm<sup>2</sup>) was divided through the numbers of adipocytes/HPF to calculate the mean surface area per adipocyte.</p></sec><sec><title>Immunohistochemistry</title><p>The presence of antigen presenting cells (APC) was evaluated by the use of Monoclonal Mouse Anti-Human HLA-DR antigen, alpha-chain clone TAL.1B5 (Code No. M0746; DakoCytomation, DakoCytomation, DK-2600, Glostrup, Denmark). This stain colours the major histocompatibility II (MHC II) molecules that are expressed on cells that serve as APC for CD4+, such as macrophages, monocytes, dendritic cells, and B cells [<xref ref-type="bibr" rid="B51">51</xref>,<xref ref-type="bibr" rid="B52">52</xref>]. Five μm-thick paraffin-embedded sections mounted on coated slides (APES, 3-aminopropyltriethoxysilane) were deparaffinised in xylene and with ethanol. Subsequently, the slides were pre-treated according to the microwave pressure cooker protocol for antigen retrieval (Citrate Buffer 10x, pH 6.0, Klinipath CBB 999, Klinipath BV, 6920 AD, Duiven, Netherlands). The immunohistochemistry was performed in an automated immunostainer (Dako, Glostrup, Denmark; S/N S38-7410-01) according to the manufactures protocol. For visualization the Envision+/HRP mouse (DAB) kit (Dako Ref K4007, DakoCytomation, DK-2600, Glostrup, Denmark) was used. Antibody diluent (Dako Ref S302283) was used to block hydrophobic interactions. Sections were counterstained with Mayer’s hemalum solution (Klinipath). A positive control (thoracic mass, high grade sarcoma) was included in each run to ensure specificity. In negative controls, the primary antibody was replaced by a nonsense antibody of similar isotype (Monoclonal Mouse Anti-Human Cytokeratin Clones AE1/AE3). A second type of negative control was carried out by using central nervous system (CNS) parenchyma with the original primary antibody (Monoclonal Mouse Anti-Human HLA-DR antigen (alpha-chain clone TAL.1B5), as within the normal CNS parenchyma, MHC expression is minimal or absent [<xref ref-type="bibr" rid="B68">68</xref>]. To calculate APC/adipocyte, number of APC/10 high HPF was divided through the number of adipocytes/10 HPF.</p></sec><sec><title>Statistical analysis</title><p>Data are reported as means ± SD and significance was set at P < 0.05. Statistical analyses were performed using IBM SPSS Statistics 20. Gene expression data were analysed with a general linear model by means of repeated measures with depot as within variable and animal as between variable, followed by a Bonferroni post hoc test when a significant difference between depots was detected. Histology data were analysed with a general linear model, univariate analysis and Tukey post hoc test. Correlation analysis (Pearson’s correlation test) was performed to identify relationships between blood parameters, histology findings and mRNA expression of cytokines.</p></sec></sec><sec><title>Abbreviations</title><p>AT: Adipose tissue; RT-qPCR: Quantitative real-time polymerase chain reaction; mRNA: Messenger ribonucleic acid; ACTB: Beta actin; HPRT1: Hypoxanthine phosphoribosyltransferase 1; TUBA4A: Tubulin; RPL 32: Ribosomal protein L32; GAPDH: Glyceraldehyde-3-phosphate dehydrogenase; SDHA: Succinate dehydrogenase complex; IL-10: Interleukin 10; IL-6: Interleukin 6; IL-1β: Interleukin 1β; TNF-α: Tumor necrosis factor α; CCL5: Chemokine ligand 5; MMP2: Matrix metalloproteinase 2; CLS: Crown like structure; SC: Subcutaneous; IR: Insulin resistance; cDNA: Complementary deoxyribonucleic acid; Cq: Transcription value; N ¼: Neck ¼; SOD: Superoxide dismutase; APC: Antigen presenting cell; HE: Hematoxylin and eosin; HPF: High power field; MHC II: Major histocompatibility complex II; APES: 3-aminopropyltriethoxysilane.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>LB was the primary author of the manuscript, responsible for the study design and performed most of the study procedures. TE participated in the study procedures and provided real-time instrument procedures. LJP, RD, GPJJ, PAH, and MH participated in the design of the project, helped to draft the manuscript and supervised the study. All authors read and approved the final manuscript.</p></sec>
Extracellular vesicles and reproduction–promotion of successful pregnancy	<p>Extracellular vesicles (EVs) are membrane-bound complexes secreted from cells under both physiological and pathological conditions. They contain proteins, nucleic acids and lipids and act as messengers for cell–cell communication and signalling, particularly between immune cells. EV research is a rapidly evolving and expanding field, and it appears that all biological fluids contain very large numbers of EVs; they are produced from all cells that have been studied to date, and are known to have roles in several reproductive processes. This review analyses the evidence for the role of EVs throughout human reproduction, starting with the paternal and maternal gametes, followed by the establishment and continuation of successful pregnancies, with specific focus, where possible, on the interaction of EVs with the maternal immune system. Importantly, variations within the EV populations are identified in various reproductive disorders, such as pre-term labour and pre-eclampsia.</p>	<contrib contrib-type="author"><name><surname>Tannetta</surname><given-names>Dionne</given-names></name></contrib><contrib contrib-type="author"><name><surname>Dragovic</surname><given-names>Rebecca</given-names></name></contrib><contrib contrib-type="author"><name><surname>Alyahyaei</surname><given-names>Zahraa</given-names></name></contrib><contrib contrib-type="author"><name><surname>Southcombe</surname><given-names>Jennifer</given-names></name></contrib><aff id="aff1"><institution>Nuffield Department of Obstetrics and Gynaecology, University of Oxford John Radcliffe Hospital Oxford</institution>, <country>UK</country></aff>	Cellular and Molecular Immunology	<sec sec-type="intro"><title>Introduction</title><p>In human pregnancy, the maternal immune system is controlled at every stage of the reproductive process, to promote tolerance to spermatozoa and the semi-allogeneic fetus. Preconception, the maturing oocyte of the pre-ovulatory follicle prepares for its imminent release, and sperm cells must negotiate the hostile female reproductive tract and avoid being phagocytosed by maternal immune cells, to increase the chances of successful fertilization of the oocyte.<sup><xref ref-type="bibr" rid="bib1">1</xref></sup> Sperm cell capacitation permits fusion with the released oocyte, enabling fertilization to take place. The resulting embryo migrates to the uterus, where it must attach to and invade the endometrium, requiring a pro-inflammatory environment.<sup><xref ref-type="bibr" rid="bib2">2</xref></sup> Embryo invasion proceeds as trophoblast signal the presence of the embryo to uterine cells (critically including immune cells which aid the process). From approximately 10 weeks of gestation, when blood flow to the placenta is established, maternal peripheral blood immune cells and immune factors will directly contact the conceptus and therefore, systemic tolerance must be established. This requires a balance between immune tolerance of the fetus and protection against infectious agents for the mother.</p><p>Reproductive failure, due to inappropriate maternal immune cell activity, is prevented by mechanisms that target both the innate and acquired immune systems. The gametes and embryo have a relatively small window during which they must evade the maternal immune system. However, the trophoblast cells, made up of the invasive extravillous trophoblast (EVT) and the multinucleated syncytiotrophoblast (STB) must interact with the maternal immune system for practically the entire length of the pregnancy. Semi-allogeneic trophoblast expressing paternal antigens should be recognized as foreign by maternal immune cells and provoke a rejection response. This is prevented by trophoblast cells having a unique expression profile of human leukocyte antigens (HLA); proteins responsible for allorecognition by natural killer (NK) cells and T cells. The STB layer is unique in being HLA null and therefore immunologically inert, while the EVT that invade the decidual layer uniquely downregulate HLA-A and -B expression and only express HLA-C and the non-classical HLA-G, -E and -F Class I MHC antigens.<sup><xref ref-type="bibr" rid="bib3">3</xref></sup> Immunomodulatory signals are also released during pregnancy as both soluble factors, such as chemokines, cytokines and steroid and protein hormones, and membrane-associated factors in the form of extracellular vesicles (EVs).</p><p>EVs are potent modulators of the immune system, with well-defined roles in immune signalling during both physiological and pathological processes.<sup><xref ref-type="bibr" rid="bib4">4</xref></sup> During the early stages of the human reproductive process, the ovarian follicle, seminal fluid, endometrium, embryo and trophoblast cells are all possible sources of EVs that have the potential to modulate maternal immune function locally. During later pregnancy, the STB of the placenta is the primary source of these EVs; STB releases EVs directly into the maternal blood constituting a major signalling mechanism between fetus and mother.<sup><xref ref-type="bibr" rid="bib5">5</xref></sup> EV signalling may be through protein or lipid ligand-receptor interactions or micro-interfering RNAs (miRNAs) which have been found in both soluble and EV associated forms in various bodily fluids (e.g., plasma, urine, saliva and breast milk).<sup><xref ref-type="bibr" rid="bib6">6</xref>,<xref ref-type="bibr" rid="bib7">7</xref></sup> These are small RNA species that regulate gene expression post-transcriptionally, as part of a novel mechanism for intercellular exchange of genetic material, which also control immune responses. The placenta expresses 46 unique miRNAs found in the primate specific chromosome 19 microRNA cluster alongside abundant ubiquitous species.<sup><xref ref-type="bibr" rid="bib8">8</xref>,<xref ref-type="bibr" rid="bib9">9</xref>,<xref ref-type="bibr" rid="bib10">10</xref>,<xref ref-type="bibr" rid="bib11">11</xref>,<xref ref-type="bibr" rid="bib12">12</xref></sup> These imprinted placental specific miRNAs are the predominant forms of miRNA in the STB layer and trophoblast cells and have been speculated to be immunomodulators in pregnancy.<sup><xref ref-type="bibr" rid="bib13">13</xref>,<xref ref-type="bibr" rid="bib14">14</xref>,<xref ref-type="bibr" rid="bib15">15</xref>,<xref ref-type="bibr" rid="bib16">16</xref></sup></p><p>Pregnancy offers a unique opportunity to study EVs in normal physiology. First, unlike any other condition, it is known exactly when the pregnancy begins and when it finishes, allowing EVs to be followed throughout the entire process. Second, many of the reproductive cellular components carry specific markers which allow the EVs to be distinguished from those produced by other cell types and thirdly, the major source of fetal-derived EVs, the placenta, is available at the end of the pregnancy for study.</p><p>This review will follow the key stages of the human reproductive process, from pre-conception to established pregnancy, outlining the different sources and subtypes of reproductive EVs, their molecular cargo and summarizing our understanding of their complex interactions with maternal immune cells or their roles in reproduction.</p></sec><sec><title>EVs released by reproductive cells and tissues—from nano- to macro-sized material</title><p>EV is a general term encompassing several different vesicle types, released by cells constitutively or in response to specific stimuli or cell stressors, including exosomes, microvesicles, apoptotic vesicles and in pathological situations, necrotic debris. EVs are primarily distinguished on the basis of their size with exosomes, microvesicles and apoptotic vesicles considered to be 30–100 nm, 100 nm–1 µm and 1–5 µm, respectively<sup><xref ref-type="bibr" rid="bib17">17</xref></sup> (<xref ref-type="fig" rid="fig1">Figure 1</xref>). Not only do these vesicles differ in size, but there are also differences with regard to their formation and protein content. Microvesicles are released from the plasma membrane by direct budding or shedding in response to cellular activation or stress. Exosomes are formed from internalized endocytic vesicles, and are constitutively secreted from the cell. Apoptotic vesicles are released from blebbing cells undergoing apoptosis. STB also releases syncytial nuclear aggregates (approx. 20–500 µm in size), in addition to other EV subtypes<sup><xref ref-type="bibr" rid="bib18">18</xref>,<xref ref-type="bibr" rid="bib19">19</xref></sup> (<xref ref-type="fig" rid="fig1">Figure 1</xref>).</p><p>There are no specific markers to distinguish the subtypes of EVs. Protein components of the ‘endosomal-sorting complexes required for transport' (ESCRT complexes) Alix and TSG101, and tetraspanins such as CD63, CD81 and CD9, are enriched in, but not always exclusive to, exosomes. Combined with size, they are used as exosomal markers, while their lower relative abundance in larger vesicles is taken as an indication of microvesicle release.<sup><xref ref-type="bibr" rid="bib20">20</xref></sup> However, EV populations are often not defined by researchers, hindering elucidation of the EV subtypes released by cells from different reproductive tissues. This is also confounded by the use of different isolation techniques, resulting in some groups only studying the larger vesicles,<sup><xref ref-type="bibr" rid="bib21">21</xref>,<xref ref-type="bibr" rid="bib22">22</xref></sup> while others focus on the smaller exosomes to the exclusion of other vesicle types.<sup><xref ref-type="bibr" rid="bib23">23</xref></sup> However, future studies separating the different types of reproductive EVs, i.e., exosomes from microvesicles, using techniques such as immunoaffinity beads or differential centrifugation and subsequent functional studies and proteomic and DNA/RNA analysis of their contents will be crucial to determine their relative functional importance.</p><p>The four main methodologies used routinely to quantify and phenotype EVs in biological samples, are: (i) enzyme-linked immunosorbent assay; (ii) flow cytometry; (iii) electron microscopy (EM); and (iv) western blotting. More recently, other techniques have been used to investigate and characterize EVs including; cryo-EM, nanoparticle tracking analysis (NTA), dynamic light scattering, resistive pulse sensing (IZON qNano), atomic force microscopy and Raman spectroscopy (as reviewed in Refs. 24–26). Flow cytometry is the most widely used method to investigate EVs, however, as flow cytometers are typically designed to examine cells, analysing small EVs is associated with a number of technical limitations and standardisation issues. One of the major limiting factors is resolving EVs from the instrument electronic noise and this can be variable between instruments. Analogue instruments can resolve EVs down to ∼500 nm in diameter,<sup><xref ref-type="bibr" rid="bib27">27</xref>,<xref ref-type="bibr" rid="bib28">28</xref></sup> whereas newer digital flow cytometers are reported to analyse EVs of ≥300 nm.<sup><xref ref-type="bibr" rid="bib29">29</xref>,<xref ref-type="bibr" rid="bib30">30</xref>,<xref ref-type="bibr" rid="bib31">31</xref></sup> Therefore, this limits standard flow cytometric analysis to large microvesicles and apoptotic bodies. The analysis of EVs of ∼100 nm has been demonstrated using a BD Influx flow cytometer, however, this instrument was specifically modified for this purpose.<sup><xref ref-type="bibr" rid="bib32">32</xref>,<xref ref-type="bibr" rid="bib33">33</xref></sup> More recently, the Apogee A50 Micro flow cytometer was marketed, featuring an optical design with an improved light scatter detection performance, resulting in enhanced signal/noise ratio and enabling the detection of EVs down to 100 nm in diameter.<sup><xref ref-type="bibr" rid="bib34">34</xref></sup></p><p>With an appropriate panel of antibodies, multicolour flow cytometry allows the identification of EVs from multiple cellular sources in a single sample, such as those found in blood samples which contain EVs from platelets, red blood cells and leukocytes in non-pregnant individuals, as well as STB-derived EVs in pregnancy<sup><xref ref-type="bibr" rid="bib29">29</xref>,<xref ref-type="bibr" rid="bib35">35</xref></sup> However, accurate sizing of EVs cannot be performed by flow cytometry. It is possible to determine the size of EVs by EM, but it is labour intensive, highly subjective and not quantitative. Although this technique is still routinely used, many researchers are now using an alternative method for EV sizing; NTA. NTA determines the size of vesicles from their Brownian motion, by tracking the movement of individual vesicles in real time, and has, for the first time, allowed the accurate sizing and counting of EVs in biological fluids.<sup><xref ref-type="bibr" rid="bib36">36</xref></sup></p></sec><sec><title>EVs and pre-pregnancy adaptation of the maternal reproductive tract</title><p>The immune system of the female reproductive tract is finely tuned to cope with reproductive processes. It must provide a robust defence against pathogenic agents whilst tolerating the presence of semen harbouring immunogenic paternal alloantigens, and subsequently provide a nurturing environment for the semiallogeneic conceptus. In this section, we will review the literature pertaining to EVs present in follicular fluids of ovulatory follicles, secreted from the endometrium, present in seminal fluid or released from the developing embryo prior to implantation, focusing on any known immune interactions.</p><sec><title>Oocyte EVs</title><p>Sperm and oocyte fusion is essential for fertilisation. The tetraspanin CD9 is known to play an important role in gamete fusion in mice, as CD9-deficient oocytes are unable to fuse with sperm.<sup><xref ref-type="bibr" rid="bib37">37</xref>,<xref ref-type="bibr" rid="bib38">38</xref></sup> A possible role for CD9 in the transfer of oocyte material to sperm has been investigated; however, this topic remains controversial.<sup><xref ref-type="bibr" rid="bib39">39</xref>,<xref ref-type="bibr" rid="bib40">40</xref>,<xref ref-type="bibr" rid="bib41">41</xref></sup> Using a transgenic mouse model, oocytes were shown to transfer CD9 to the sperm head <italic>via</italic> vesicles described as ‘exosome-like'.<sup><xref ref-type="bibr" rid="bib41">41</xref></sup> These CD9-associated EVs were localized to the perivitelline space (PVS) of the oocyte and were shown to be between 50 and 250 nm in diameter and express two known exosome components—ganglioside GM3 and heat shock protein 90. Another study also confirmed the presence of EVs of exosome size (50–150 nm) in the oocyte PVS, some of which were associated with CD9.<sup><xref ref-type="bibr" rid="bib39">39</xref></sup> There are conflicting reports as to whether the accumulation of EVs in the PVS is altered in CD9 null oocytes and whether EV production and secretion from the oocyte is CD9 dependent.<sup><xref ref-type="bibr" rid="bib39">39</xref>,<xref ref-type="bibr" rid="bib41">41</xref></sup> Sperm cultured in medium containing EVs from fluorescently labelled CD9 null oocytes capture these EVs and in turn the sperm acquires fluorescence, thereby suggesting that oocyte EVs are transported to sperm, but this process is independent of CD9.<sup><xref ref-type="bibr" rid="bib39">39</xref></sup> One study showed sperm fusion with wildtype oocytes or CD9 null oocytes cultured in conditioned medium containing CD9-associated EVs isolated from wild-type oocytes.<sup><xref ref-type="bibr" rid="bib41">41</xref></sup> These data suggest that CD9-associated EVs are released from the oocyte prior to fertilisation and are required for sperm fusion. Contrary to this report, two independent studies have failed to replicate the ability of CD9 null oocytes to fuse with sperm after co-incubation with wild-type oocytes or medium containing CD9-associated EVs.<sup><xref ref-type="bibr" rid="bib39">39</xref>,<xref ref-type="bibr" rid="bib40">40</xref></sup></p><p>The tetraspanin CD81 is expressed on the surface of the mouse oocyte<sup><xref ref-type="bibr" rid="bib42">42</xref></sup> and surrounding somatic cells<sup><xref ref-type="bibr" rid="bib43">43</xref></sup> and may also play a role in sperm-oocyte membrane fusion.<sup><xref ref-type="bibr" rid="bib42">42</xref>,<xref ref-type="bibr" rid="bib43">43</xref>,<xref ref-type="bibr" rid="bib44">44</xref></sup> Subcellular expression of CD9 and CD81 in the mouse oocyte demonstrates that these are predominantly localized in the PVS and zona pellucida respectively. It has been proposed that CD9 is primarily produced by the oocyte, whereas CD81 is primarily produced in the surrounding cumulus cells.<sup><xref ref-type="bibr" rid="bib44">44</xref></sup> A model has been proposed where, the localisation of CD9 and CD81 to their respective oocyte compartments is essential for sperm fusion, and as sperm penetrate the PVS both CD9 and CD81 are secreted <italic>via</italic> exosomes and attach to sperm.<sup><xref ref-type="bibr" rid="bib44">44</xref></sup></p><p>Given the size of the oocyte and surrounding somatic cells, and the potential therefore to only produce a limited number of EVs, it would not be surprising to find that oocyte/somatic cell-derived EVs only play a role in close proximity within the ovarian follicle, rather than a broader role modulating the uterine lumen environment. Their limited numbers may also help to explain the need to contain EVs within the oocyte PVS and zona pellucida.</p></sec><sec><title>Follicular fluid EVs</title><p>Follicular fluid contains a plethora of biochemical and metabolic substances that regulate oocyte maturation and follicle growth<sup><xref ref-type="bibr" rid="bib45">45</xref></sup> and recently it was reported that follicular fluid also contains EVs.<sup><xref ref-type="bibr" rid="bib46">46</xref>,<xref ref-type="bibr" rid="bib47">47</xref>,<xref ref-type="bibr" rid="bib48">48</xref>,<xref ref-type="bibr" rid="bib49">49</xref></sup> Ovarian follicular fluid EVs were first described in the horse.<sup><xref ref-type="bibr" rid="bib46">46</xref></sup> This study identified miRNAs within follicular fluid EVs, identical to those in the somatic cells (granulosa and cumulus). These EVs bind and are taken up by granulosa cells both <italic>in vitro</italic> and <italic>in vivo</italic> and therefore may constitute a new form of ovarian cell–cell communication. Significant variation in levels of follicular fluid miRNA-associated EVs in old versus young mares also suggest that these may be indicative of age-related decline in oocyte quality.<sup><xref ref-type="bibr" rid="bib46">46</xref></sup> Similarly bovine follicular fluid contains exosome-associated miRNAs.<sup><xref ref-type="bibr" rid="bib49">49</xref></sup> Bovine granulosa cells take-up follicular fluid exosome-associated miRNAs <italic>via</italic> endocytosis <italic>in vitro</italic>, significantly increasing the level of endogenous miRNAs and changing mRNA expression of target genes in granulosa cells known to be involved in follicle development. These follicular fluid exosome-associated miRNAs may also play a role in regulating oocyte growth as they are differentially expressed in follicles containing growing versus fully grown oocytes.<sup><xref ref-type="bibr" rid="bib49">49</xref></sup> Using differential centrifugation and NTA, our laboratory has shown the presence of exosome and microvesicle sized EVs in human follicular fluid (<xref ref-type="fig" rid="fig2">Figure 2</xref>). Human follicular fluid EVs are polydisperse and include both microvesicle and exosome sized EVs, with a mean diameter of ∼220 nm (<xref ref-type="fig" rid="fig2">Figure 2ai</xref>). Two other studies have also confirmed the presence of human follicular fluid derived EVs.<sup><xref ref-type="bibr" rid="bib47">47</xref>,<xref ref-type="bibr" rid="bib48">48</xref></sup> Human follicular fluid EVs contain numerous miRNAs, some of which target genes that are important in regulating pathways of reproduction, metabolism and endocrine function.<sup><xref ref-type="bibr" rid="bib48">48</xref></sup> However, as yet, possible interactions of follicular fluid EVs and maternal immune cells have not been investigated.</p></sec><sec><title>Endometrial EVs</title><p>Fertilisation of the oocyte occurs in the oviduct, and a novel role for exosomes residing in the oviduct of mice has been proposed, termed ‘oviductosomes'. Oviductosomes express a Ca<sup>2+</sup> regulatory protein, PMCA4a which is transferred to sperm. PMCA4a has roles in sperm storage in the oviduct, capacitation and the acrosome reaction, therefore it is proposed these oviductosomes may have an important role in fertility.<sup><xref ref-type="bibr" rid="bib50">50</xref></sup></p><p>EVs, predominantly 50–150 nm in diameter, have been identified in human uterine fluid and dissociated mucus.<sup><xref ref-type="bibr" rid="bib51">51</xref></sup> Most of the EVs found in the uterine cavity are likely to be of exosome origin, derived from the endometrial epithelium, as the two tetraspanins CD9 and CD63 are present on the apical surfaces of endometrial epithelial cells.<sup><xref ref-type="bibr" rid="bib51">51</xref></sup> It is hypothesized that the endometrial epithelial cells or the blastocyst acquire endometrial exosomes to improve implantation.<sup><xref ref-type="bibr" rid="bib51">51</xref></sup></p><p>Endometrial remodelling by matrix metalloproteinases is essential during each menstrual cycle in preparation for embryo implantation. Using human endometrial epithelial cell lines, a possible role has emerged for endometrial epithelial EVs in regulating this process.<sup><xref ref-type="bibr" rid="bib51">51</xref>,<xref ref-type="bibr" rid="bib52">52</xref>,<xref ref-type="bibr" rid="bib53">53</xref></sup> Microvesicles from a human uterine epithelial cell line were shown to contain the extracellular matrix metalloprotease inducer (EMMPRIN (CD147)) which stimulated uterine stromal fibroblast matrix metalloproteinase production.<sup><xref ref-type="bibr" rid="bib52">52</xref></sup> Microvesicle EMMPRIN secretion is stimulated by the ovarian hormones estradiol and progesterone and also by the pro-inflammatory cytokine IL-1β, thereby suggesting that its secretion is regulated as part of an inflammatory response.<sup><xref ref-type="bibr" rid="bib52">52</xref></sup> It is proposed that EMMPRIN is secreted as a soluble protein in response to degradation of microvesicles.<sup><xref ref-type="bibr" rid="bib52">52</xref></sup> EMMPRIN has also been shown to be released in microvesicles in response to G protein-coupled receptor 30 stimulation.<sup><xref ref-type="bibr" rid="bib53">53</xref></sup></p><p>The analysis of miRNAs from the human uterine epithelial cell line ECC1 demonstrated sorting of certain miRNAs into EVs, showing >200 miRNAs were co-expressed by the ECC1 parent cells and the ECC1 EVs, 13 specific to EVs and 5 specific to the parent cells.<sup><xref ref-type="bibr" rid="bib51">51</xref></sup> Predicted target genes of the EV miRNAs included those known to play a role in regulating embryo implantation such as extracellular matrix receptor interactions, adherens and tight junctional proteins and the Jak-STAT signalling and VEGF signalling pathways.<sup><xref ref-type="bibr" rid="bib51">51</xref></sup></p></sec><sec><title>Seminal fluid EVs</title><p>Seminal fluid contains large numbers of EVs, the majority of which in humans originate from the epithelial cells lining the acinar ducts of the prostate gland and are called prostasomes, although other epithelial cells of the male reproductive tract also secrete EVs, such as from the epididymis (epididymosomes) and vesicular glands, reviewed elsewhere.<sup><xref ref-type="bibr" rid="bib1">1</xref>,<xref ref-type="bibr" rid="bib54">54</xref></sup> Extensive analysis of prostasomes has been performed, summarized by Ronquist<italic>.</italic><sup><xref ref-type="bibr" rid="bib54">54</xref></sup> We have also used NTA to determine the size of prostasomes. As previously reported,<sup><xref ref-type="bibr" rid="bib54">54</xref></sup> prostasomes are very small in size, with a modal peak at ∼120 nm (<xref ref-type="fig" rid="fig2">Figure 2aii</xref>). Prostasomes play an important role in fertility; they bind to sperm and are thought to enhance their motility and ability for capacitation—they are enriched in divalent cations known to be involved with sperm motility, such as Ca<sup>2+</sup>, and contain enzymes which control their levels. They also transfer hydrolases such as ecto-diadenosine polyphosphate hydrolase and are rich in sphingomyelin with a high cholesterol/phospholipid ratio, all of which regulate the acrosome reaction. They also contain fragments of genomic DNA, although it is not clear if this has a functional consequence. Chromogranin B confers bactericidal properties to prostasomes and importantly they also prevent immune cell recognition of sperm in the female reproductive tract and have immune modulatory properties. Prostasomes inhibit lymphocyte proliferation and macrophage/neutrophil functions.<sup><xref ref-type="bibr" rid="bib55">55</xref></sup> Direct inhibition of NK cells by prostasomal expression of CD48, the ligand for the CD244 (NK activating receptor 2B4), has been identified.<sup><xref ref-type="bibr" rid="bib56">56</xref></sup> They also contain the complement regulatory proteins CD59 (a GPI-anchored membrane attack complex inhibitory protein) and CD46 (a cofactor for proteolytic inactivation of C3b and C4b)—both of which protect sperm from lysis mediated by the female complement system.<sup><xref ref-type="bibr" rid="bib57">57</xref>,<xref ref-type="bibr" rid="bib58">58</xref></sup> High levels of reactive oxygen species in semen samples are associated with infertility. It is possible that contaminating neutrophils are a source of reactive oxygen species generation, and prostasomes are thought to reduce reactive oxygen species production.<sup><xref ref-type="bibr" rid="bib59">59</xref></sup> Finally, Galectin-3 (a multifaceted, immunomodulatory lectin) is expressed on prostasomes and probably has wide ranging functions within the female reproductive tract.<sup><xref ref-type="bibr" rid="bib60">60</xref></sup></p></sec><sec><title>Pre-implantation embryo EVs</title><p>EVs from the pre-implantation embryo have not yet been reported, but it is likely that the embryo has the capacity to release them. Although embryo culture supernatants can be obtained from <italic>in vitro</italic> fertilization (IVF) clinics, we have found that the <italic>in vitro</italic> fertilization (IVF) culture media alone contains high levels of EVs, probably from the nutrients supplemented for embryo growth, making detection of specific embryo-derived EVs challenging.</p></sec></sec><sec><title>EVs and adaptation of the maternal system during pregnancy</title><p>Following implantation and throughout the remainder of pregnancy, trophoblast cells form the interface between the maternal immune cells and the fetus. Differentiation of the trophectoderm cell lineage gives rise to two forms of trophoblast cells, villous cytotrophoblast (CTB) and EVT. EVT proliferate in the tips of the anchoring villi attached to the uterine wall, then acquire invasive characteristics and migrate into the decidua and, as pregnancy progresses, into the myometrium. Both interstitial and endovascular subtypes of EVT form part of a coordinated response with maternal uterine NK and macrophage cells to adapt blood flow to the needs of the placenta and fetus, by destroying the muscular walls and replacing the endothelial lining of the distal portions of the uterine spiral arteries with EVT.<sup><xref ref-type="bibr" rid="bib61">61</xref>,<xref ref-type="bibr" rid="bib62">62</xref>,<xref ref-type="bibr" rid="bib63">63</xref></sup> The CTB fuse into multinucleated STB covering the villi of the placenta that are bathed in nutrient rich secretions from endometrial glands until towards the end of the first trimester, when maternal blood flow to the placenta is established.<sup><xref ref-type="bibr" rid="bib64">64</xref></sup> This then brings into play the largest maternal/fetal interface—the STB of the human haemochorial placenta in direct contact with the maternal blood. STB becomes the dominant site of conceptus EV release and the release of STB EVs into the maternal circulation extends this interface beyond the uterus and out into the maternal systemic vasculature.<sup><xref ref-type="bibr" rid="bib65">65</xref></sup></p><p>Taking into account the type of trophoblast EVs and the source of the immune cells used to study their interaction, maternal immune cell modulation and the role played by trophoblast EVs during pregnancy will be discussed (literature for interactions with adaptive and innate immune cells are summarized in <xref rid="tbl1" ref-type="table">Tables 1</xref> and <xref rid="tbl2" ref-type="table">2</xref>, respectively).</p></sec><sec><title>EVT EVs</title><p>Release of EVs by EVT during early pregnancy, has been demonstrated both <italic>in vivo</italic> and <italic>in vitro</italic> through detection of HLA-G positive EVs. HLA-G is expressed on EVT but not STB. Cultured first trimester explants and the EVT like cell line Swan71 release HLA-G-positive EVs, and HLA-G-positive EVs have been detected in the maternal circulation.<sup><xref ref-type="bibr" rid="bib66">66</xref>,<xref ref-type="bibr" rid="bib67">67</xref>,<xref ref-type="bibr" rid="bib68">68</xref></sup> Using flow cytometry, circulating HLA-G-positive EVs were identified throughout pregnancy, with amounts decreasing towards term.<sup><xref ref-type="bibr" rid="bib67">67</xref></sup> Detection of HLA-G-positive EVs was also reported in the third trimester of pregnancy;<sup><xref ref-type="bibr" rid="bib68">68</xref></sup> however both accounts of detectable levels of circulating HLA-G positive vesicles are surprising given the very small surface area of endovascular EVT exposed to the large maternal blood volume.<sup><xref ref-type="bibr" rid="bib69">69</xref></sup> To date, only one study has characterized the morphology of EVs released by EVT using Swan71 cell line and a protocol designed to enrich for exosomes.<sup><xref ref-type="bibr" rid="bib70">70</xref></sup> Density (sucrose gradient), morphology (EM) and size (dynamic light scattering) analysis strongly suggested that Swan71 cells release exosomes. This finding, together with reports that HLA-G-positive vesicles can be detected by flow cytometry, suggests that native EVT release both exosomes and microvesicles.</p><p>Evidence for immunomodulatory effects of EVT-derived EVs is also limited (<xref rid="tbl1" ref-type="table">Tables 1</xref> and <xref rid="tbl2" ref-type="table">2</xref>). Circulating HLA-G-positive EVs present in plasma from healthy term pregnant women have been reported to bind at low levels to T lymphocytes <italic>ex vivo.</italic><sup><xref ref-type="bibr" rid="bib68">68</xref></sup> The same study showed that EVs isolated from healthy term pregnancy plasma modestly decreased peripheral T lymphocyte and Jurkat cell STAT3 phosphorylation.<sup><xref ref-type="bibr" rid="bib68">68</xref></sup> Potential to affect immune cell function is also suggested by the observation that HLA-G-positive exosomes from first trimester explants carry immunomodulatory proteins B7-H1 and B7-H3, which have previously been shown to modulate effector T cell function. B7-H1, also known as PD-L1, is implicated in promoting maternal/fetal tolerance.<sup><xref ref-type="bibr" rid="bib66">66</xref></sup> HLA-G itself also has immunosuppressive properties thought to protect the fetus from immune rejection; HLA-G binds with high affinity to the NK inhibitory receptor LILRB1 inhibiting NK cell killing activity.<sup><xref ref-type="bibr" rid="bib71">71</xref>,<xref ref-type="bibr" rid="bib72">72</xref></sup> Vesicles, thought to be exosomes, isolated from the EVT-like cell line Swan71, have been shown to stimulate a pro-inflammatory cytokine and chemokine profile in both the THP-1 monocytic cell line and cultures of primary macrophages,<sup><xref ref-type="bibr" rid="bib73">73</xref>,<xref ref-type="bibr" rid="bib74">74</xref></sup> suggesting monocytic pro-inflammatory as well as lymphocytic immunosuppressive activity of EVT exosomes.</p><p>It is not known whether the early STB layer releases any vesicular material at the time when the developing intervillous space is filled with endometrial gland secretions, and whether these EVs would interact with maternal immune cells. Interstitial trafficking of STB derived EVs through the decidua cannot be ruled out, especially during the very early stages of pregnancy when the STB forms within the decidual layer, but this has not yet been investigated. However, STB EVs from the late first trimester onwards, around the time when placental blood flow is established, are by far the most studied EVs derived from reproductive tissue. This is due to the availability of human placental tissue from first trimester terminations and at term.<sup><xref ref-type="bibr" rid="bib69">69</xref>,<xref ref-type="bibr" rid="bib75">75</xref></sup> Term placentas are particularly useful as they can yield large quantities of STB EVs derived from a fresh, primary tissue, especially if the placenta is obtained from an elective caesarean section and has therefore not been subjected to the stresses of labour. It is important to remember however that placentas from normal pregnancies are not normally available prior to term; therefore, identification of STB in the maternal blood will be the only way we can study placenta EVs throughout later gestations.</p><p>We have used NTA to determine the size of STB EVs prepared by perfusion of term placentas (<xref ref-type="fig" rid="fig2">Figure 2</xref>). Sizes of STB EVs isolated from placental perfusate range from approximately 40 nm upwards, with the majority (>90%) being less than 1 µm and 70% being less than 300 nm in diameter, suggesting that these EVs are also predominantly a mixture of exosomes and microvesicles.<sup><xref ref-type="bibr" rid="bib18">18</xref>,<xref ref-type="bibr" rid="bib31">31</xref></sup> Using differential centrifugation, placental perfusion-derived STB EVs can also be fractionated to give enriched STB microvesicle and STB exosome preparations, characterized using NTA (<xref ref-type="fig" rid="fig2">Figure 2bi</xref> and bii; STB microvesicles and <xref ref-type="fig" rid="fig2">Figure 2ci and 2cii</xref>; STB exosomes), transmission EM (<xref ref-type="fig" rid="fig2">Figure 2biii</xref>; STB microvesicles and <xref ref-type="fig" rid="fig2">Figure 2ciii</xref>; STB exosomes) and western blotting for exosome markers CD63 and Alix (<xref ref-type="fig" rid="fig2">Figure 2d</xref>) (Tannetta and Dragovic, unpublished data).</p><sec><title>Identification of syncytiotrophoblast EVs in the maternal circulation</title><p>The deportation of STB material into the maternal circulation has been recognized for many years.<sup><xref ref-type="bibr" rid="bib76">76</xref></sup> Increased levels of STB EVs were found in the uterine vein blood compared to the peripheral blood, consistent with their placental origin.<sup><xref ref-type="bibr" rid="bib77">77</xref></sup> Levels measured by STB EV enzyme-linked immunosorbent assay, which captures STB EV using an antibody to placental alkaline phosphatase (PLAP), a STB marker, showed increases with advancing pregnancy <sup><xref ref-type="bibr" rid="bib78">78</xref></sup> and labour <sup><xref ref-type="bibr" rid="bib79">79</xref></sup> returning to non-pregnant levels in most cases by 48 h post-delivery. Flow cytometry has also been used to confirm the particulate nature of the STB material pelleted from maternal plasma using two STB-specific monoclonal antibodies, one that binds PLAP and another called ED822 which recognizes an as yet unknown antigen on the apical surface of the STB.<sup><xref ref-type="bibr" rid="bib29">29</xref>,<xref ref-type="bibr" rid="bib35">35</xref>,<xref ref-type="bibr" rid="bib67">67</xref>,<xref ref-type="bibr" rid="bib77">77</xref>,<xref ref-type="bibr" rid="bib78">78</xref></sup> PLAP-positive EVs are found circulating throughout pregnancy, with levels increasing towards term.<sup><xref ref-type="bibr" rid="bib78">78</xref></sup> PLAP-positive EVs have also been isolated directly from pregnancy plasma, taken between 26 and 28 weeks gestation, using agarose beads coated with a PLAP-specific antibody.<sup><xref ref-type="bibr" rid="bib80">80</xref></sup></p><p>The composition of EVs in terms of the proportions of exosomes, microvesicles, apoptotic bodies and syncytial nuclear aggregates will have an important bearing on their functional characteristics and distribution throughout the body. As such, circulating STB EVs give only a glimpse of the material released from the placental surface. Syncytial nuclear aggregates have been identified in uterine vein and inferior vena cava blood. However, due to their size, syncytial nuclear aggregates are mostly trapped in the pulmonary capillary bed,<sup><xref ref-type="bibr" rid="bib81">81</xref></sup> while smaller STB EVs are able to pass through the lung capillaries and enter the peripheral circulation.<sup><xref ref-type="bibr" rid="bib77">77</xref></sup> Further loss of STB EVs, filtered out by organs such as the liver and spleen, is also possible, although this has not been investigated.<sup><xref ref-type="bibr" rid="bib82">82</xref></sup> Meanwhile, phagocytic immune and endothelial cells constantly clear cellular debris from the circulation that could include STB derived material.<sup><xref ref-type="bibr" rid="bib78">78</xref></sup> This raises questions such as: do STB EVs have target cells with which they interact?, what biological effects do the STB EVs have on the cells and organs they come into contact with? and what is the significance of the material not removed from the circulation?</p></sec><sec><title>Functional effects of syncytiotrophoblast EVs throughout pregnancy</title><p>STB EVs have the potential to modulate maternal immune cell function at a local and systemic level (<xref rid="tbl1" ref-type="table">Tables 1</xref> and <xref rid="tbl2" ref-type="table">2</xref>). STB EVs are bound at detectable levels to monocytes from the second trimester and increase further at term <italic>in vivo.</italic><sup><xref ref-type="bibr" rid="bib78">78</xref></sup> Studies carried out <italic>in vitro</italic> also showed that monocytes and B cells rapidly bound and internalized STB EVs, suggesting possible functional interactions.<sup><xref ref-type="bibr" rid="bib83">83</xref></sup> Intriguingly, NK cell and T-cell binding of STB EVs has not been reported, even though several studies, outlined below, have shown STB EVs to have immunosuppressive effects on the function of these cell types.</p><p>Prevention of fetal rejection, in part by the suppression of maternal cell mediated immune responses, has long been recognized. Evidence continues to grow that placental EVs form part of a range of immunosuppressive factors released by the placenta that are involved in this process. The Y chromosome-linked minor histocompatibility antigen DDX3Y has been shown to be released from first trimester placental explant bound to STB debris and the authors speculate that this may lead to the induction of antigen specific regulatory CD8<sup>+</sup> T cells, conferring immunological tolerance to the fetus.<sup><xref ref-type="bibr" rid="bib84">84</xref></sup> Mid-first to early-second trimester explants also release STB exosomes with biologically active MHC class I chain-related (MIC A/B) proteins and UL-16 binding proteins on their surface, able to initiate internalization of NK cell activating receptor NKG2D on NK cells, cytotoxic T lymphocytes and γδ T cells, reducing cell surface NKG2D expression and subsequent cytotoxic capacity.<sup><xref ref-type="bibr" rid="bib23">23</xref>,<xref ref-type="bibr" rid="bib85">85</xref></sup> T cell responses have also been shown to be significantly inhibited by several different preparations of placental EVs, measured by PHA and mixed lymphocyte response induced proliferation,<sup><xref ref-type="bibr" rid="bib86">86</xref>,<xref ref-type="bibr" rid="bib87">87</xref>,<xref ref-type="bibr" rid="bib88">88</xref></sup> Fas ligand and TRAIL-mediated lymphocyte and activated peripheral blood mononuclear cell (PBMC) apoptosis and CD3-zeta loss.<sup><xref ref-type="bibr" rid="bib80">80</xref>,<xref ref-type="bibr" rid="bib87">87</xref>,<xref ref-type="bibr" rid="bib89">89</xref>,<xref ref-type="bibr" rid="bib90">90</xref></sup> Immunosuppressive activity has also been demonstrated with much larger shed STB material. Phagocytosis of large apoptotic mononuclear trophoblast and syncytial nuclear aggregates from cultured first trimester placental explants stimulated IL-10 release and indoleamine 2,3-dioxygenase (IDO) expression, and decreased IL-1β secretion in monocytic U937 cells.<sup><xref ref-type="bibr" rid="bib91">91</xref></sup> In addition, large trophoblast debris from term placenta explants decreased surface expression of MHC class II molecules and inflammatory cytokine release, with a concomitant increase in anti-inflammatory cytokines and IDO expression, from primary macrophages.<sup><xref ref-type="bibr" rid="bib22">22</xref></sup> Therefore, larger apoptotic trophoblast debris may have an integral role in tolerizing maternal immune cells to fetal alloantigens.<sup><xref ref-type="bibr" rid="bib21">21</xref></sup></p><p>While the immunosuppressive effects of placental EVs on maternal immune cells suggest a role for STB EVs in the mechanism by which the placenta avoids immune rejection, the situation is more complicated than this because maternal innate immune responses are also activated in normal pregnancy and bring about a systemic inflammatory state.<sup><xref ref-type="bibr" rid="bib92">92</xref></sup> This may seem counterintuitive; however, controlled inflammation, in the face of suppressed T and NK cell-mediated immune responses, may be beneficial in helping the mother fight infection. There is growing evidence that placental EVs play a role in the maternal systemic inflammatory response. We and others have shown that preparations of STB EVs stimulate PBMC and monocytes to release a range of proinflammatory cytokines (including TNFα, MIP-1α, IL-1α, IL-1β, IL-6, IL-8, IL-12, IL-18 and IFN-γ).<sup><xref ref-type="bibr" rid="bib73">73</xref>,<xref ref-type="bibr" rid="bib74">74</xref>,<xref ref-type="bibr" rid="bib78">78</xref>,<xref ref-type="bibr" rid="bib83">83</xref>,<xref ref-type="bibr" rid="bib93">93</xref></sup> Reports of the effects of STB EVs on granulocyte function are limited, but term placental EV preparations have been shown to directly stimulate neutrophils, resulting in increased superoxide production and the formation of neutrophil extracellular lattices.<sup><xref ref-type="bibr" rid="bib94">94</xref>,<xref ref-type="bibr" rid="bib95">95</xref></sup></p><p>As discussed above, STB EVs have a wide range of functional activities suggesting that they carry a variety of bioactive molecules. Several EV-associated biologically active factors have already been identified, a summary of which is shown in <xref ref-type="fig" rid="fig3">Figure 3</xref>. A crucial step now is to further define which molecules are present on the different vesicle types and which are responsible for the functional effects. We have performed proteomic analysis on perfused placental EVs using multidimensional protein identification technology.<sup><xref ref-type="bibr" rid="bib96">96</xref></sup> Over 2000 proteins were identified and included potential immunomodulatory molecules such as alarmins (HSP70, high mobility group box 1, fetal haemoglobin and galectin 3), immunoregulatory molecules (CD200, CD147 and galectin 1) and complement regulatory molecules (CD55 and CD59) (Tannetta <italic>et al.</italic>, unpublished observations). Very few studies have specifically investigated placental vesicle-associated miRNA.</p><p>Several placenta-specific miRNAs are released in exosomes by the trophoblast cell line BeWo and isolated human term trophoblast.<sup><xref ref-type="bibr" rid="bib13">13</xref>,<xref ref-type="bibr" rid="bib14">14</xref></sup> Using Affymetrix GeneChipR miRNA arrays we also identified five placental specific miRNAs in term placental perfusion-derived EVs: 517a, 512-3p, 517b, 518b and 519a (Tannetta <italic>et al.</italic>, unpublished observations), which have previously been reported in the placenta and maternal circulation.<sup><xref ref-type="bibr" rid="bib10">10</xref>,<xref ref-type="bibr" rid="bib13">13</xref></sup> Using qRTPCR, Cronqvist <italic>et al</italic>.<sup><xref ref-type="bibr" rid="bib20">20</xref></sup> also demonstrated placental miRNA in placental perfusion-derived EVs. Moreover, isolated trophoblast from term placenta, which fuse in culture to form STB, release exosomes able to transfer viral resistance to recipient cells by inducing autophagy through the delivery of chromosome 19 microRNA cluster miRNAs,<sup><xref ref-type="bibr" rid="bib97">97</xref></sup> demonstrating yet another mechanism of STB EVs modulation of maternal immune responses.</p></sec></sec><sec><title>Amniotic fluid EVs</title><p>The amniotic fluid is unsurprisingly rich in EVs, as fetal waste products are secreted into this medium and fetal urine contains high levels of EVs. Recently a role has emerged for amniotic fluid EVs in controlling inflammation in the fetal compartment. Exosomes are detected in second trimester amniotic fluids and contain the inducible heat shock protein 72, which is a known inhibitor of immune activity.<sup><xref ref-type="bibr" rid="bib98">98</xref>,<xref ref-type="bibr" rid="bib99">99</xref></sup> Amniotic fluid exosomes are also phagocytosed by the human monocytic precursor cell line THP-1, inducing cytokine production such as IL-1β, TNF-α and IL-6 leading to STAT-3 activation and an immunosuppressive phenotype.<sup><xref ref-type="bibr" rid="bib100">100</xref></sup> However, maternal immune cells do not contact amniotic fluid and therefore this result may be a reflection of the immune-inhibitory nature of exosomes rather than an amniotic fluid EV-specific phenotype.</p></sec><sec><title>Clinical perspectives</title><p>EVs have been extensively studied in a wide variety of disease pathologies, including autoimmune disorders, inflammatory diseases and cancer. EVs are also implicated in pathological processes of pregnancy, particularly preeclampsia (PE); however, a role for EVs in other reproductive disorders such as pre-term labour is also emerging. In developed countries about 5%–7% of births are pre-term. Interestingly, plasma levels of exosomes at 28–30 weeks are reduced in women who deliver preterm compared to women who deliver at term, and these exosomes contain less of the immune inhibitory molecule FasL, suggesting impaired suppression of cytotoxic T-cell activity.<sup><xref ref-type="bibr" rid="bib89">89</xref></sup> Anti-sperm antibodies, which can occur in human semen or the serum of both men and women, have been linked with male infertility, although the clinical association remains controversial.<sup><xref ref-type="bibr" rid="bib101">101</xref></sup> Anti-sperm antibodies also bind to prostasomes,<sup><xref ref-type="bibr" rid="bib102">102</xref></sup> so could alter their function. Several antigens have been identified on prostasomes, such as the prolactin-inducible protein, which has various immunological functions.<sup><xref ref-type="bibr" rid="bib103">103</xref></sup> Although implicated in modulation of the immune response, as yet there are no known clinical associations between prostasomes and fertility. Research is focused towards using prostasomes as a biomarker for prostate cancer, as prostate-specific antigen is known to be detected on prostasomes, and a small study has shown that prostate cancer patients have prostasomes in their blood plasma.<sup><xref ref-type="bibr" rid="bib104">104</xref></sup> In addition, the complement evading protein found on prostasomes CD59, is elevated in cancer cells which may aid tumour immune evasion.<sup><xref ref-type="bibr" rid="bib54">54</xref></sup> A recent study examining human follicular fluid and its supernatant in healthy controls and polycystic ovarian syndrome patients identified miRNAs in EVs and the supernatant.<sup><xref ref-type="bibr" rid="bib48">48</xref></sup> Two miRNAs present in follicular fluid EV and the supernatant; miR-132 and miR-320, which regulate estradiol concentrations, were significantly decreased in polycystic ovarian syndrome patients compared to healthy controls.<sup><xref ref-type="bibr" rid="bib48">48</xref></sup> Undoubtedly, further research into functional effects of EVs associated with reproduction will yield other clinical observations.</p><p>Most research on reproductive cell EVs has focused on PE. PE is a disorder of human pregnancy which affects 2.5%–3.0% of women, with potential to be lethal or detrimental to long-term health for both the mother and baby.<sup><xref ref-type="bibr" rid="bib96">96</xref>,<xref ref-type="bibr" rid="bib105">105</xref></sup> PE develops in two stages; the first (preclinical) stage comprises poor spiral artery remodelling (8–18 weeks), leading to dysfunctional perfusion and placental oxidative stress.<sup><xref ref-type="bibr" rid="bib106">106</xref></sup> This stimulates the release of placental factors into maternal blood that cause the second, clinical stage (after 20 weeks) characterized by the maternal syndrome of systemic vascular inflammation, that underlies the symptoms of hypertension, proteinuria, oedema, activation of the coagulation system and in severe cases, eclampsia, typified by convulsions.<sup><xref ref-type="bibr" rid="bib107">107</xref></sup></p><p>Several studies, using enzyme-linked immunosorbent assay methods, have shown increased levels of placental EVs in the maternal circulation in PE, with levels correlated to disease severity.<sup><xref ref-type="bibr" rid="bib77">77</xref>,<xref ref-type="bibr" rid="bib108">108</xref>,<xref ref-type="bibr" rid="bib109">109</xref></sup> Increased EV shedding also appears to be key to the development of the maternal syndrome of PE as increased levels were specific to PE complicated by fetal growth restriction and were not evident in normotensive fetal growth restriction, despite similar placental pathology.<sup><xref ref-type="bibr" rid="bib108">108</xref></sup> A significant increase in placental EV shedding in PE women during labour, may also play a part in postpartum worsening of the disease.<sup><xref ref-type="bibr" rid="bib79">79</xref></sup> Results of flow cytometry studies have been less consistent with some studies showing increased plasma levels of placental EVs in PE, while others found no difference between PE and normal pregnancy.<sup><xref ref-type="bibr" rid="bib29">29</xref>,<xref ref-type="bibr" rid="bib35">35</xref>,<xref ref-type="bibr" rid="bib67">67</xref>,<xref ref-type="bibr" rid="bib78">78</xref>,<xref ref-type="bibr" rid="bib110">110</xref></sup></p><p>Qualitative as well as quantitative differences in placental EVs shed in PE, such as changes in the types of EVs released and the molecular cargo carried by the EVs, could alter their biological function. There are limited studies addressing this issue due to availability of tissue from well-defined PE pregnancies. NTA analysis of PE and normal placental perfusion-derived EVs showed a significant increase in vesicle size in PE, suggesting a shift in the balance towards shedding of more STB microvesicles which could alter their overall functional effects to a more pro-inflammatory, anti- angiogenic and procoagulant state.<sup><xref ref-type="bibr" rid="bib31">31</xref>,<xref ref-type="bibr" rid="bib111">111</xref></sup> Treatment of PBMC with EVs from PE placenta explants caused significantly higher production of several pro-inflammatory cytokines and chemokines, including IL-1β, compared to normal placental EVs.<sup><xref ref-type="bibr" rid="bib112">112</xref></sup> Increased placental EV stimulated neutrophil activation is also suggested by higher levels of superoxide production induced by PE placenta EVs and increased formation of neutrophil extracellular lattices in the intervillous space of PE placentae.<sup><xref ref-type="bibr" rid="bib88">88</xref>,<xref ref-type="bibr" rid="bib94">94</xref></sup> Possible functional mediators include peroxidized lipids due to increased placental oxidation in PE.<sup><xref ref-type="bibr" rid="bib113">113</xref></sup> Excessive activation of the coagulation system and increased platelet activation are also features of PE.<sup><xref ref-type="bibr" rid="bib114">114</xref></sup> PE placental-perfusion EVs have higher functional tissue factor levels than those from normal placentas, implicating them in the excessive activation of the clotting system seen in this disorder.<sup><xref ref-type="bibr" rid="bib115">115</xref>,<xref ref-type="bibr" rid="bib116">116</xref></sup> Placental EVs have also been shown to affect the function of endothelial cells, inhibiting their proliferation and growth as a monolayer <italic>in vitro</italic><sup><xref ref-type="bibr" rid="bib117">117</xref>,<xref ref-type="bibr" rid="bib118">118</xref>,<xref ref-type="bibr" rid="bib119">119</xref></sup> and inhibiting the relaxation of preconstricted blood vessels <italic>ex vivo.</italic><sup><xref ref-type="bibr" rid="bib117">117</xref>,<xref ref-type="bibr" rid="bib119">119</xref>,<xref ref-type="bibr" rid="bib120">120</xref></sup> Furthermore, when human umbilical vein endothelial cells are cultured with placental EVs the culture supernatants can secondarily activate neutrophils, demonstrating the potential for a vicious cycle of inflammatory activation.<sup><xref ref-type="bibr" rid="bib121">121</xref></sup> To date, no differential effects of PE and control placental EVs on endothelial cells has been shown; however, the disruptive effects of placental EVs on endothelial cell function demonstrated <italic>in vitro</italic> and <italic>ex vivo</italic> and discussed above suggest a contribution of placental EVs to the endothelial dysfunction characteristic to the maternal syndrome of PE.<sup><xref ref-type="bibr" rid="bib117">117</xref></sup></p></sec><sec sec-type="discussion"><title>Discussion</title><p>The field of EV research has grown exponentially in the last 10 years and continues to grow with new findings confirming the involvement of EVs in normal physiology and disease. This is also true for EVs from reproductive tissues and the role they play in the successful establishment of pregnancy. However, by taking an overview of the literature regarding reproductive EVs, it becomes apparent that further work is needed to better understand the source and role of EVs in human reproduction. As outlined in this review, EVs have been identified in a variety of reproductive fluids or are known to be produced by certain human reproductive cells and to carry a multitude of functional moieties (<xref ref-type="fig" rid="fig3">Figure 3</xref>). However, it is not known if the sperm themselves produce EVs. Similarly, it is not yet known if human female gametes secrete EVs, although evidence from murine studies suggests that oocytes do produce EVs which bind to sperm and aid fertilisation. It is also not known if the pre-implantation embryo secretes EVs, although, as both the oocyte (in mouse) and trophoblast cells secrete EVs, it is likely that the pre-implantation embryo also has this capacity.</p><p>Research into the targeting of reproductive tissue EVs to the maternal immune system also has far to go. The different reproductive EVs will encounter highly specialized maternal immune cell niches, such as the immune cells of the uterine epithelial mucosa (oocyte, follicular fluid, semen and pre-implantation embryo EVs), decidua (early STB and interstitial EVT EVs) and peripheral circulation (endovascular EVT and STB EVs). Each niche has its own specialized immune cell population and cytokine and chemokine microenvironment suggesting niche-specific effects of EVs. Proteomic and miRNA screening of EVs has identified many candidate molecules carried by various EV types; as we have described here, however, functional studies are still required to confirm their association. Also the downstream consequences of immune cell modulation in each niche will be very different, with more local effects in the uterine lumen and decidua, whereas effects of STB EVs in the maternal circulation could have implications for multiple organs and vascular cell types. This presents challenges in designing models to better understand the role of reproductive EVs in establishing successful pregnancy. These include modelling of the specialized immune cells found in each niche, such as uterine NK cells of the decidua and immune cells resident in organs (such as Kupffer cells in the liver). However, studies to date have focused on using peripheral blood immune cells, which are easier to obtain, but have different functions. Length of exposure to each reproductive EV population may also need to be taken into account. Multiple exposures to semen EVs can occur before a pregnancy is established, unlike STB EVs where each exposure is unique but slow and sustained, gradually increasing from approximately 10 weeks of gestation, with the onset of placental perfusion with maternal blood, to reach a maximum at term. However, each niche may be important in driving maternal immune cell activation and tolerance at specific stages of pregnancy. Further research is therefore required to fully understand the role each EV type and subtype plays in the promotion of a successful pregnancy.</p><p>STB-derived EVs are the most abundant and most studied type of reproductive EVs. Results to date suggest that STB EVs released from the late first trimester onwards have a tolerizing effect on the maternal adaptive immune response that potentiates successful pregnancies, although this is not yet proven, and no <italic>in vivo</italic> experiments have been performed. Elevated circulating levels in PE are also implicated in contributing to exacerbated maternal systemic innate immune cell activation and vascular dysfunction in PE. Therefore, measurement of STB EVs in the maternal circulation could assist with the detection and management of PE. Biomarkers currently being investigated include sFlt-1, endoglin and PLGF,<sup><xref ref-type="bibr" rid="bib122">122</xref>,<xref ref-type="bibr" rid="bib123">123</xref></sup> but as STB EVs contain thousands of proteins, many of which are unique to PE, novel and improved biomarkers could also be discovered. Further understanding of STB EVs in normal pregnancy and PE may also lead to new therapeutic options for women suffering from PE such as neutralizing specific pathogenic EVs, or removal of STB EVs through apheresis. Inhibiting maternal inflammation and reducing vascular damage could allow the pregnancy to continue long enough to avoid the need for very early pre-term delivery, removing the associated risks to both the baby and mother.</p><p>This is an exciting time for reproductive EV research. With ongoing improvements to isolation methods, discovery of better markers of EV subtypes and the development of more sensitive techniques to aid detection and characterisation of EVs, further advances will continue to be made in the field. In addition to this, more appropriate models, reflecting the <italic>in vivo</italic> cellular environment into which the reproductive EVs are released and will function, will be developed to better test their biological effects. With increasing understanding of reproductive EV function comes the great potential for the use of reproductive EVs as biomarkers and therapeutic targets in disorders of both male and female reproduction as well as obstetrics.</p></sec>
Immune cells in term and preterm labor	<p>Labor resembles an inflammatory response that includes secretion of cytokines/chemokines by resident and infiltrating immune cells into reproductive tissues and the maternal/fetal interface. Untimely activation of these inflammatory pathways leads to preterm labor, which can result in preterm birth. Preterm birth is a major determinant of neonatal mortality and morbidity; therefore, the elucidation of the process of labor at a cellular and molecular level is essential for understanding the pathophysiology of preterm labor. Here, we summarize the role of innate and adaptive immune cells in the physiological or pathological activation of labor. We review published literature regarding the role of innate and adaptive immune cells in the cervix, myometrium, fetal membranes, decidua and the fetus in late pregnancy and labor at term and preterm. Accumulating evidence suggests that innate immune cells (neutrophils, macrophages and mast cells) mediate the process of labor by releasing pro-inflammatory factors such as cytokines, chemokines and matrix metalloproteinases. Adaptive immune cells (T-cell subsets and B cells) participate in the maintenance of fetomaternal tolerance during pregnancy, and an alteration in their function or abundance may lead to labor at term or preterm. Also, immune cells that bridge the innate and adaptive immune systems (natural killer T (NKT) cells and dendritic cells (DCs)) seem to participate in the pathophysiology of preterm labor. In conclusion, a balance between innate and adaptive immune cells is required in order to sustain pregnancy; an alteration of this balance will lead to labor at term or preterm.</p>	<contrib contrib-type="author"><name><surname>Gomez-Lopez</surname><given-names>Nardhy</given-names></name><xref ref-type="aff" rid="aff1">1</xref><xref ref-type="aff" rid="aff2">2</xref></contrib><contrib contrib-type="author"><name><surname>StLouis</surname><given-names>Derek</given-names></name><xref ref-type="aff" rid="aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Lehr</surname><given-names>Marcus A</given-names></name><xref ref-type="aff" rid="aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Sanchez-Rodriguez</surname><given-names>Elly N</given-names></name><xref ref-type="aff" rid="aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Arenas-Hernandez</surname><given-names>Marcia</given-names></name><xref ref-type="aff" rid="aff1">1</xref></contrib><aff id="aff1"><label>1</label><institution>Department of Obstetrics & Gynecology and Immunology & Microbiology, Wayne State University</institution>, Detroit, MI, <country>USA</country></aff><aff id="aff2"><label>2</label><institution>Perinatology Research Branch NICHD/NIH, School of Medicine, Wayne State University</institution>, Detroit, MI, <country>USA</country></aff>	Cellular and Molecular Immunology	<sec sec-type="intro"><title>Introduction</title><p>Pregnancy demonstrates the capabilities of the human immune system. The fetus, a semi-allogeneic graft, grows and develops within the mother without succumbing to immunological rejection, a process which depends on the proper establishment of fetomaternal tolerance.<sup><xref ref-type="bibr" rid="bib1">1</xref></sup> This tolerance is initiated by the presentation of the paternal–fetal antigen from semen and is facilitated by seminal plasma factors.<sup><xref ref-type="bibr" rid="bib2">2</xref>,<xref ref-type="bibr" rid="bib3">3</xref>,<xref ref-type="bibr" rid="bib4">4</xref></sup> Antigen is processed by dendritic cells (DCs) and then presented to T cells in the uterine draining lymph nodes.<sup><xref ref-type="bibr" rid="bib3">3</xref>,<xref ref-type="bibr" rid="bib5">5</xref></sup> As a result, antigen-specific regulatory T cells (Tregs) proliferate in order to create peripheral tolerance towards fetal antigens and allow conceptus implantation.<sup><xref ref-type="bibr" rid="bib6">6</xref>,<xref ref-type="bibr" rid="bib7">7</xref></sup> Treg numbers are maintained through pregnancy, creating a tolerogenic anti-inflammatory state or hyporesponsiveness towards paternal antigens until late gestation.<sup><xref ref-type="bibr" rid="bib7">7</xref>,<xref ref-type="bibr" rid="bib8">8</xref>,<xref ref-type="bibr" rid="bib9">9</xref>,<xref ref-type="bibr" rid="bib10">10</xref></sup> During late pregnancy, we have proposed that circulating maternal leukocytes (innate and adaptive) are recruited into reproductive tissues (cervix and myometrium) and to the maternal/fetal interface (decidual tissues) by chemotactic processes,<sup><xref ref-type="bibr" rid="bib11">11</xref>,<xref ref-type="bibr" rid="bib12">12</xref>,<xref ref-type="bibr" rid="bib13">13</xref>,<xref ref-type="bibr" rid="bib14">14</xref>,<xref ref-type="bibr" rid="bib15">15</xref>,<xref ref-type="bibr" rid="bib16">16</xref></sup> where a pro-inflammatory state develops and leads to labor and delivery of the baby.<sup><xref ref-type="bibr" rid="bib17">17</xref>,<xref ref-type="bibr" rid="bib18">18</xref>,<xref ref-type="bibr" rid="bib19">19</xref></sup> It is thought that the premature activation of this pro-inflammatory pathway can lead to a breakdown of fetomaternal tolerance and play a role in the induction of labor, which subsequently can result in preterm birth.<sup><xref ref-type="bibr" rid="bib20">20</xref>,<xref ref-type="bibr" rid="bib21">21</xref>,<xref ref-type="bibr" rid="bib22">22</xref></sup></p><p>Preterm birth is a major determinant of neonatal mortality and morbidity.<sup><xref ref-type="bibr" rid="bib23">23</xref></sup> In 2011, 11.7% of all births in the United States were diagnosed as preterm.<sup><xref ref-type="bibr" rid="bib23">23</xref></sup> Among problems occurring after preterm birth are chronic respiratory illnesses, neurodevelopmental disorders and long-term cognitive impairment.<sup><xref ref-type="bibr" rid="bib24">24</xref>,<xref ref-type="bibr" rid="bib25">25</xref>,<xref ref-type="bibr" rid="bib26">26</xref></sup> However, the mechanisms that lead to preterm birth/labor are poorly understood. The main goal of this review is to summarize the innate and adaptive immune cell components that participate in term and in preterm labor, clarifying the contributions of resident, adaptive leukocytes to the physiological or pathological activation of parturition. Achieving a deeper understanding of the innate and adaptive immune cell components involved in preterm labor might allow us to develop strategies to prolong pregnancy and thereby improve pregnancy outcomes.</p></sec><sec><title>Innate immune cells in term and preterm labor</title><p>Labor is an inflammatory process.<sup><xref ref-type="bibr" rid="bib27">27</xref>,<xref ref-type="bibr" rid="bib28">28</xref></sup> A number of studies in humans and mice have reported the presence of inflammatory neutrophils and macrophages in the uterus, decidua, cervix and fetal membranes during labor.<sup><xref ref-type="bibr" rid="bib27">27</xref>,<xref ref-type="bibr" rid="bib29">29</xref>,<xref ref-type="bibr" rid="bib30">30</xref>,<xref ref-type="bibr" rid="bib31">31</xref>,<xref ref-type="bibr" rid="bib32">32</xref>,<xref ref-type="bibr" rid="bib33">33</xref>,<xref ref-type="bibr" rid="bib34">34</xref></sup> The spreading and homing of these granulocytes is facilitated by chemokines and cellular adhesion molecules.<sup><xref ref-type="bibr" rid="bib16">16</xref></sup> Additionally, mast cells are present in the uterus and cervix during late gestation and may contribute to the process of labor.<sup><xref ref-type="bibr" rid="bib35">35</xref>,<xref ref-type="bibr" rid="bib36">36</xref>,<xref ref-type="bibr" rid="bib37">37</xref>,<xref ref-type="bibr" rid="bib38">38</xref></sup> Uterine contractions, cervical ripening and dilation, and rupture of the fetal membranes (ROM) are processes which must occur in a timely fashion for a successful delivery.<sup><xref ref-type="bibr" rid="bib39">39</xref></sup> Innate immune cells have been linked to these processes by various studies, and this section aims to discuss the possible roles for these cells in the processes of term and preterm labor.</p><sec><title>Neutrophils</title><p>Neutrophil numbers are higher in the circulation of women who undergo labor than in those who do not undergo labor.<sup><xref ref-type="bibr" rid="bib40">40</xref></sup> Labor-related granulocytes are activated since they exhibit an increased ability to migrate.<sup><xref ref-type="bibr" rid="bib40">40</xref></sup> The presence of neutrophils in reproductive tissues at term and their ability to migrate to this region during labor have been well documented in both humans and rodents.<sup><xref ref-type="bibr" rid="bib11">11</xref>,<xref ref-type="bibr" rid="bib27">27</xref>,<xref ref-type="bibr" rid="bib29">29</xref>,<xref ref-type="bibr" rid="bib41">41</xref></sup> Neutrophils participate in the process of labor by releasing pro-inflammatory cytokines and secreting matrix metalloproteinases (MMPs);<sup><xref ref-type="bibr" rid="bib42">42</xref>,<xref ref-type="bibr" rid="bib43">43</xref>,<xref ref-type="bibr" rid="bib44">44</xref>,<xref ref-type="bibr" rid="bib45">45</xref>,<xref ref-type="bibr" rid="bib46">46</xref>,<xref ref-type="bibr" rid="bib47">47</xref></sup> however, their role in each anatomical compartment seems to be unique and will be discussed further below.</p><p>In the myometrium, mRNA levels of CXCL8, a neutrophil chemoattractant, are higher in women at term during labor than in women without labor,<sup><xref ref-type="bibr" rid="bib28">28</xref></sup> suggesting that neutrophils are more abundant in the myometrium during labor. However, it was recently demonstrated in a murine model of infection-induced preterm birth (intrauterine administration of lipopolysaccharide (LPS)) that there is no increase in the neutrophil numbers (Gr-1<sup>+</sup> cells) in the myometrium 6 h after LPS administration.<sup><xref ref-type="bibr" rid="bib48">48</xref></sup> Preliminary studies in our laboratory put forward evidence on the role of myometrial neutrophils during LPS-induced preterm birth. We found that LPS administration in the peritoneum causes high rates of preterm birth and this is associated with increases in proportion and in absolute numbers of neutrophils in the myometrium (NGL, unpublished data).<sup><xref ref-type="bibr" rid="bib49">49</xref></sup> This discrepancy between studies may be due to the fact that we collected tissues prior to delivery (24 h after LPS injection) instead of 6 h after the intrauterine administration of LPS. Nonetheless, the potential role of myometrial neutrophils in the process of labor needs continuing exploration.</p><p>Neutrophils are proposed to play a central role in the cervical ripening process,<sup><xref ref-type="bibr" rid="bib42">42</xref>,<xref ref-type="bibr" rid="bib43">43</xref>,<xref ref-type="bibr" rid="bib46">46</xref>,<xref ref-type="bibr" rid="bib50">50</xref></sup> although recent studies have indicated that this is likely not the case.<sup><xref ref-type="bibr" rid="bib51">51</xref>,<xref ref-type="bibr" rid="bib52">52</xref></sup> In mice, neutrophil numbers in the cervix do not vary from 15 days postcoitum (dpc) to the time of cervical ripening (late 18 dpc).<sup><xref ref-type="bibr" rid="bib52">52</xref></sup> It has been consistently reported that there are no differences in the number of cervical neutrophils between women without labor with cervical ripening and women who had not undergone the ripening process.<sup><xref ref-type="bibr" rid="bib51">51</xref></sup> However, the number of neutrophils is higher in the cervixes of women who have just completed a vaginal delivery following spontaneous labor at term than in women who were in the first trimester of pregnancy.<sup><xref ref-type="bibr" rid="bib51">51</xref></sup> This finding supports the new hypothesis that neutrophil function is required shortly after parturition, in the phase of postpartum tissue repair.<sup><xref ref-type="bibr" rid="bib52">52</xref>,<xref ref-type="bibr" rid="bib53">53</xref></sup></p><p>There have been several studies in mice implicating decidual neutrophils in the process of infection-induced preterm birth.<sup><xref ref-type="bibr" rid="bib34">34</xref>,<xref ref-type="bibr" rid="bib48">48</xref></sup> A large influx of neutrophils into the decidua and myometrium is observed during LPS-induced preterm labor and during term labor; however, this increment was not seen in a non-infectious model of preterm birth (caused by mifepristone).<sup><xref ref-type="bibr" rid="bib41">41</xref>,<xref ref-type="bibr" rid="bib54">54</xref></sup> Another study reported a sevenfold increase in neutrophils in the decidua after 6 h of intrauterine LPS administration.<sup><xref ref-type="bibr" rid="bib48">48</xref></sup> Despite these findings, the role of neutrophils as a causative agent of preterm labor is questioned since the depletion of these cells does not alter the timing or success of labor and does not prevent LPS-induced preterm birth.<sup><xref ref-type="bibr" rid="bib48">48</xref>,<xref ref-type="bibr" rid="bib52">52</xref></sup> Neutrophil depletion prior to LPS administration did, however, reduce the amount of a key pro-inflammatory cytokine, IL-1β, in the uteroplacental tissues.<sup><xref ref-type="bibr" rid="bib48">48</xref></sup> This finding is relevant since systemic administration of IL-1β leads to preterm birth in mice.<sup><xref ref-type="bibr" rid="bib55">55</xref></sup> These results suggest that neutrophils are not a necessary component in infection-induced preterm birth, yet they may be required in inflammation-induced preterm birth.</p><p>In human decidual tissues, the number of neutrophils was higher in women with preterm labor associated with chorioamnionitis than in women with term gestations (with and without labor) and in women with spontaneous preterm labor/birth without chorioamnionitis.<sup><xref ref-type="bibr" rid="bib34">34</xref></sup> The maternal origin of these decidual leukocytes (e.g., neutrophils) in preterm labor/birth associated with acute chorioamnionitis was proven by FISH.<sup><xref ref-type="bibr" rid="bib56">56</xref></sup> Maternal cells could be recruited into this maternal/fetal interface by decidual-derived chemokines, such as CXCL8.<sup><xref ref-type="bibr" rid="bib11">11</xref>,<xref ref-type="bibr" rid="bib12">12</xref></sup> Human decidual neutrophils release several inflammatory mediators and MMPs, which degrade the extracellular matrix of the fetal membranes during both term and preterm labor.<sup><xref ref-type="bibr" rid="bib44">44</xref>,<xref ref-type="bibr" rid="bib47">47</xref>,<xref ref-type="bibr" rid="bib57">57</xref>,<xref ref-type="bibr" rid="bib58">58</xref>,<xref ref-type="bibr" rid="bib59">59</xref>,<xref ref-type="bibr" rid="bib60">60</xref></sup> Taken together, these data suggest that decidual neutrophils contribute to the physiological ROM and pathological preterm premature rupture of membranes (PPROM) during term and preterm labor.</p></sec><sec><title>Macrophages</title><p>Macrophages are among the primary innate immune cells that contribute to the processes of term and preterm labor, and their roles have been studied in humans, mice and rats. Macrophages are significant during late gestation primarily due to their secretory products, which include MMPs, IL-1β, IL-6, TNF-α and nitric oxide (NO).<sup><xref ref-type="bibr" rid="bib61">61</xref>,<xref ref-type="bibr" rid="bib62">62</xref>,<xref ref-type="bibr" rid="bib63">63</xref></sup> These versatile leukocytes are being extensively studied to deepen our understanding of the parturition process. We discuss below the possible effector actions of macrophages in term and preterm labor.</p><p>Macrophages play a relevant role in the uterus during parturition. In mice, the number of uterine macrophages at 15 dpc (4 days prior to birth) was significantly higher than in non-pregnant controls though these numbers dropped to non-pregnant levels one day prior to birth.<sup><xref ref-type="bibr" rid="bib33">33</xref></sup> This trend for macrophages to decrease immediately prior to labor correlates with another study, performed in rats, which found that NO synthesis in the uterus was elevated during pregnancy but reduced during term labor.<sup><xref ref-type="bibr" rid="bib64">64</xref></sup> NO, which can be produced by macrophages,<sup><xref ref-type="bibr" rid="bib61">61</xref></sup> has been demonstrated to inhibit myometrial contractions.<sup><xref ref-type="bibr" rid="bib65">65</xref></sup> Altogether, these results suggest that a decrease in macrophages, and the resultant reduction in NO, is required for the onset of labor.</p><p>Although the aforementioned studies indicate that macrophage numbers decrease in the uterus prior to labor, a study in rats found concentrations of CCL2, a monocyte/macrophage chemoattractant, increased in the myometrium near the time of delivery in comparison to earlier points of gestation and during RU486-induced preterm labor.<sup><xref ref-type="bibr" rid="bib66">66</xref></sup> Additionally, macrophages may exert effects on the uterus during parturition through the release of pro-inflammatory cytokines, such as TNFA, which are able to upregulate uterine activation proteins,<sup><xref ref-type="bibr" rid="bib67">67</xref></sup> allowing the uterus to prepare for labor. These findings suggest that macrophages are instead recruited into the uterus during labor.</p><p>Ripening and dilation of the cervix are the next steps in parturition after the initiation of uterine contractions;<sup><xref ref-type="bibr" rid="bib39">39</xref>,<xref ref-type="bibr" rid="bib53">53</xref></sup> an inflammatory response has been associated with these processes.<sup><xref ref-type="bibr" rid="bib31">31</xref>,<xref ref-type="bibr" rid="bib51">51</xref></sup> During pregnancy at term, but before the onset of labor, women with a ripened cervix were found to have greater numbers of cervical macrophages in comparison to women who were not undergoing cervical ripening.<sup><xref ref-type="bibr" rid="bib51">51</xref></sup> A murine model similarly found an increased proportion of macrophages in the cervix the day before birth (18 dpc) in comparison to mid/late gestation (15 dpc).<sup><xref ref-type="bibr" rid="bib31">31</xref></sup> A large number of cervical macrophages was also found in antepartum and in LPS-induced preterm labor.<sup><xref ref-type="bibr" rid="bib21">21</xref>,<xref ref-type="bibr" rid="bib68">68</xref></sup> These data suggest the possible involvement of macrophages in cervical remodeling.</p><p>Although their presence suggests that macrophages play a role in the cervix during the process of labor, the characterization of these cells also supports this theory. Murine cervical macrophages expressing markers associated with adhesion (CD11b<sup>high</sup>) and migration (CD54) were lower prior to birth (18 dpc) than in mid/late gestation(15 dpc).<sup><xref ref-type="bibr" rid="bib31">31</xref></sup> However, macrophages that express markers associated with MMP activation (CD147) and cell matrix remodeling (CD169) are significantly higher at 18 dpc than at 15 dpc.<sup><xref ref-type="bibr" rid="bib31">31</xref></sup> These results suggest that cervical macrophages are probably not migrating or binding to vessels prior to birth, but are instead remodeling and degrading the extracellular matrix,<sup><xref ref-type="bibr" rid="bib31">31</xref></sup> which are important processes in ripening of the human cervix.<sup><xref ref-type="bibr" rid="bib69">69</xref>,<xref ref-type="bibr" rid="bib70">70</xref></sup> The fact that cervical leukocytes (e.g., macrophages) secrete MMP-9 at term pregnancy<sup><xref ref-type="bibr" rid="bib71">71</xref></sup> and that macrophage depletion prevents LPS-induced preterm birth in mice,<sup><xref ref-type="bibr" rid="bib21">21</xref></sup> suggests that macrophages are a main source of MMP-9 and contribute to the process of labor at both term and preterm stages. Moreover, macrophage-derived cytokines IL-1β and TNF-α increase the levels of MMP-1, MMP-3 and MMP-9,<sup><xref ref-type="bibr" rid="bib72">72</xref></sup> which may be another pathway whereby macrophages participate in the cervical ripening process. Despite the evidence above, it is important to point out that several studies have contrarily suggested that macrophages are not necessary for cervical ripening in mice,<sup><xref ref-type="bibr" rid="bib52">52</xref>,<xref ref-type="bibr" rid="bib68">68</xref>,<xref ref-type="bibr" rid="bib73">73</xref>,<xref ref-type="bibr" rid="bib74">74</xref></sup> but play a role in postpartum repair.<sup><xref ref-type="bibr" rid="bib52">52</xref>,<xref ref-type="bibr" rid="bib53">53</xref></sup> Further research on the human cervix during labor and preterm labor must be performed in order to come to a definitive conclusion.</p><p>Macrophages could also play a role in the rupture of the fetal membranes since macrophages are recruited by these tissues<sup><xref ref-type="bibr" rid="bib11">11</xref></sup> and produce MMP-9.<sup><xref ref-type="bibr" rid="bib63">63</xref></sup> MMP-9 concentrations are significantly increased in the fetal membranes during labor, preterm labor and PPROM,<sup><xref ref-type="bibr" rid="bib58">58</xref>,<xref ref-type="bibr" rid="bib75">75</xref>,<xref ref-type="bibr" rid="bib76">76</xref>,<xref ref-type="bibr" rid="bib77">77</xref></sup> which directly links this enzyme to physiological ROM and pathological PPROM. Additionally, pro-inflammatory cytokines released by macrophages during labor can regulate the further release of MMPs<sup><xref ref-type="bibr" rid="bib78">78</xref></sup> by the fetal membranes, suggesting another mechanism whereby macrophages may contribute to ROM and PPROM.</p><p>Macrophages also reside in the decidua near or during the time of labor.<sup><xref ref-type="bibr" rid="bib14">14</xref>,<xref ref-type="bibr" rid="bib34">34</xref></sup> In human decidual tissues, macrophage proportions are higher at term than in preterm gestations without labor.<sup><xref ref-type="bibr" rid="bib14">14</xref></sup> Macrophage tissue density is even greater in decidua from women who delivered term and preterm with labor in comparison to women who delivered at term without labor.<sup><xref ref-type="bibr" rid="bib34">34</xref></sup> In mice, the proportion of decidual macrophages increases prior to birth (18 dpc) in comparison to mid/late gestation (15 dpc).<sup><xref ref-type="bibr" rid="bib41">41</xref></sup> Altogether, these results suggest that decidual macrophages have a role prior to the onset of labor.</p><p>Macrophages are also implicated in the etiology of preterm labor since CCL2 concentrations are increased in the amniotic fluid of women delivering preterm, both in the presence and absence of intra-amniotic infection, in comparison to women delivering at term.<sup><xref ref-type="bibr" rid="bib79">79</xref></sup> One of the most significant indicators of the role of macrophages in preterm labor was the demonstration that the depletion of macrophages in pregnant mice protected these animals from LPS-induced preterm birth.<sup><xref ref-type="bibr" rid="bib21">21</xref></sup> Ultimately, macrophages are potentially involved in several processes during parturition. The precise role of this cell type in labor remains disputed, yet much evidence gives credibility to their putative roles. Further studies are required to fully elucidate the roles of macrophages in the physiological process of labor and the pathological induction of preterm labor.</p><p>Currently, we are investigating the role of macrophages during preterm birth using animal models. Our preliminary data suggest that the plasticity of these cells at the maternal/fetal interface is unique, and that besides participating in the process of labor, macrophages play a central role in the maintenance of fetomaternal tolerance during late pregnancy.</p></sec><sec><title>Mast cells</title><p>Mast cells (MCs) are also important innate immune effector cells during late gestation and labor due to their secretion of mediators.<sup><xref ref-type="bibr" rid="bib16">16</xref>,<xref ref-type="bibr" rid="bib80">80</xref></sup> Fast-acting MC mediators are histamine, serotonin, heparin, proteoglycans, proteases, prostaglandins and leukotrienes.<sup><xref ref-type="bibr" rid="bib81">81</xref></sup> MCs also secrete the long-term modulators IL-1β, IL-3, IL-5, IL-6 and TNF-α.<sup><xref ref-type="bibr" rid="bib36">36</xref></sup> Moreover, human MCs induce the expression of endothelial adhesion molecules,<sup><xref ref-type="bibr" rid="bib82">82</xref></sup> and express several chemokine receptors.<sup><xref ref-type="bibr" rid="bib83">83</xref></sup> This combination of MC recruitment and up-regulation of cellular adhesion molecules allows MCs to localize within the uterus and cervix, where they may play a role in the development of a pro-inflammatory environment. Due to their presence and actions in cervical tissue during late gestation, MCs and histamine have been associated with the stimulation of cervical contractility;<sup><xref ref-type="bibr" rid="bib38">38</xref></sup> however, MCs have been detected in higher proportions in postpartum than during late gestational cervix which indicates a greater role for this cell type in postpartum uterine cervical repair than during labor.<sup><xref ref-type="bibr" rid="bib84">84</xref></sup> We therefore focus the section below on discussing the role of MCs and their mediators in the uterus during term and preterm labor.</p><p>MC degranulation releases mediators which likely play major roles in the process of labor by remodeling uterine smooth muscle cells and stimulating uterine contractions.<sup><xref ref-type="bibr" rid="bib35">35</xref>,<xref ref-type="bibr" rid="bib36">36</xref>,<xref ref-type="bibr" rid="bib37">37</xref>,<xref ref-type="bibr" rid="bib85">85</xref>,<xref ref-type="bibr" rid="bib86">86</xref></sup> The release of histamine and serotonin has been linked to uterine contractility since MCs reside adjacent to smooth muscle in the myometrium.<sup><xref ref-type="bibr" rid="bib35">35</xref>,<xref ref-type="bibr" rid="bib87">87</xref></sup> Indeed, during murine pregnancy MCs are more abundant in the myometrium than in the endometrium.<sup><xref ref-type="bibr" rid="bib88">88</xref></sup> The degranulation of MCs <italic>in vitro</italic>, utilizing compound 48/80, induces greater uterine contractility in tissue from late gestation than in non-pregnant uterine tissue.<sup><xref ref-type="bibr" rid="bib36">36</xref></sup> In addition, the tissue density of human uterine MCs is greater during pregnancy than in the non-pregnant state, which also suggests that uterine MCs modulate myometrial contractility during late pregnancy.<sup><xref ref-type="bibr" rid="bib89">89</xref></sup> A link between MCs and allergy has been suggested as a mechanism of preterm labor/birth<sup><xref ref-type="bibr" rid="bib37">37</xref>,<xref ref-type="bibr" rid="bib90">90</xref></sup> since MCs are one of the cells effecting immediate hypersensitivity reactions and allergic disease<sup><xref ref-type="bibr" rid="bib81">81</xref></sup> and allergies play a central role in uterine contractions.<sup><xref ref-type="bibr" rid="bib37">37</xref>,<xref ref-type="bibr" rid="bib90">90</xref></sup> Furthermore, pre-treatment of guinea pigs with a histamine H<sub>1</sub> receptor antagonist decreased the rate of preterm birth induced by an allergic reaction.<sup><xref ref-type="bibr" rid="bib91">91</xref></sup> This finding suggests a vital role for histamine, and therefore MCs, in the processes of term and preterm labor.</p><p>A recent study contradicts the notion of MC involvement in labor. This study found that, in human myometrial tissues, the abundance of MCs was similar at mid-pregnancy and during labor.<sup><xref ref-type="bibr" rid="bib86">86</xref></sup> In MC deficient Kit<sup>W-sh</sup> mice, labor still occurs and leukocyte recruitment into the myometrium is not different from wild-type controls.<sup><xref ref-type="bibr" rid="bib86">86</xref></sup> A possible explanation is that MCs are not the sole leukocyte recruiters,<sup><xref ref-type="bibr" rid="bib16">16</xref></sup> and the pro-inflammatory cascade can be upregulated by other leukocyte subpopulations even in the absence of MCs. Further research is needed in order to clarify the role of mast cells during term and preterm labor.</p></sec></sec><sec><title>Adaptive immune cells in term and preterm labor</title><p>The adaptive immune system creates memory and responds to specific antigens. During pregnancy, the adaptive immune limbs of both the mother and the fetus must tolerate each other in order to maintain pregnancy until term. A breakdown of this fetomaternal tolerance may lead to labor. In term pregnancy, lack of the tolerogenic state results in physiologic labor. However, a premature retreat of this tolerogenic state might lead to preterm labor.</p><sec><title>T cells</title><p>During pregnancy, maternal T cells recognize fetal antigens through interactions with antigen-presenting cells.<sup><xref ref-type="bibr" rid="bib92">92</xref></sup> Fetal antigen-specific T cells maintain fetomaternal immune tolerance across pregnancy.<sup><xref ref-type="bibr" rid="bib7">7</xref></sup> We previously proposed that maternal circulating T cells infiltrate into the maternal/fetal interface prior to delivery and during labor at term.<sup><xref ref-type="bibr" rid="bib11">11</xref>,<xref ref-type="bibr" rid="bib93">93</xref></sup> Decidual T cells are activated and have both a regulatory and an effector phenotype.<sup><xref ref-type="bibr" rid="bib94">94</xref>,<xref ref-type="bibr" rid="bib95">95</xref>,<xref ref-type="bibr" rid="bib96">96</xref>,<xref ref-type="bibr" rid="bib97">97</xref></sup> The next section further addresses the putative roles of specific T-cell subsets during late pregnancy and in term and preterm labor.</p><sec><title><italic>Effector T cells</italic></title><p>We recently provided evidence that decidual CD4<sup>+</sup> T cells are involved in term parturition.<sup><xref ref-type="bibr" rid="bib14">14</xref></sup> Specifically, we demonstrated that decidual CD4<sup>+</sup> T cells are more abundant in term than in preterm gestations without labor. These T cells express CD45RO, but not CD45RA, which suggests that they are memory cells that were generated early in pregnancy when fetal–antigen presentation occurs.<sup><xref ref-type="bibr" rid="bib7">7</xref>,<xref ref-type="bibr" rid="bib14">14</xref>,<xref ref-type="bibr" rid="bib92">92</xref></sup> We also demonstrated that decidual CD4<sup>+</sup> T cells express IL-1β, TNF-α and MMP-9 during spontaneous labor at term.<sup><xref ref-type="bibr" rid="bib14">14</xref></sup> The fact that decidual T cells express activation markers such as CD25<sup><xref ref-type="bibr" rid="bib98">98</xref></sup> and labor mediators implicated in both term and preterm labor<sup><xref ref-type="bibr" rid="bib17">17</xref>,<xref ref-type="bibr" rid="bib29">29</xref>,<xref ref-type="bibr" rid="bib55">55</xref>,<xref ref-type="bibr" rid="bib58">58</xref>,<xref ref-type="bibr" rid="bib75">75</xref>,<xref ref-type="bibr" rid="bib77">77</xref>,<xref ref-type="bibr" rid="bib99">99</xref>,<xref ref-type="bibr" rid="bib100">100</xref>,<xref ref-type="bibr" rid="bib101">101</xref>,<xref ref-type="bibr" rid="bib102">102</xref></sup> suggests that the adaptive limb of the immune system participates during labor.</p><p>Additionally, we demonstrated that during term labor T cells are preferentially recruited into the rupture zone of the fetal membranes by chemotactic processes facilitated by CXCL10 and CCL5.<sup><xref ref-type="bibr" rid="bib13">13</xref>,<xref ref-type="bibr" rid="bib14">14</xref>,<xref ref-type="bibr" rid="bib93">93</xref></sup> However, T-cell attraction to the rupture zone was significantly diminished in premature ROM cases.<sup><xref ref-type="bibr" rid="bib13">13</xref></sup> These data suggest that T-cell recruitment into the maternal/fetal interface is required for term pregnancy, and the dysregulation of this recruitment may lead to pathological rupture of membranes.</p><p>Th17 cells (CD3<sup>+</sup>CD4<sup>+</sup>IL-17A<sup>+</sup>) also congregate in human decidua,<sup><xref ref-type="bibr" rid="bib103">103</xref></sup> and their tissue density is higher in cases of chorioamnionitis than in cases without chorioamnionitis.<sup><xref ref-type="bibr" rid="bib104">104</xref></sup> This finding further supports the idea that pro-inflammatory adaptive immune cells at the maternal/fetal interface are associated with chorioamnionitis, which can lead to preterm labor/birth. Studies in our laboratory are currently exploring the potential role for this T-cell subset in preterm labor using LPS-induced and RU486-induced preterm birth models.</p><p>Fetal T cells might also play a role during preterm labor. Memory fetal T cells (CD45RO<sup>+</sup>RA<sup>−</sup>) are present in higher proportions in cord blood from cases of preterm labor compared to term labor.<sup><xref ref-type="bibr" rid="bib105">105</xref></sup> Fetal T cells are also activated (CD25<sup>+</sup>CD69<sup>+</sup>) during preterm labor.<sup><xref ref-type="bibr" rid="bib106">106</xref></sup> Indeed, acute chorioamnionitis, a leading cause of preterm deliveries, is associated with an increase in cord blood T-cell chemokines (CXCL9, -10 and -11).<sup><xref ref-type="bibr" rid="bib107">107</xref></sup> These results suggest that fetal T cells can contribute to the pathophysiology of preterm labor.</p><p>Cytotoxic T cells (CTLs) are present at the maternal/fetal interface in term gestations in the absence of labor, where they express perforin and granzyme B.<sup><xref ref-type="bibr" rid="bib95">95</xref>,<xref ref-type="bibr" rid="bib97">97</xref>,<xref ref-type="bibr" rid="bib108">108</xref></sup> In placenta, CTLs are abundant in cases with villitis of unknown etiology and express T-cell chemokine receptors (CXCR3 and CCR5).<sup><xref ref-type="bibr" rid="bib107">107</xref></sup> In peripheral circulation, CD300a<sup>+</sup> CTLs have an effector memory phenotype, and their proportion is higher in women with chronic chorioamnionitis than in women without this lesion.<sup><xref ref-type="bibr" rid="bib109">109</xref></sup> Taken together, these data suggest that CTLs may participate in pathological inflammation associated with preterm birth, but their role during spontaneous labor at term and preterm requires further exploration.</p></sec></sec><sec><title>Tregs</title><p>There are two main Treg subsets: thymic Tregs (tTregs) and extrathymic or peripheral Tregs (pTregs). During pregnancy, CD4<sup>+</sup> pTregs have been categorized into four subsets: DR<sup>high+</sup>CD45RA<sup>−</sup>, DR<sup>low+</sup>CD45RA<sup>−</sup>, DR<sup>−</sup>CD45RA<sup>−</sup> and naïve DR<sup>−</sup>CD45RA<sup>+</sup>.<sup><xref ref-type="bibr" rid="bib110">110</xref></sup> The proportion of each subset seems to be relevant in the pathophysiology of pregnancy complications such as preterm labor. Women with preterm labor have a reduced proportion of naïve DR<sup>−</sup>CD45RA<sup>+</sup> Tregs, accompanied by higher proportions of DR<sup>−</sup>CD45RA<sup>−</sup> and DR<sup>low+</sup>CD45RA<sup>−</sup> Tregs within their total pTreg pool.<sup><xref ref-type="bibr" rid="bib110">110</xref>,<xref ref-type="bibr" rid="bib111">111</xref></sup> Indeed, the suppressive activity of pTregs is strongly reduced in term and preterm labor,<sup><xref ref-type="bibr" rid="bib111">111</xref></sup> which is correlated with a reduction in the expression of HLA-DR in preterm cases.<sup><xref ref-type="bibr" rid="bib112">112</xref></sup> This suggests that the lack of suppressive function during late pregnancy could trigger the onset of parturition at term and preterm gestations.<sup><xref ref-type="bibr" rid="bib113">113</xref></sup></p><p>At term pregnancy, Tregs are found at the maternal/fetal interface, have a unique phenotype (CD4<sup>+</sup>CD25<sup>bright</sup>FoxP3<sup>+</sup>CD69<sup>+</sup>HLA-DR<sup>+</sup>CTLA-4<sup>+</sup>), and exhibit suppressive function <italic>in vitro</italic>.<sup><xref ref-type="bibr" rid="bib108">108</xref>,<xref ref-type="bibr" rid="bib114">114</xref></sup> However, the role of decidual Tregs remains undetermined. Currently, we are investigating the function and phenotypic characteristics of these cells during term and preterm labor.</p><p>Mechanistic studies have successfully demonstrated that the systemic ablation of Tregs by targeting FoxP3<sup>+</sup> cells leads to pregnancy failure during early gestation.<sup><xref ref-type="bibr" rid="bib6">6</xref>,<xref ref-type="bibr" rid="bib7">7</xref>,<xref ref-type="bibr" rid="bib115">115</xref></sup> A separate study suggested that regulatory T cells are not required in late pregnancy;<sup><xref ref-type="bibr" rid="bib116">116</xref></sup> however, the targeted depletion of CD25<sup>+</sup> cells is not specific for Tregs. Preliminary data from our laboratory demonstrates that systemic depletion of FoxP3<sup>+</sup> cells during late gestation does lead to pregnancy complications (NGL, unpublished data).</p><p>Additional unpublished data from our laboratory demonstrate that LPS-induced preterm labor causes an expansion of CD4<sup>+</sup> Tregs in the spleen and thymus but a reduction of uterine CD4<sup>+</sup> Tregs (NGL, unpublished data).<sup><xref ref-type="bibr" rid="bib117">117</xref></sup> We also found that the administration of vaginal progesterone, a clinical strategy to prevent preterm birth in women with a short cervix,<sup><xref ref-type="bibr" rid="bib118">118</xref></sup> increases the proportion of decidual CD4<sup>+</sup> Tregs (NGL, unpublished data).<sup><xref ref-type="bibr" rid="bib119">119</xref></sup> Altogether, these data suggest that preterm birth is characterized by altered proportions of CD4<sup>+</sup> Tregs at the maternal/fetal interface and that natural progesterone can restore the number of these cells during late pregnancy, preventing preterm birth.</p></sec><sec><title>B cells</title><p>A few years ago, we suggested a role for B cells during term labor since the fetal membranes from laboring women who delivered at term exhibit B-cell attraction <italic>in vitro</italic>.<sup><xref ref-type="bibr" rid="bib12">12</xref>,<xref ref-type="bibr" rid="bib16">16</xref></sup>Current preliminary data demonstrate that B cells are indeed present in the decidua and cord blood at term and preterm stages (NGL, unpublished data). However, the role of B cells in the processes of term and preterm labor is still under investigation.</p><p>Several studies have linked various B cell subsets to pregnancy. B1 cells are present in lower proportions in maternal blood during pregnancy and return to non-pregnant proportions post-partum.<sup><xref ref-type="bibr" rid="bib120">120</xref></sup> However, B2 cell frequencies are unchanged by pregnancy in the peripheral blood of women.<sup><xref ref-type="bibr" rid="bib120">120</xref></sup> Regulatory B cells exist during early and late gestation and release IL-10.<sup><xref ref-type="bibr" rid="bib121">121</xref>,<xref ref-type="bibr" rid="bib122">122</xref></sup> Regulatory B cells are potential immune players in the development of immunological tolerance, and their presence during mid-gestation may be important in sustaining pregnancy until labor. B10 cells suppress TNF-α secretion by CD4<sup>+</sup> T cells during pregnancy,<sup><xref ref-type="bibr" rid="bib122">122</xref></sup> and this may regulate the inflammatory state prior to labor. Furthermore, B cells isolated from term placentas produce increased amounts of asymmetric IgG upon stimulation by IL-6, IL-10 and IL-4.<sup><xref ref-type="bibr" rid="bib123">123</xref></sup> Therefore, an abnormal disruption of B cell-derived cytokine and asymmetric antibody production could play a part in disturbing fetal tolerance and possibly in eliciting preterm labor.</p></sec></sec><sec><title>Bridges between the innate and adaptive immune systems in term and preterm labor</title><p>Immune tolerance involves both the innate and adaptive immune systems. Therefore, fetomaternal tolerance must involve the participation of immune cells that bridge the innate and adaptive immune systems, such as DCs and natural killer T (NKT) cells. The roles of these cells during late gestation, labor and preterm labor are discussed below.</p><sec><title>NKT cells in term and preterm labor</title><p>NKT cells are a unique lymphocyte subpopulation that express markers and characteristics of both the adaptive and innate limbs of the immune system. NKT cells recognize lipid antigens presented by the non-polymorphic CD1D molecule,<sup><xref ref-type="bibr" rid="bib124">124</xref></sup> which is expressed by trophoblast cells, placenta, and choriocarcinoma cell lines.<sup><xref ref-type="bibr" rid="bib125">125</xref>,<xref ref-type="bibr" rid="bib126">126</xref></sup> There are two types of NKT cells, type I and type II.<sup><xref ref-type="bibr" rid="bib127">127</xref></sup> Type I NKT or invariant NKT (iNKT) cells can be activated by the marine-derived glycolipid α-galactosylceramide,<sup><xref ref-type="bibr" rid="bib124">124</xref></sup> and this molecule has been utilized to explore the role of iNKT cell activation during pregnancy <italic>in vivo</italic>.<sup><xref ref-type="bibr" rid="bib128">128</xref>,<xref ref-type="bibr" rid="bib129">129</xref>,<xref ref-type="bibr" rid="bib130">130</xref></sup> Therefore, this section will primarily focus on iNKT cells.</p><p>During murine pregnancy, NK1.1<sup>+</sup>TCRαβ<sup>+</sup> NKT cells have been observed in the decidua and uterus primarily in early gestation, although they are still present near term (14–18 dpc).<sup><xref ref-type="bibr" rid="bib129">129</xref>,<xref ref-type="bibr" rid="bib131">131</xref></sup> Murine NK1.1<sup>+</sup>CD3<sup>+</sup> NKT cell proportions were higher in the livers of pregnant mice during late gestation (16 dpc) than in non-pregnant controls.<sup><xref ref-type="bibr" rid="bib132">132</xref></sup> iNKT cells can secrete large quantities of IL-4 (Th2) and IFN-γ (Th1) upon TCR activation,<sup><xref ref-type="bibr" rid="bib133">133</xref></sup> and their activation has been shown to have roles in activating NK cells, B cells and T cells.<sup><xref ref-type="bibr" rid="bib134">134</xref></sup> As a result of their immunological effects and their presence during gestation, iNKT cells may participate in pathological or physiological responses during late gestation.</p><p>iNKT cells have been involved in the induction of an increased cytotoxic state during human pregnancy complications such as preeclampsia.<sup><xref ref-type="bibr" rid="bib135">135</xref></sup> The proportion of iNKT cells expressing activation markers (CD69<sup>+</sup>), perforin and IFN-γ is increased in the blood of pre-eclamptic women in comparison to pregnant women without this pathology.<sup><xref ref-type="bibr" rid="bib135">135</xref></sup> Although the aforementioned study did not address preterm labor, these results indicate that pro-inflammatory iNKT cells are increased during late gestational pregnancy complications; this Th1-like environment could potentially disrupt fetomaternal tolerance and lead to preterm labor.</p><p>The role of iNKT cells in the induction of LPS-induced preterm birth has been studied in iNKT cell deficient (Jα18<sup>−/−</sup>) mice.<sup><xref ref-type="bibr" rid="bib136">136</xref></sup> The injection of LPS at 15 dpc caused preterm birth in wild-type mice but not in iNKT cell-deficient mice,<sup><xref ref-type="bibr" rid="bib136">136</xref></sup> suggesting that iNKT cells modulate the process of labor induced by microbial products. Conversely, the stimulation of iNKT cells <italic>in vivo</italic> by injection of α-galactosylceramide during late gestation (16 dpc) induced early preterm birth (17 dpc),<sup><xref ref-type="bibr" rid="bib128">128</xref></sup> which may be due to an expansion of NK1.1<sup>+</sup>TCRαβ<sup>+</sup> NKT cells in the uterus.<sup><xref ref-type="bibr" rid="bib129">129</xref></sup> In contrast, studies conducted in our laboratory found that the activation of iNKT cells during late gestation (16 dpc) through α-galactosylceramide administration induced late preterm birth (birth at 18 dpc, 28 h post-injection), which is relevant since 70% of all preterm births in woman fall under this category (NGL, unpublished data).<sup><xref ref-type="bibr" rid="bib130">130</xref></sup> Current experiments in our laboratory are addressing the immune mechanisms whereby iNKT cell activation leads to late preterm birth.</p></sec><sec><title>DCs in term and preterm labor</title><p>DCs are specialized in antigen recognition and presentation. DCs exhibit properties that include induction of antigen-specific T-cell activation, T-cell suppression, Treg generation and peripheral tolerance.<sup><xref ref-type="bibr" rid="bib137">137</xref>,<xref ref-type="bibr" rid="bib138">138</xref></sup> Lymphoid CD8α<sup>+</sup> DCs (DCs1) induce a Th1 response whereas myeloid CD8α DCs (DCs2) elicit a Th2 response.<sup><xref ref-type="bibr" rid="bib139">139</xref>,<xref ref-type="bibr" rid="bib140">140</xref></sup> A third type of DCs is the inflammatory DCs which initiates a Th1 response as well.<sup><xref ref-type="bibr" rid="bib141">141</xref></sup> Due to their immunomodulatory properties, these three subsets of DCs are relevant in the study of fetomaternal tolerance and inflammation during labor and preterm labor.</p><p>DCs contribute to fetomaternal tolerance during early pregnancy.<sup><xref ref-type="bibr" rid="bib142">142</xref></sup> In mice, uterine DCs have a DC2 phenotype at 15 dpc,<sup><xref ref-type="bibr" rid="bib143">143</xref></sup> which suggests that these cells contribute to the tolerogenic state by inducing a local anti-inflammatory (Th2) response during late gestation. Later in pregnancy (17.5 dpc), the predominant DC subset in the uterus is CD11c<sup>+</sup>CD8α<sup>−</sup>MHCII<sup>−</sup> (immature phenotype).<sup><xref ref-type="bibr" rid="bib144">144</xref></sup> The fact that immature DCs express the anti-inflammatory cytokine IL-10,<sup><xref ref-type="bibr" rid="bib144">144</xref></sup> a potential early biomarker of preterm birth,<sup><xref ref-type="bibr" rid="bib145">145</xref></sup> suggests that these cells may participate in the etiology of preterm labor. Moreover, in T and B cell-deficient mice (<italic>Rag1<sup>−/−</sup></italic>) injected with LPS to induce preterm birth, uterine DC activation was observed,<sup><xref ref-type="bibr" rid="bib146">146</xref></sup> suggesting the participation of DCs in the induction of labor. Further research is needed in order to establish a role for DCs during late gestation, labor and preterm labor.</p></sec></sec><sec sec-type="conclusions"><title>Conclusions</title><p>During late pregnancy, paternal–fetal antigen-specific memory T cells (including Tregs) participate in the maintenance of fetomaternal peripheral tolerance. Collectively, these cells create an anti-inflammatory environment which will sustain pregnancy. We suggest the following pathway could lead to labor: (1) activation of innate and adaptive immune cells increases their migratory ability; (2) reproductive tissues and the maternal/fetal interface actively recruit the activated cells through the release of chemokines such as CXCL10, CXCL8, CCL2 and CCL5; and (3) infiltrating leukocytes amplify the pro-inflammatory microenvironment at the maternal/fetal interface leading to labor. A triggered stimulus (e.g., infection/inflammation, sterile inflammation, stress, <italic>etc.</italic>) can cause the premature activation of this pathway, eliciting a shift from an anti-inflammatory to a pro-inflammatory microenvironment and consequently preterm labor (<xref ref-type="fig" rid="fig1">Figure 1</xref>).</p><p>An overview of the innate and adaptive immune cells in reproductive tissues and at the maternal/fetal interface during term and preterm labor is shown in <xref ref-type="fig" rid="fig2">Figure 2</xref>. Neutrophils are present in the cervix, myometrium, fetal membranes and decidua at term pregnancy; however, their density increases in the myometrium and decidua in term labor and infection-associated preterm labor. Neutrophils are present in the cervix and participate in the repair process during the postpartum period. Macrophages are present in the cervix, myometrium, fetal membranes and decidua at term pregnancy and their density increases in all these tissues, except the cervix, during term and preterm labor. Cervical macrophages also seem to participate in postpartum repair processes. Mast cells are found in cervical and myometrial tissues during late gestation; however, their roles during term and preterm labor are unclear. Effector CD4<sup>+</sup> T cells are present in decidual tissues during term labor, and decidual Th17 cells also seem to be involved in the pathology of preterm labor. CTLs are found in term pregnancy and in placental tissues in cases with villitis of unknown etiology; however, their role during labor is unknown. The fetal membranes exhibit B-cell recruitment during term labor, and B cells are found in decidual tissues and cord blood; however, their role in preterm labor is still under investigation. Finally, in myometrial tissues, NKT cell and DC activation seem to be involved in the pathophysiology of preterm labor.</p><p>Overall, collaboration between the innate and adaptive limbs of the immune system is required to sustain pregnancy until term. A disruption of either limb at term may lead to physiological labor, and an untimely disruption could result in pathological preterm labor. Research targeting the immune cells involved in the process of labor might reveal new strategies to prevent preterm labor and consequently preterm birth.</p></sec>
Inhibition of heat-induced apoptosis in rat small intestine and IEC-6 cells through the AKT signaling pathway	<sec><title>Background</title><p>As the world warms up, heat stress is becoming a major cause of economic loss in the livestock industry. Long-time exposure of animals to hyperthermia causes extensive cell apoptosis, which is harmful to them. AKT and AKT-related serine–threonine kinases are known to be involved in signaling cascades that regulate cell survival, but the mechanism remains elusive. In the present study, we demonstrate that phosphoinositide 3-kinase (PI3K) /AKT signal pathway provides protection against apoptosis induced by heat stress to ascertain the key point for treatment.</p></sec><sec><title>Results</title><p>Under heat stress, rats showed increased shedding of intestinal epithelial cells. These rats also had elevated levels of serum cortisol and improved expression of heat shock proteins (Hsp27, Hsp70 and Hsp90) in response to heat stress. Apoptosis analysis by TUNEL assay revealed a higher number of villi epithelial cells that were undergoing apoptosis in heat-treated rats than in the normal control. This is supported by gene expression analysis, which showed an increased ratio of Bax/Bcl-2 (p < 0.05), an important indicator of apoptosis. During heat-induced apoptosis, more AKTs were activated, showing increased phosphorylation. An increase of BAD phosphorylation, which is an inhibitory modification, ensued. In rat IEC-6 cell line, a significant higher level of AKT phosphorylation was observed at 2 h after heat exposure. This coincided with a marked reduction of apoptosis.</p></sec><sec><title>Conclusion</title><p>Together, these results suggest that heat stress caused damages to rat jejunum and induced apoptosis to a greater degree. HSPs and pro-survival factors were involved in response to heat stress. Among them, AKT played a key role in inhibiting heat-induced apoptosis.</p></sec>	<contrib contrib-type="author" id="A1"><name><surname>Gao</surname><given-names>Zhimin</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>gaozhimin123@126.com</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Liu</surname><given-names>Fenghua</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>iufenghua1209@126.com</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Yin</surname><given-names>Peng</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>leilaleila@126.com</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Wan</surname><given-names>Changrong</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>changrongwan@126.com</email></contrib><contrib contrib-type="author" id="A5"><name><surname>He</surname><given-names>Shasha</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>hss_sara@126.com</email></contrib><contrib contrib-type="author" id="A6"><name><surname>Liu</surname><given-names>Xiaoxi</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>liuxiaoxi_06@163.com</email></contrib><contrib contrib-type="author" id="A7"><name><surname>Zhao</surname><given-names>Hong</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>zhaohong2556@126.com</email></contrib><contrib contrib-type="author" id="A8"><name><surname>Liu</surname><given-names>Tao</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>liutaoshuier168@163.com</email></contrib><contrib contrib-type="author" corresp="yes" equal-contrib="yes" id="A9"><name><surname>Xu</surname><given-names>Jianqin</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>xujianqincau@126.com</email></contrib><contrib contrib-type="author" corresp="yes" equal-contrib="yes" id="A10"><name><surname>Guo</surname><given-names>Shining</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>shining@scau.edu.cn</email></contrib>	BMC Veterinary Research	<sec><title>Background</title><p>Heat stress is a common stressful factor that affects many biological systems. Research over the past decade has demonstrated that hyperthermia causes various damages to the animal body, including injuries in the central nervous system [<xref ref-type="bibr" rid="B1">1</xref>] and adrenal glands [<xref ref-type="bibr" rid="B2">2</xref>], reduction of thyroid physiology in lactating cows [<xref ref-type="bibr" rid="B3">3</xref>], and gastrointestinal hyperpermeability [<xref ref-type="bibr" rid="B4">4</xref>]. The integrity (both structural and functional) of the small intestine is essential for absorption of nutrients. However, it can also be jeopardized by hyperthermia. Especially, hyperthermia causes damages to the tips of intestinal villi, where epithelial cells renewal requires a large amount of energy [<xref ref-type="bibr" rid="B5">5</xref>]. Under high temperature, the blood flow to the small intestine is reduced significantly to increase that to essential organs such as brain and cardiac. This greatly impairs the small intestinal villus epithelial cells [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B6">6</xref>], and induces excessive apoptosis of them.</p><p>Apoptosis, also known as programmed cell death, is a physiological suicide mechanism by which cells die under strict control [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B8">8</xref>]. It is characterized by specific features, including nuclear fragmentation, DNA fragmentation, and apoptotic body formation. The formed apoptotic bodies are rapidly phagocytosed by neighboring cells or macrophages, without causing a damaging inflammatory response [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. A lot of researches demonstrate that as a critical media of apoptosis, heat stress would induce apoptosis in cells [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B12">12</xref>]. Although apoptosis is a normal physiological process, in excess it is pathologic [<xref ref-type="bibr" rid="B13">13</xref>].</p><p>PI3K/AKT signaling has been reported to block apoptosis induced by diverse apoptotic stimuli, and promotes cell survival in a variety of apoptotic paradigms [<xref ref-type="bibr" rid="B14">14</xref>-<xref ref-type="bibr" rid="B16">16</xref>]. However, little is known about its role in heat-induced apoptosis. In this signaling pathway, AKT is the primary mediator. It has a number of downstream substrates that may contribute to tumor genesis. In the presence of survival factors, AKT becomes activated, which in turn phosphorylates and inactivates components of the apoptotic machinery, such as Bad. Bad and other Bcl-2 family members are known to function as critical regulators of apoptosis pathways, acting to either inhibit (Bcl-2, Bcl-xl) or promote (Bak, Bad) cell death [<xref ref-type="bibr" rid="B17">17</xref>]. Thus, AKT may serve to repress apoptosis by inhibiting the activities of pro-apoptotic proteins.</p><p>From our previous study on heat-stress, we hypothesized that cell apoptosis in small intestine play crucial role under state of heat-tress. To investigate heat-induced apoptosis in rat small intestine and IEC-6 cells, and to examine the role of AKT in this apoptosis, the rats were simulated in hyperthermia. After heat exposure, the morphological changes were detected by electron microscopes. Apoptotic cells were examined by TUNEL assay. Our results suggest an effect of AKT on suppressing apoptosis triggered by heat stress, so that AKT would be as a target for treatment for a more general aim of this study is to improve animal growth.</p></sec><sec sec-type="methods"><title>Method</title><sec><title>Animals</title><p>All experimental protocols were approved by the Committee for the Care and Use of Experimental Animals at Beijing University of Agriculture.</p><p>Twelve healthy male Sprague Dawley (SD) rats (BW200 ± 20 g) were obtained from Beijing Vital River Laboratory, Animal Technology Co., Beijing, China, and raised (25°C, 60% relative humidity) for 7 days freedom to water and food. Then the rats were divided into two groups, control and heat stress group.</p></sec><sec><title>Treatment and sampling</title><p>Rats in the control group were raised at an atmosphere of 25°C, 60% relative humidity (RH); the heat stress group was housed at 40°C, 60% RH between 11:00 am and 1:00 pm daily for three consecutive days. Rats from both groups were sacrificed by broken neck immediately after the heat exposure period.</p><p>Rat body temperatures were recorded after heat exposure using a thermistor probe connected to a digital thermometer. Their body weights were also recorded. Blood samples were immediately collected after execution, and centrifuged at 3,000 × g for 10 min. Intestinal tissue samples of duodenum, jejunum and ileum were collected afterwards. The samples of each tissue were divided into two parts: One was fixed in 10% buffered formalin phosphate for histological analysis; the other was stored at -80°C.</p></sec><sec><title>IEC-6 cell culture, cell treatment and morphology observation</title><p>IEC-6 cells (CRL21592, obtained from Peking Union Medical College) were grown in Dulbecco’s Modified Eagle Medium (DMEM) containing 5% (v/v) fetal bovine serum (HyClone, USA), 2 mg/L insulin, 50 IU/ml penicillin and 50 mg/ml streptomycin. The cells in control group were strictly regulated at 37°C and 5% CO<sub>2</sub>, while cells in heat-stressed group were exposed to 42°C and 5% CO<sub>2</sub> for 15 minutes, 30 minutes, 1 hour, 2 hours, 4 hours and 8 hours. Changes in cell morphology were observed using phase-contrast microscope (IX71/IX2, Olympus).</p></sec><sec><title>Morphological analysis and apoptosis detection</title><p>The formalin-fixed tissues were embedded in paraffin and transversely sectioned (5 mm thick). After deparaffinization and dehydration, the sections were stained by hematoxylin and eosin.</p><p>Apoptotic cells were visualized with TUNEL kit (Promega G7310, Madison, WI, USA) following the manufacturer’s protocol. Briefly, after deparaffinization and dehydration, protein digestion was done by incubating tissue sections in 20 mg/ml proteinase K for 15 minutes at room temperature. Sufficient rTdT reaction mix was prepared before for both control and stress groups. One hundred microliters of reaction mix per slide will adequately cover the section. After the reaction of Terminal Deoxynucleotidyl Transferase Recombinant(rTdT), sections were covered with plastic cover slips, incubating at 37°C for 60 minutes inside a humidified chamber; reactions were terminated by immersing the slides in sodium citrate (SSC). Then, sections were incubating in Horseradish Peroxidase (HPR). Finally, the sections were colored by diaminobenzidine (DAB) at room temperature. Microstructures of the small intestine were observed using a BH2 Olympus microscope (DP71, Olympus, Japan).</p></sec><sec><title>RT-PCR</title><p>Expression of HSP70, HSP90 and HSP27were quantitatively determined using real-time PCR. Quantitative PCR analysis was carried out using the DNA Engine Mx3000P® fluorescence detection system against a double- stranded DNA - specific fluorescent dye (Stratagene, USA) according to optimized PCR protocols. β-actin was amplified in parallel with the target genes providing a control.</p><p>The system included 1 μl cDNA, 10 μl mix, 0.3 μl Rox, 6.7 μl Diethypyrocarbonate (DEPC) for every sample, then PCR was carried out as follows: one cycle of denaturation at 94°C for 5 min; denaturation at 94°C for 30 s, annealing at 60°C 30 s; a total of 40 cycles for 1 min at 72°C for 30 s; followed by one cycle of 5 min at 72°C for the final extension. The primes were listed in the Table <xref ref-type="table" rid="T1">1</xref>.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Primer sequences for real-time PCR</p></caption><table frame="hsides" rules="groups" border="1"><colgroup><col align="left"/><col align="left"/><col align="center"/><col align="center"/></colgroup><thead valign="top"><tr><th align="left"><bold>Description</bold></th><th align="left"><bold>Accession number</bold></th><th align="center"><bold>Primer sequence</bold></th><th align="center"><bold>Product (bp)</bold></th></tr></thead><tbody valign="top"><tr><td rowspan="2" align="left" valign="top">β-actin<hr/></td><td rowspan="2" align="left" valign="top">NM_031144<hr/></td><td align="center" valign="bottom">Forward: TTGTCCCTGTATGC CTCTGG<hr/></td><td rowspan="2" align="center" valign="top">218<hr/></td></tr><tr><td align="center" valign="bottom">Reverse: ATGTCACGCACGATTTCCC<hr/></td></tr><tr><td rowspan="2" align="left" valign="top">HSP27<hr/></td><td rowspan="2" align="left" valign="top">NM_031970.3<hr/></td><td align="center" valign="bottom">Forward: GGCAAGCACGAAGAAAGG<hr/></td><td rowspan="2" align="center" valign="top">269<hr/></td></tr><tr><td align="center" valign="bottom">Reverse: GATGGGTAGCAAGCTGAAGG<hr/></td></tr><tr><td rowspan="2" align="left" valign="top">HSP70<hr/></td><td rowspan="2" align="left" valign="top">NM_031971.2<hr/></td><td align="center" valign="bottom">Forward: CGTGCCCGCCTACTTCA<hr/></td><td align="center" valign="bottom">280<hr/></td></tr><tr><td align="center" valign="bottom">Reverse: CACCAGCCGGTTGTCGA<hr/></td><td align="center" valign="bottom"> <hr/></td></tr><tr><td rowspan="2" align="left" valign="top">HSP90<hr/></td><td rowspan="2" align="left" valign="top">NM_175761.2<hr/></td><td align="center" valign="bottom">Forward: GTCCCGGTGCGGTTAGTCACG<hr/></td><td rowspan="2" align="center" valign="top">70<hr/></td></tr><tr><td align="center" valign="bottom">Reverse: TTGGGTCTGGGTTTCCTCAGGC<hr/></td></tr><tr><td rowspan="2" align="left" valign="top">Bcl-2<hr/></td><td rowspan="2" align="left" valign="top">NM_001191.2<hr/></td><td align="center" valign="bottom">Forward: CTGGGAGGAGAAGATGC<hr/></td><td rowspan="2" align="center" valign="top">126<hr/></td></tr><tr><td align="center" valign="bottom">Reverse: ACCTTTGTTCCACGACCCATAG<hr/></td></tr><tr><td rowspan="2" align="left" valign="top">Bax</td><td rowspan="2" align="left" valign="top">NM_017059.2</td><td align="center" valign="bottom">Forward: CAGGACGCATCCACCAAGAA<hr/></td><td rowspan="2" align="center" valign="top">114</td></tr><tr><td align="center">Reverse: GGGTCCCGAAGTAGGAAAGG</td></tr></tbody></table></table-wrap></sec><sec><title>Protein extraction and Western blot</title><p>Tissus and cells were harvested and lysed in lysis buffer (5 μL phosphatase inhibitors, 1 μL protease inhibitor and 5 μL 100 mM PMSF). After incubation (30 min, on ice), lysates were centrifuged (10,000 g, 5 min, 4°C). The supernatant was removed and the protein concentration was measured using a BCA protein assay reagent according to the manufacturer’s instructions.</p><p>Equivalent amounts of protein were subjected to SDS-PAGE electrophoresis and then electroblotted onto nitrocellulose membrane, in which process, the concentration of gel is 12%. The membrane was incubated with primary antibody and then IRDye 800CW-conjugated secondary antibody, and the infrared fluorescence image was obtained using Odyssey infrared imaging system (Li-Cor Bioscience, Lincoln, NE, USA).</p><p>The antibody used in the western blot: rabbit anti-AKT (1:1000; #4685 Cell Signaling Technology Inc.), rabbit anti-phospho-AKT (1:2000; #4060 Cell Signaling Technology Inc.), rabbit anti-Bad (1:1000; #9239 Cell Signaling Technology Inc.), rabbit anti-phospho-Bad (1:1000; #4366 Cell Signaling Technology Inc.), GAPDH(1:1000; Cell Signaling Technology Inc.).</p></sec><sec><title>Statistical analysis</title><p>All data are presented as the mean ± SD. Statistical analysis was performed by independent-sample T-tests using SPSS 11.5 AP-value of < 0.05 was considered significant. Microarray analysis was conducted using a Molecule Annotation System.</p></sec></sec><sec><title>Result and discussion</title><sec><title>Assessment of heat stress</title><p>In research of heat stress, the rectal temperature is generally considered one of indexes of heat stress assessment, after heat exposure, the rectal temperature increased significantly [<xref ref-type="bibr" rid="B18">18</xref>], and our study showed similar result of previous research (Figure <xref ref-type="fig" rid="F1">1</xref>).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Heat stress induced serious affect on physiology.</bold> The weight of rats decreased significantly after heat stress, and rectal temperature significantly increased. Values represent the mean±SD, n=6 rats for each group. p<0.05 control versus heat stress.</p></caption><graphic xlink:href="1746-6148-9-241-1"/></fig><p>The intestinal epithelium provides a physical barrier between the luminal contents and the interior environment of the body and protects the body against entrance of bacteria, bacterial toxins, and other unwanted macromolecules. In rats, about 30% of the total blood flow goes to the small intestine. However, when they are exposed to heat for a long time, this rate reduces significantly to increase the cerebral blood flow for heat dissipation [<xref ref-type="bibr" rid="B19">19</xref>]. The decrease of blood flow to the small intestine in this situation results in ischemia and shedding of intestinal epithelial cells, and it was most serious on the third day [<xref ref-type="bibr" rid="B6">6</xref>]. The damage on intestinal would result in the reduction of food, and then induce decrease of body weight (Figure <xref ref-type="fig" rid="F1">1</xref>). Another index, glucocorticoid, which is critical for successful adaptation [<xref ref-type="bibr" rid="B20">20</xref>], is considered to be a good indicator of stress response intensity, particularly in its acute phase. Thus, the increased glucocorticoid levels observed in heat-treated rats in this study may be indicative of higher stress during heat exposure (Figure <xref ref-type="fig" rid="F2">2</xref>). Similarly, our results revealed severe shedding of epithelial cells in heat-treated rats, suggesting that the small intestines of these rats were damaged by heat stress (Figure <xref ref-type="fig" rid="F3">3</xref>).</p><fig id="F2" position="float"><label>Figure 2</label><caption><p><bold>Changes of glucocorticoid concentrations between control and heat-stressed group.</bold> In the stressed group, the serum corticosterone increased significantly. Values represent the mean ±SD, n=6 rats for each group. p<0.05 heat-stressed versus control.</p></caption><graphic xlink:href="1746-6148-9-241-2"/></fig><fig id="F3" position="float"><label>Figure 3</label><caption><p><bold>Photomicrographs of hematoxylin and eosin-stained sections of control (A) and heat-treated (B) stressed groups.</bold> After treated by hyperthermia, the integrity of small intestine was damaged (jejunum), with desquamation at the top of the intestinal villi and exposure of the lamina propria(indicated by arrows).</p></caption><graphic xlink:href="1746-6148-9-241-3"/></fig><p>Since first reported by Ritossa [<xref ref-type="bibr" rid="B21">21</xref>], knowledge about HSPs has been increasing. They are known to limit the damage caused by stress, and promote cellular recovery [<xref ref-type="bibr" rid="B22">22</xref>]. Consistently, in this study, the expression of HSP 27, HSP 70 and HSP 90 were elevated after heat exposure. These data underscore the role of HSPs in cellular resistance to heat and in heat adaption (Figure <xref ref-type="fig" rid="F4">4</xref>).</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>Expression of rat HSP genes which were detected by RT-PCR.</bold> The expression of HSP genes increased significantly after heat exposure. Values represent the mean ± SD, n = 6 rats for each group. p <0.05 heat-stressed versus control.</p></caption><graphic xlink:href="1746-6148-9-241-4"/></fig></sec><sec><title>Apoptosis induced by heat stress</title><p>During apoptosis, apoptotic cells finally break up into “apoptotic bodies” and are phagocytosed by phagocytes or neighboring cells. TUNEL assay is always considered a usual method to detect apoptotic cells [<xref ref-type="bibr" rid="B23">23</xref>-<xref ref-type="bibr" rid="B25">25</xref>]. To proof the effect of heat stress caused on the cells, it is crucial to detect apoptotic cells, the results that revealed much higher apoptotic rates in heat-treated rats was proofed heat stress caused damage to intestine cells (Figure <xref ref-type="fig" rid="F5">5</xref>). Moreover, apoptotic cells in these rats migrated to the bottom of villi. Given that the intestinal epithelial cells differentiate from cells in the crypts at the bottom of villi, these results may suggest a much more severe apoptosis in heat-treated rats. Gene expression analysis further supported the TUNEL assay results by showing a higher Bax/Bcl-2 ratio in heat-treated rats. Bcl-2 family members are key regulators of apoptosis. They can either repress (eg. Bcl-2) or promote (eg. Bax) apoptosis [<xref ref-type="bibr" rid="B26">26</xref>]. However, heterodimerization between Bax and Bcl-2 may negate the function of either protein. The relative ratio of Bax/Bcl-2 is known to be an important indicator of apoptosis. An excess of Bcl-2 homodimers promotes cell survival [<xref ref-type="bibr" rid="B27">27</xref>], whereas an excess of Bax homodimers promotes apoptosis. Thus, the ratio of Bax/Bcl-2 determines whether a cell will die or survive [<xref ref-type="bibr" rid="B28">28</xref>]. In agreement with a previous report [<xref ref-type="bibr" rid="B29">29</xref>], our results showed that the ratio of Bax/Bcl-2 increased significantly after heat exposure. This ratio was obtained on the third day of heat treatment, because it is the time when the most serious damages were observed in our preliminary experiments. Interestingly, in this study, the significant increase of Bax/Bcl-2 ratio in heat-treated rats was coincident with a considerable amount of villi epithelial cells undergoing apoptosis (Figure <xref ref-type="fig" rid="F6">6</xref>). Therefore, our data are consistent with the observation that a considerably increased Bax/Bcl-2 ratio is associated with the peak period of apoptosis. Taken together, these results suggest that heat stress induced the apoptosis of villi epithelial cells, accompanied by down-regulation of Bcl-2 gene and up-regulation of Bax gene.</p><fig id="F5" position="float"><label>Figure 5</label><caption><p><bold>Apoptosis in jejunum of non-treated and heat-treated rats.</bold> The apoptotic cells of stressed group <bold>(B,D)</bold> was increased after stress compared with control <bold>(A,C)</bold> (indicated by arrows).(<bold>A</bold>,<bold>B</bold> 40×; <bold>C</bold>,<bold>D</bold> 200×).</p></caption><graphic xlink:href="1746-6148-9-241-5"/></fig><fig id="F6" position="float"><label>Figure 6</label><caption><p><bold>Ratio of Bax/Bcl-2 before and after heat stress detected by RT-PCR.</bold> The ratio Bax/Bcl-2 was significantly higher in heat-treated group than in control. Values represent the mean ± SD, n = 6 rats for each group. p <0.05 heat-stressed versus control.</p></caption><graphic xlink:href="1746-6148-9-241-6"/></fig></sec><sec><title>Expression of AKT in vivo and in vitro</title><p>AKT plays key roles in regulating cell growth, survival and metabolism [<xref ref-type="bibr" rid="B30">30</xref>]. It was first discovered as an oncogene within the mouse leukemia virus [<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B32">32</xref>] and as a homolog of protein kinase C [<xref ref-type="bibr" rid="B33">33</xref>]. Thereafter, there have been many exciting breakthroughs elucidating the mechanism of upstream regulation of AKT [<xref ref-type="bibr" rid="B34">34</xref>-<xref ref-type="bibr" rid="B36">36</xref>]. AKT promotes cell survival through the phosphoinositide 3-kinase (PI3K) pathway. After phosphorylation, AKT phosphorylates Bad (serine-136) and inhibits the pro-apoptosis effect [<xref ref-type="bibr" rid="B15">15</xref>], inactive Bad promotes apoptosis by binding to Bcl-xl protein, phosphorylated Bad in turn interacts with 14-3-3 proteins to promote cell survival [<xref ref-type="bibr" rid="B37">37</xref>]. In the present study, we found a key role for the activation of PI3K/AKT in the apoptosis induced by heat stress. This was accompanied by an increase in Bad phosphorylation, which is an inhibitory modification of Bad (Figure <xref ref-type="fig" rid="F7">7</xref>). Together, these results indicate that in response to heat stress, AKT is activated to inhibit apoptosis and promote cell survival. Although AKT may play roles in the whole process of heat treatment, our in vitro study revealed significantly higher phosphorylation of AKT and BAD at 2 h of heat exposure (Figure <xref ref-type="fig" rid="F7">7</xref>). At this time, the numbers of apoptotic cells was also significantly reduced, compared with the number of apoptotic cells after heat stressed 4 hour (Figure <xref ref-type="fig" rid="F7">7</xref>). The previous research of our lab demonstrated that heat-stress induced apoptosis in IEC-6 cells after treated 4 hours, and lead to necrosis treated more time [<xref ref-type="bibr" rid="B38">38</xref>]. The present study suggested that PI3K/AKT pathway protects villi epithelial cells from apoptosis at certain points in the apoptotic process. Collectively, our data support a role of AKT in antagonizing apoptosis of villi epithelial cells through the PI3K/AKT pathway. Heat stress may stimulate the activity of AKT to repress apoptosis.</p><fig id="F7" position="float"><label>Figure 7</label><caption><p><bold>Effect of AKT on heat-induced apoptosis. (A)</bold> Phosphorylation of AKT and BAD in rat small intestine was significantly increased after heat exposure. <bold>(B)</bold> AKT and Bad proteins extracted from rat IEC-6 cells. Cells were treated with heat for 15 min, 30 min, 1 h, 2 h, 4 h and 8 h, respectively. The phosphorylation of AKT and BAD were significantly higher at 2 h of heat exposure than at other time points. <bold>(C)</bold> was observed after heat exposure. Compared to control, heat stress caused damages to cell morphology. The damage was more serious at 4 h <bold>(b</bold>,<bold>d)</bold> of heat exposure than at 2 h<bold>(a,c)</bold>(<bold>a</bold>,<bold>b</bold> 200×; <bold>c</bold>,<bold>d</bold> 400×). p <0.05, * p <0.01 heat-stressed versus control.</p></caption><graphic xlink:href="1746-6148-9-241-7"/></fig></sec></sec><sec sec-type="conclusions"><title>Conclusions</title><p>In conclusion, the results in this present study suggest that heat stress affected growth of rats, caused damages to the small intestine and induced shipping and apoptosis of epithelial cells. This study also demonstrates that PI3K/AKT signal pathway was involved in the resistance mechanisms of apoptosis induced heat stress. In addition, in IEC-6 cell lines, a significant higher level of AKT phosphorylation was observed at 2 h after heat exposure, this indicates the PI3K/AKT pathway has an effect on an early period.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>Z was the student who conducted the study, and wrote the manuscript. P, C, S, H, T and X collected the materials and culture cell lines. F, X and S reviewed the manuscript and the quality of the written English. All authors read and approved the final manuscript.</p></sec><sec><title>Authors’ information</title><p>Zhimin Gao and Fenghua Liu are Joint first authors.</p></sec>
In vitro synthesis of tensioned synoviocyte bioscaffolds for meniscal fibrocartilage tissue engineering	<sec><title>Background</title><p>Meniscal injury is a common cause of lameness in the dog. Tissue engineered bioscaffolds may be a treatment option for meniscal incompetency, and ideally would possess meniscus- like extracellular matrix (ECM) and withstand meniscal tensile hoop strains. Synovium may be a useful cell source for meniscal tissue engineering because of its natural role in meniscal deficiency and its <italic>in vitro</italic> chondrogenic potential. The objective of this study is to compare meniscal -like extracellular matrix content of hyperconfluent synoviocyte cell sheets (“HCS”) and hyperconfluent synoviocyte sheets which have been tensioned over wire hoops (tensioned synoviocyte bioscaffolds, “TSB”) and cultured for 1 month.</p></sec><sec><title>Results</title><p>Long term culture with tension resulted in higher GAG concentration, higher chondrogenic index, higher collagen concentration, and type II collagen immunoreactivity in TSB versus HCS. Both HCS and TSB were immunoreactive for type I collagen, however, HCS had mild, patchy intracellular immunoreactivity while TSB had diffuse moderate immunoreactivity over the entire bisocaffold. The tissue architecture was markedly different between TSB and HCS, with TSB containing collagen organized in bands and sheets. Both HCS and TSB expressed alpha smooth muscle actin and displayed active contractile behavior. Double stranded DNA content was not different between TSB and HCS, while cell viability decreased in TSB.</p></sec><sec><title>Conclusions</title><p>Long term culture of synoviocytes with tension improved meniscal- like extra cellular matrix components, specifically, the total collagen content, including type I and II collagen, and increased GAG content relative to HCS. Future research is warranted to investigate the potential of TSB for meniscal tissue engineering.</p></sec>	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Warnock</surname><given-names>Jennifer J</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>jennifer.warnock@oregonstate.edu</email></contrib><contrib contrib-type="author" id="A2"><name><surname>Baker</surname><given-names>Lindsay</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>lbaker@westvet.net</email></contrib><contrib contrib-type="author" id="A3"><name><surname>Ballard</surname><given-names>George A</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I3">3</xref><email>adamballard549@yahoo.com</email></contrib><contrib contrib-type="author" id="A4"><name><surname>Ott</surname><given-names>Jesse</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>jesse.ott@oregonstate.edu</email></contrib>	BMC Veterinary Research	<sec><title>Background</title><p>Meniscal injuries are a common cause of painful stifle arthritis and joint dysfunction in dogs and humans [<xref ref-type="bibr" rid="B1">1</xref>-<xref ref-type="bibr" rid="B6">6</xref>]. Despite intensive research over several decades, a cure for meniscal deficiency has not been found. Thus tissue engineering is being investigated as a means for inducing meniscal healing or regeneration, through producing fibrocartilage neotissues in the laboratory, utilizing cell culture, scaffold use, and <italic>in vitro</italic> biomechanical stimulation.</p><p>Tissue engineering scaffolds mechanically support cell growth and guide tissue formation, and have been surgically implanted in dogs [<xref ref-type="bibr" rid="B7">7</xref>-<xref ref-type="bibr" rid="B9">9</xref>]. An ideal cell scaffold for meniscal tissue engineering and <italic>in vivo</italic> implantation would have physiologically relevant, meniscus like extracellular matrix (ECM) and biomechanical properties. Synthetic polymer scaffolds such as porous polyurethane [<xref ref-type="bibr" rid="B10">10</xref>] poly vinyl alcohol hydrogel [<xref ref-type="bibr" rid="B11">11</xref>], polyglycolic acid mesh [<xref ref-type="bibr" rid="B12">12</xref>], 75:25 poly(lactic-co-glycolic acid)[<xref ref-type="bibr" rid="B13">13</xref>], l-lactide ϵ-caprolactone co-polymer [<xref ref-type="bibr" rid="B14">14</xref>], poly(L-co-D,L-lactic acid)/poly(caprolactone-triol) [<xref ref-type="bibr" rid="B15">15</xref>], small intestinal submucosa [<xref ref-type="bibr" rid="B8">8</xref>], collagen gels [<xref ref-type="bibr" rid="B16">16</xref>], and pure collagen scaffolds [<xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B18">18</xref>], have been investigated for the treatment of meniscal deficiency. While these implants partially replace meniscal function, scaffolds may be complicated by progressive chemical breakdown, structural weakening, lack of lubricity, inadequate integration with the recipients’ tissues [<xref ref-type="bibr" rid="B19">19</xref>], articular cartilage contact damage [<xref ref-type="bibr" rid="B20">20</xref>] and inflammatory reactions [<xref ref-type="bibr" rid="B18">18</xref>] incited by allogenic or xenogenic scaffold content [<xref ref-type="bibr" rid="B21">21</xref>].</p><p>As harvest of autologous meniscal cells would cause patient injury, this study used canine osteoarthritic, autogenous joint- origin synoviocytes obtained from arthroscopic debris collected during clinically indicated stifle surgery. To avoid additional patient morbidity, synoviocytes from normal, unaffected joints were not utilized; additionally, normal canine synovium has been found to have less chondrogenic potential compared to infrapatellar fat pad [<xref ref-type="bibr" rid="B22">22</xref>]. In rats, mice, and humans, synovium is superior to other cell sources, including periosteum, bone marrow, muscle, and adipose, for in vitro chondrogenesis [<xref ref-type="bibr" rid="B23">23</xref>-<xref ref-type="bibr" rid="B25">25</xref>]. In the dog, osteoarthritic joint-origin synovial cells can produce components of meniscal fibrocartilage <italic>in vitro</italic>[<xref ref-type="bibr" rid="B26">26</xref>,<xref ref-type="bibr" rid="B27">27</xref>] and are readily obtained minimally invasively [<xref ref-type="bibr" rid="B28">28</xref>]. In vivo, canine osteoarthritic joint- origin synoviocytes are responsible for cellular repopulation of meniscal grafts [<xref ref-type="bibr" rid="B29">29</xref>] and undergo spontaneous fibrochondrogenesis [<xref ref-type="bibr" rid="B30">30</xref>], including formation of a meniscal-like regenerate in meniscectomized stifles [<xref ref-type="bibr" rid="B31">31</xref>]. Synovial pedicle grafts and free synovial grafts have been used to achieve partial avascular meniscal healing in dogs and rabbits [<xref ref-type="bibr" rid="B32">32</xref>-<xref ref-type="bibr" rid="B35">35</xref>] and are superior to muscle flaps and synthetic meshes in dogs [<xref ref-type="bibr" rid="B33">33</xref>]. Use of living, autologous bioscaffolds with these regenerative properties could be advantageous over use of synthetic scaffolds for augmenting meniscal healing or treating meniscal loss.</p><p>Recently, scaffold-free tissue-engineered constructs cultured from synovial mesenchymal stem cells were used to successfully induce healing of meniscal defects [<xref ref-type="bibr" rid="B36">36</xref>] in pigs. Thus, given this success in the pig, and to avoid the complications of scaffold use, the objective of this study is to produce autologous fibrocartilage bioscaffolds, which in the future may be utilized in the dog as surgical implants without synthetic, xenogenic, or allogenic components. We hypothesize that the long term application of tension to synoviocytes will result in formation of a collagenous bioscaffold, with greater cell viability, and increased meniscal-like extracellular matrix content and architecture, versus synoviocytes grown short term in hyperconfluent monolayer culture.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Tissue harvest</title><p>With informed owner consent, synovium was obtained from 10 dogs with naturally occurring clinical osteoarthritis as per institutional Animal Care and Use Committee Protocol. Dogs were treated for degeneration of the cranial cruciate ligament and medial meniscal injury via exploratory arthroscopy, partial meniscectomy if indicated, and tibial plateau leveling osteotomy or lateral tibiofabellar suture. Synovial villi were arthroscopically harvested during routine partial synovectomy, only as clinically required, using a tissue shaver (Stryker, San Jose, CA) with a 3.5 mm aggressive shaver blade run at 1800 rpm [<xref ref-type="bibr" rid="B28">28</xref>]. Dogs with a history of steroid administration, or dogs with concurrent disease processes were excluded from the study.</p><p>Harvested synovial villi were immediately placed in a 50 mL polypropylene tube containing 40 mL of Dulbeccos’ Modified Eagle’s Media (DMEM) with 10% fetal bovine serum (FBS), warmed to 37°C. The tube was then transported immediately to the laboratory and centrifuged at 313 g, media was decanted, and resultant tissue pellet weighed, and transferred by pipette to a digestion solution as described below.</p></sec><sec><title>Cell culture</title><p>Tissue fragments were completely digested with sterile Type 1A clostridial collagenase 10 mg/mL in RPMI 1640 solution over 2–6 hours at 37°C. Tissue was deemed to be completely digested when no ECM could be visualized grossly. Cells were cultured in monolayer for 4 passages to isolate Type B fibroblast-like synoviocytes [<xref ref-type="bibr" rid="B37">37</xref>] and Type C intermediate synoviocytes [<xref ref-type="bibr" rid="B37">37</xref>]. The following media formulation was used for the duration of culture: high glucose DMEM with phenol red pH indicator, supplemented with 17.7% FBS, 0.021 mg/mL glycine, 0.025 mg/mL L-alanine, 0.037 mg/mL L- asparagine, 0.038 mg/mL L-aspartic acid, 0.042 mg/mL L-glutamic acid, 0.033 mg/mL L-proline, 0.030 mg/mL L-serine, 0.23 mg/mL pyruvate, 0.52 mg/mL L-glutamine, 6.75 mg/mL HEPES buffer, 177.0 units/mL penicillin, 177.0 μg/mL streptomycin, and 0.44 ug/mL amphoterocin (supplemented DMEM, “sDMEM”). The flasks were incubated at canine body temperature, at 37.8°C, and at 5% CO<sub>2</sub>, 95% humidity, with sterile media change performed every 24 hours.</p><p>For each dog, synoviocytes were passaged when 95% monolayer confluence was reached. For all dogs at 4<sup>th</sup> passage a mean of 6,300,000 ±10,000 cells per flask were seeded into eighteen 150 cm<sup>2</sup> flasks. These 4<sup>th</sup> passage cells were allowed to become hyperconfluent monolayer cell sheets (“HCS”), defined as cells overlapping each other in greater than 100% confluency (Figure <xref ref-type="fig" rid="F1">1</xref>A). When HCS began to spontaneously contract off the corners of the flask floor (Figure <xref ref-type="fig" rid="F1">1</xref>B), they were completely dislodged off the flask floors by gentle pushing with the pipette tip. At this time 3 HCS per dog were harvested for tissue analyses (as described below), 6 were required for use in another unrelated study, and the remaining 9 were used to make tensioned synoviocyte bioscaffolds (TSB).</p><fig id="F1" position="float"><label>Figure 1</label><caption><p><bold>Representative hyperconfluent cell sheets: Phase contrast microscopy of 4</bold><sup><bold>th </bold></sup><bold>passage hyperconfluent cell sheet (“HCS”; A), 10X objective magnification, bar = 100 μm.</bold> Gross appearance of a representative hyperconfluent 4<sup>th</sup> passage cells sheet at commencement of spontaneous sheet contraction (arrows), indicating time for sheet harvest to synthesize tensioned synoviocyte bioscaffolds <bold>(B)</bold>.</p></caption><graphic xlink:href="1746-6148-9-242-1"/></fig><p>To synthesize TSB (Figure <xref ref-type="fig" rid="F2">2</xref>), each HCS was moved from the flask to a 6-well plate without touching the flask nozzle, using a 10 mL pipettor and gentle continuous suction. The HCS was gently discharged into a 10 mL media well, filled with the above described media to prevent cell sheet desiccation. A 2.0 cm diameter 22ga wire hoop was placed over the cell sheet. While grasping the twist on the wire hoop with a tissue forceps, the hyperconfluent cell sheet was pulled around the hoop using Bishop Harmon forceps. The hypercondfluent cell sheet was pulled over the hoop three times with approximately 0.5 N of tension to avoid tearing, making a tensioned bioscaffold. Then the completed TSB was placed on the bottom of the well with the free end down to prevent unraveling. TSB were cultured for an additional 30 days with daily media changes and then harvested for tissue analyses as described below.</p><fig id="F2" position="float"><label>Figure 2</label><caption><p>Illustration of tensioned synoviocyte bioscaffold synthesis: 1) the hyperconfluent cell sheet (“HCS”) is moved to the cell culture well using a pipette tip; 2) the wire hoop is placed over the HCS; 3) while holding the wire twist with a forceps, the HCS is doubled over one end of the wire hoop; 4) the ends of the doubled over HCS are grasped with forceps and tension is applied perpendicular to the wire hoop; 5) the HCS is wrapped over the opposite side of the wire hoop; 6) the triple wrap is finished, creating a tensioned synoviocyte bioscaffold with six cell layers.</p></caption><graphic xlink:href="1746-6148-9-242-2"/></fig></sec><sec><title>Tissue analyses</title><p>Tissue analyses of HCS and TSB examined presence of ECM which is functionally critical in the normal meniscus: type I collagen which accounts for the preponderance of meniscal collagen [<xref ref-type="bibr" rid="B38">38</xref>]; type II collagen, which accounts for a small amount primarily localized to the axial meniscus, [<xref ref-type="bibr" rid="B38">38</xref>] α- smooth muscle actin (ASM), [<xref ref-type="bibr" rid="B39">39</xref>-<xref ref-type="bibr" rid="B41">41</xref>]; and glycosaminoglycans (GAG) [<xref ref-type="bibr" rid="B42">42</xref>-<xref ref-type="bibr" rid="B44">44</xref>], including aggrecan [<xref ref-type="bibr" rid="B45">45</xref>].</p></sec><sec><title>Cell viability</title><p>During monolayer culture, cell viability counts were performed using the trypan blue exclusion assay [<xref ref-type="bibr" rid="B46">46</xref>] at each passage on all dogs. One HCS each from 3 dogs and one TSB each from all dogs were washed three times in sterile phosphate buffered saline and immersed in 4 μM ethidium homodimer and 6 μM acetomethoxy calcein (calcein –AM) solution (Ethidium homodimer and Calcein AM Live/ Dead Viability Assay, Invitrogen, Carlsbad, CA) for 20 minutes at 37.8°C, 5% CO2, 95% humidity. Live and dead cell counts were performed by hand on 6 regions (3 of the periphery and 3 of the center) of each construct, where cells could be clearly visualized in one plane, using a laser microscope (Eclipse Ti-u Laser Microscope, Nikkon, Japan). Due to the complex three –dimensional nature of the neotissues, these cell counts provided an estimate of cell viability.</p></sec><sec><title>Histology</title><p>Formalin-fixed and paraffin embedded sections of one HCS and two TSB per dog were stained with Hematoxylin and Eosin, Masson’s trichrome, and Toluidine Blue.</p><p>For immunohistochemistry, tissues were sectioned at 4–5 μm and sections were collected on charged slides and baked at 60°C for 1 hour. Slides were rehydrated through two washes of xylene, two washes of 100% ethanol, and one wash of 80% ethanol and water. Slides to be stained for collagen were pretreated with a pepsin digestion of 0.1% pepsin in 0.1 M HCl at 37°C for 5 minutes. All slides were rinsed in Tris buffered saline (TBS), placed on an autostainer and washed in TBS followed by 3% H<sub>2</sub>O<sub>2</sub> in TBS for 5 minutes. Then serum-free protein block (Serum Free Protein Block, Dako, Carpinteria CA, #X0909) was applied for 10 minutes, with excess blown off. The primary antibodies were diluted in a proprietary antibody diluent (Antibody Diluent, Dako, Carpinteria, CA) to the following concentrations: Collagen I (rabbit Collagen Type I Antibody (#AB749P), Millipore, Temecula, CA) 1:100, Collagen II (rabbit Collagen Type II Antibody (#AB746P), Millipore, Temecula, CA), 1:100; alpha smooth muscle actin (mouse Alpha Smooth Muscle Actin Antibody (#M0851), Dako, Carpinteria, CA)1:30; and mouse or rabbit universal negative control antibodies (Universal Negative Control, Dako, Carpinteria, CA) were used to test for non-specific immunoreactivity. Antibodies were applied to experimental neotissues or control tissues for 30 minutes at room temperature. Positive control tissues (Envision + HRP, Dako, Carpinteria, CA) type II collagen included canine articular cartilage, meniscus, and tracheal cartilage; for type I collagen, meniscus, tendon, and skin; small intestine for α- smooth muscle actin; and lymph node for macrophages. After washing in TBS, secondary antibodies were applied for 30 minutes at room temperature, then washed again with TBS. The chromogen (Nova Red, Vector Laboratories, Burlingame, CA) was applied for 5 minutes as directed by the manufacturer, and washed in deionized water followed by hematoxylin for 5 minutes, rinsed again in deionized water, then rinsed in TBS, and coverslipped.</p><p>Histologic specimens were examined completely at 10x and 20x objective magnification (Zeiss Microscope, Thornwood, NY) and images of each section were captured by a digital camera (Olympus DP-70 Digital Camera, Olympus, Melville, NY). Immunoreactivity staining intensity for each was rated as originally described by Wakshlag et al. [<xref ref-type="bibr" rid="B47">47</xref>] with a few modifications: immunoreactivity was localized to intracellular or extracellular staining and intensity was described as weak, moderate, and strong staining. Intracellular and extracellular immunoreactivity was described as being rare if <10% of the cells or ECM area was positively stained, patchy if 10-50% of cells or ECM were stained, and diffuse if >50% of cells or ECM was stained.</p></sec><sec><title>Tissue weight</title><p>One HCS and one TSB per dog were lyophilized and a dry weight obtained. Samples were digested in 1.0 ml Papain Solution (2 mM Dithiothreitol and 300ug/ml Papain) at 60°C in a water bath for 24 hours. This papain digest solution was used to obtain double stranded DNA (dsDNA), GAG, and collagen content of the HCS and TSB.</p></sec><sec><title>DNA quantification</title><p>Double stranded DNA quantification assay (The Quant-iT PicoGreen™ Assay, Invitrogen Carlsbad, CA) was performed per manufacturer’s instructions; double stranded DNA extracted from bovine thymus was used to create standards of 1,000, 100, 10, and 1 ng/mL. Standard and sample fluorescence was read by a fluorometer (Qubit, Invitrogen, Carlsbad CA) at 485 nm excitation/ 528 nm emission and dsDNA was determined based on the standard curve.</p></sec><sec><title>Biochemical extracellular matrix assays</title><p>Glycosaminglycan content was determined by the di-methyl-methylene blue sulfated glycosaminoglycan assay [<xref ref-type="bibr" rid="B48">48</xref>] using a spectrophotometer (Synergy HT– KC4 Spectrophotometric Plate Reader and FT4software, BioTec, Winooski, VT). The Chondrogenic Index was calculated using the following equation: [μg GAG/ ug dsDNA] [<xref ref-type="bibr" rid="B49">49</xref>] and [μg collagen/ ug dsDNA] to identify chondrogenic cellular activity of each tested culture type. Collagen content was determined by Erlich’s hydroxyproline assay, as described by Reddy et al. [<xref ref-type="bibr" rid="B50">50</xref>] Hydroxyproline content was converted to collagen content using the equation: [μg hydroxyproline × dilution factor/ 0.13 = μg collagen] (Ignat’eva et al. [<xref ref-type="bibr" rid="B51">51</xref>]), because hydroxyproline is approximately 13% of the amino acids in human meniscal collagen [<xref ref-type="bibr" rid="B52">52</xref>]. Collagen Index was calculated using the following equation: [μg collagen/ ug dsDNA] to determine cellular collagen production. GAG and collagen content were also standardized to tissue dry weight and expressed as% dry weight [<xref ref-type="bibr" rid="B53">53</xref>].</p></sec><sec><title>Statistical methods</title><p>Data was analyzed using a paired 2-tailed Student’s t-test using statistical software (Graphpad Prism, La Jolla, CA). All data is reported as mean ± Standard Error of the Mean (SEM). Statistical significance was declared at <italic>P</italic> ≤ 0.05.</p></sec></sec><sec sec-type="results"><title>Results</title><p>The mean age of dogs was 5.4 years, (range 2–8 years). Breeds represented included: Labrador Retriever (3), Boston Bull Terrier, Australian Shepherd, Rottweiler (2), Doberman Pincer, Labrador cross, and mixed breed; 5 dogs were male neutered, 4 dogs were female spayed, and 1 was an intact female. As observed by a Diplomate of the American College of Veterinary Surgeons –Small Animal, all dogs had synovitis and osteophytosis, and grade 1–2 Outerbridge cartilage lesions [<xref ref-type="bibr" rid="B54">54</xref>].</p><sec><title>Cell culture, cell viability, and cellularity</title><p>Cell culture: The mean wet weight of harvested synovium was 1.59 g, (range: 0.15 g- 4.8 g). As found previously [<xref ref-type="bibr" rid="B28">28</xref>], the preponderance of cells harvested arthroscopically were red blood cells. A mean of 2.27 × 10<sup>6</sup> ± 6.6 × 10<sup>5</sup> synoviocytes per dog were obtained at harvest. Time from plating at 4<sup>th</sup> passage to hyperconfluency and commencement of spontaneous contraction off the plate floor was a median of 5 days (range 2.3 -8 days).</p><p>Timing of TSB synthesis was critical and needed to be performed upon first observation of cell sheet contraction as described above. Failure to immediately harvest HCS resulted in the formation of contracted masses. By the end of the 1st week of culture, TSB were able to be moved without unraveling. Long term culture with tension resulted in a sheet- like appearance of TSB, with some TSB appearing more translucent (Figure <xref ref-type="fig" rid="F3">3</xref>A) while others appeared more opaque (Figure <xref ref-type="fig" rid="F3">3</xref>B). Of the 9 attempted TSB made per dog, an average of 2.3 TSB per dog partially contracted off their wire hoops, forming an incomplete sheet or contracted mass, (Figure <xref ref-type="fig" rid="F3">3</xref>C) and were not analyzed in this study. One dog, a 7 year old male neutered Doberman pincer, produced hyperconfluent cell sheets that were thin and fragile and disintegrated upon manipulation, precluding formation of TSB in this individual. For the first 2 weeks after TSB were synthesized, the culture well media had become light orange by the time the daily media change was due, indicating a drop in pH.</p><fig id="F3" position="float"><label>Figure 3</label><caption><p>Tensioned synoviocyte bioscaffolds: Gross appearance of two representative tensioned synoviocyte bioscaffolds, (“TSB”; A and B); a TSB that contracted itself apart to form an incomplete TSB (C).</p></caption><graphic xlink:href="1746-6148-9-242-3"/></fig><p>Viability: For monolayer culture cell viability was 99% for passages 1–3. Mean cell viability of 4<sup>th</sup> passage monolayer cells was 98.8% ±0.4 versus passages 1–3 (P = 0.78). Mean cell viability of HCS was 93.3% ±1.7 compared to 4<sup>th</sup> passage cells (P = 0.04). TSB viability cell counts represented an estimation due to their 3 –dimensional structure. Cell viability of TSB was 74.9% ±5.6, representing a decrease from 4<sup>th</sup> passage cells and HCS (P = 0.005 and P = 0.01, respectively).</p><p>Double Stranded DNA Content: Percent tissue DNA content standardized to dry weight was 0.18% ± 0.0002 for HCS and 0.20% ± 0.0003 for TSB (P = 0.5494).</p><p>Hematoxalin and Eosin Histologic Analysis: HCS contained fibroblastic type cells with light, homogenous, eosinophilic ECM. TSB had dense ECM organized in bands and sheets. TSB cells appeared round to fusiform with the long axis parallel to the vector of tension (Figure <xref ref-type="fig" rid="F4">4</xref>).</p><fig id="F4" position="float"><label>Figure 4</label><caption><p><bold>Histologic Analysis: Hematoxylin and Eosin stain (“H + E”), Toluidine Blue stain for GAG (“TB”), and Masson’s Trichrome stain for collagen (“MT”) of hyperconfluent cell sheets (“HCS”) and tensioned synoviocyte bisocaffolds (“TSB”).</bold> 10X objective magnification, bar = 100 μm. Note the difference in tissue architecture between the thin HCS and dense bands and sheets of extracellular matrix in TSB.</p></caption><graphic xlink:href="1746-6148-9-242-4"/></fig></sec><sec><title>Glycosaminoglycan content</title><p>Dimethylmethylene Blue (DMMB) Assay: Percent GAG per dry weight was 1% ± 0.0005 for HCS and 1.8% ±0.001 for TSB (P = 0.001). On a cellular level, TSB synoviocytes produced more GAG per cell (per the Chondrogenic Index), at 10.7 ± 1.4 versus 6.2 ±0.7 for HCS (P = 0.0052).</p><p>Toluidine Blue Histologic Analysis: All HCS had light homogenous GAG deposition in between cell layers. TSB contained regional GAG deposition (Figure <xref ref-type="fig" rid="F3">3</xref>). In TSB from 2 dogs, some of the cells within GAG deposits were rounded in shape and were located in pseudo lacunae, similar to cells of the axial meniscus or articular cartilage.</p></sec><sec><title>Collagen content</title><p>Hydroxyproline Assay: TSB had higher collagen content at 13.1% ±0.02 versus the 5.7% ±0.01 of HCS (P = 0.0137). There was a trend that TSB synoviocytes produced more collagen per cell (per the Collagen Index), at 78.1 ±19.4 versus HCS at 34.7 ±7.1 (P = 0.05).</p><p>Trichrome Blue Histologic Analysis: Hyperconfluent monolayer synoviocyte sheets contained light collagen accumulation between cell layers. TSB contained dense collagen bands and sheets with cells oriented parallel to the vector of tension (Figure <xref ref-type="fig" rid="F4">4</xref>).</p><p>Collagen Immunohistochemistry: Mild patchy intracellular immunoreactivity to type I collagen was observed in all HCS. TSB 8 of 9 dogs had moderate to strong type I intracellular immunoreactivity over 50% of the cell population, with diffuse moderate immunoreactivity over the entire bioscaffold (Figure <xref ref-type="fig" rid="F5">5</xref>). One dog, a 7.3 year old female spayed mixed breed had mild intracellular and moderate extracellular immunoreactivity over 50% of the cell population and ECM area.</p><fig id="F5" position="float"><label>Figure 5</label><caption><p><bold>Immunohistochemistry analysis: Immunohistochemistry for collagens type I and II (“Col1 and Col2”), and alpha smooth muscle actin (“ASM”), of hyperconfluent cell sheets (“HCS”) and tensioned synoviocyte bisocaffolds (“TSB”).</bold> Immunohistochemistry negative controls are delineated by “NC.” 10X objective magnification, bar = 100 μm; Nova Red chromogen. In this example, TSB has moderate extracellular matrix staining for type I collagen, moderate intracellular immunoreactivity to type II collagen, and strong intracellular immunoreactivity to alpha smooth muscle actin.</p></caption><graphic xlink:href="1746-6148-9-242-5"/></fig><p>All HCS were negative for type II collagen immunoreactivity. In 7 dogs TSB had moderate to strong intracellular type II collagen immunoreactivity in 10-50% of cells while 1 dog had mild immunoreactivity in 10% of cells (Figures <xref ref-type="fig" rid="F5">5</xref>). When examined at higher magnification these positively staining cells were grouped in clusters and had variable size and shape (Figure <xref ref-type="fig" rid="F6">6</xref>). TSB contained mild immunoreactivity to type II collagen over 10-50% of the bioscaffold ECM in 4 dogs and less than 10% in 3 dogs. One dog, an 8 year old male neutered Labrador, was negative for ECM type II collagen immunoreactivity, and the above listed 7.3 year old female spayed Labrador was also negative for intracellular and ECM type II collagen immunoreactivity.</p><fig id="F6" position="float"><label>Figure 6</label><caption><p><bold>Evidence of mesenchymal progenitor cells: Immunohistochemistry for collagen type II in a tensioned synoviocyte bioscaffold (“TSB Col2”), showing variably shaped, large, fibroblastic cells (denoted by </bold><bold><italic>*</italic></bold><bold>) with moderate intracellular immunoreactivity to type II collagen; these cells are likely synovial mesenchymal progenitor cells.</bold> 20X objective magnification, bar = 50 μm; Nova Red chromagen.</p></caption><graphic xlink:href="1746-6148-9-242-6"/></fig></sec><sec><title>Alpha smooth muscle actin content</title><p>Strong intracellular immunoreactivity to ASM was expressed in all HCS and TSB (Figure <xref ref-type="fig" rid="F5">5</xref>).</p></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>Of the many synthetic and biologic materials available to create scaffolds for meniscal tissue engineering, use of type I collagen has the closest functional significance. Type I collagen is the principal functional ECM component of the menisci, which is organized into circumferential bands to convert weight bearing forces into tensile hoop strains [<xref ref-type="bibr" rid="B52">52</xref>,<xref ref-type="bibr" rid="B55">55</xref>,<xref ref-type="bibr" rid="B56">56</xref>]. Because tension is a principle biomechanical stimulus for type I collagen formation <italic>in vivo</italic>[<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B57">57</xref>,<xref ref-type="bibr" rid="B58">58</xref>] and <italic>in vitro</italic>[<xref ref-type="bibr" rid="B59">59</xref>,<xref ref-type="bibr" rid="B60">60</xref>], tension was utilized in the present study with the goal of producing a bioscaffold that is rich in type I collagen. Synoviocytes are mechanosensitive to tension [<xref ref-type="bibr" rid="B61">61</xref>,<xref ref-type="bibr" rid="B62">62</xref>]; synoviocytes cultured in monolayer at 50-60% confluence and exposed to static tension increase hyaluronic acid production by 57% [<xref ref-type="bibr" rid="B61">61</xref>]. In the present study long term culture with tension increased collagen content relative to HCS Long term culture with tension also produced tissue architecture consisting of bands and sheets of collagen, with longitudinally oriented cells, which is closer to the histologic appearance of the meniscus versus the architecture of HCS. This is the first report that static tension over long term culture can increase collagen content towards a fibrocartilage tissue type in synovial fibroblasts.</p><p>The concentration of collagen in normal rabbit synovium per dry weight is approximately 11% [<xref ref-type="bibr" rid="B63">63</xref>], which is slightly lower than the 13% collagen of TSB, and higher than the 5% of HCS. The lower collagen content of HCS may reflect the diseased origin of our synovial tissues or may be a result of the artificial monolayer culture environment. At this time the ideal collagen content of meniscal implants is not known, but maximizing collagen content would likely improve the strength and durability of a surgical implant.</p><p>GAG and type II collagen are major ECM synthesis targets in meniscal tissue engineering because they are functionally critical components of the meniscus, particularly in the non- healing, axial, avascular zone [<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B45">45</xref>]. In the present study TSB were 1.8% GAG, and as observed histologically, the majority of TSB contained type II collagen. In contrast, native synovium produces collagen types I,III, and VI [<xref ref-type="bibr" rid="B64">64</xref>], and contains only 0.7% of GAG per dry weight [<xref ref-type="bibr" rid="B63">63</xref>]. In this study, regional deposition of GAG and type II collagen were synthetic products of mesenchymal synoviocyte progenitor cells [<xref ref-type="bibr" rid="B22">22</xref>] (Figure <xref ref-type="fig" rid="F6">6</xref>), the numbers of which were increased by the high concentration of FBS and long term culture [<xref ref-type="bibr" rid="B65">65</xref>,<xref ref-type="bibr" rid="B66">66</xref>]. The variation seen in type II collagen formation between dogs may have been due to variable numbers of mesenchymal progenitor cells per dog, and their varying differentiation potential within each individual [<xref ref-type="bibr" rid="B67">67</xref>]. Another potential mechanism for GAG and type II collagen deposition involves biomechanical stimulation. Type II collagen and GAG form when synoviocytes are exposed to compressive loads in vitro [<xref ref-type="bibr" rid="B68">68</xref>], which may have been generated in the TSB during tensioning due to Poisson’s effect. Poisson’s effect states that when a material is tensioned, it contracts transversely to the vector of pull (which can be observed when stretching a rubber band, for example), thereby forming regions of compression within the material. In addition, FBS does contain chondrogenic growth factors, which support joint growth of the fetal calf in utero; however the quantity and proportion of these factors were not determined in this study.</p><p>In meniscal development, the undifferentiated meniscal primordia are highly cellular with minimal ECM [<xref ref-type="bibr" rid="B69">69</xref>], which mature into fibrocartilage with few cells and dense ECM. Because tissue differentiation is highly dependent on cell density [<xref ref-type="bibr" rid="B70">70</xref>], we attempted to recapitulate this developmental process with long term culture of highly cellular HCS tensioned as TSB. While ECM did increase in TSB (Figure <xref ref-type="fig" rid="F3">3</xref>), total dsDNA concentration, as a measure of tissue cellularity, was not different between HCS and TSB. Further, cell viability dropped over time, possibly due to apoptosis resulting from prolonged cell culture. Additionally, culture of highly cellular TSB in 9 mL of media with once daily media changes limited nutrient delivery and resulted in a daily pH drop, contributing to cell mortality.</p><p>Canine synovium expresses alpha smooth muscle actin [<xref ref-type="bibr" rid="B71">71</xref>]. In the present study, the presence of ASM in HCS and TSB explains the contractile behavior of HCS, maintenance of tension of TSB, as well as contraction of some TSB off their wire hoops. This contractile behavior of a bioscaffold implant could be utilized to assist in apposition of meniscal wound edges for augmentation of meniscal healing.</p><p>In prior research, we found that culture of canine synoviocytes as HCS in monolayer over a 30 day period resulted in spontaneously contracted masses with poor cell viability, poor ECM formation, and poor handling characteristics [<xref ref-type="bibr" rid="B28">28</xref>]. However, in the present study, we did not repeat long term monolayer culture as a control to TSB, which is a weakness in our study.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>In conclusion, we partially accept our hypothesis, as addition of tension over longer term culture in TSB resulted in higher GAG and collagen content, versus HCS. DNA content was not different between TSB and HCS. However, cell viability dropped over time in TSB. Thus, TSB is a viable model for future <italic>in vitro</italic> meniscal tissue engineering studies, but further investigation is required to reduce TSB cell mortality, increase collagen content, and reduce inter-patient variability of these constructs.</p></sec><sec><title>Abbreviations</title><p>ASM: Alpha smooth muscle actin; Calcein –AM: Acetomethoxy- calcein; DMEM: Dulbecco’s modified Eagle’s media; hgDMEM: High glucose Dulbecco’s modified Eagle’s media; sDMEM: Supplemented Dulbecco’s modified Eagle’s media; ECM: Extracellular matrix; FBS: Fetal bovine serum; GAG: Glycosaminoglycans; HCS: Hyperconfluent monolayer cell sheet; IHC: Immunohistochemistry; SEM: Standard error of the mean; TBS: Tris buffered saline; TSB: Tensioned synoviocyte bioscaffold.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors’ contributions</title><p>JW conceived of and designed the study, participated in cell culture and data acquisition, and drafted the manuscript. LB and GB were DVM students of the College of Veterinary Medicine at the time of this study, and carried out cell culture, tissue analysis assays, and helped with study design. JO organized and performed assays and coordindated study participants. All authors read and approved the final manuscript.</p></sec>
Interactive effects of nitrogen addition, warming and invasion across organizational levels in an old-field plant community	<p>Response to global change is dependent on the level of biological organization (e.g. the ecologically relevant spatial scale) in which species are embedded. For example, individual responses can affect population-level responses, which, in turn, can affect community-level responses. Although relationships are known to exist among responses to global change across levels of biological organization, formal investigations of these relationships are still uncommon. I conducted an exploratory analysis to identify how nitrogen addition and warming by open top chambers might affect plants across spatial scales by estimating treatment effect size at the leaf level, the plant level and the community level. Moreover, I investigated if the presence of <italic>Pityopsis aspera</italic>, an experimentally introduced plant species, modified the relationship between spatial scale and effect size across treatments. I found that, overall, the spatial scale significantly contributes to differences in effect size, supporting previous work which suggests that mechanisms driving biotic response to global change are scale dependent. Interestingly, the relationship between spatial scale and effect size in both the absence and presence of experimental invasion is very similar for nitrogen addition and warming treatments. The presence of invasion, however, did not affect the relationship between spatial scale and effect size, suggesting that in this system, invasion may not exacerbate or attenuate climate change effects. This exercise highlights the value of moving beyond integration and scaling to the practice of directly testing for scale effects within single experiments.</p>	<contrib contrib-type="author"><name><surname>Gornish</surname><given-names>Elise S.</given-names></name><xref ref-type="aff" rid="af1">1</xref><xref ref-type="aff" rid="af2">2</xref><xref ref-type="corresp" rid="cor1">*</xref></contrib><aff id="af1"><label>1</label><addr-line>Department of Biological Sciences</addr-line>, <institution>Florida State University</institution>, <addr-line>Tallahassee, FL 32304</addr-line>, <country>USA</country></aff><aff id="af2"><label>2</label><addr-line>Present address</addr-line>: <addr-line>Plant Sciences</addr-line>, <institution>University at California</institution>, <addr-line>Davis 95616</addr-line>, <country>USA</country></aff>	AoB Plants	<sec sec-type="intro" id="s1"><title>Introduction</title><p>Organisms respond to global environmental changes in many ways, including modifications in phenology (e.g. <xref rid="PLU061C12" ref-type="bibr">Edwards and Richardson 2004</xref>; <xref rid="PLU061C39" ref-type="bibr">Moller 2008</xref>), decreases in species richness (e.g. <xref rid="PLU061C21" ref-type="bibr">Hansen <italic>et al.</italic> 2001</xref>) and species abundance (e.g. <xref rid="PLU061C17" ref-type="bibr">Gilbert <italic>et al.</italic> 2008</xref>), and rapid evolution (<xref rid="PLU061C43" ref-type="bibr">Parmesan 2006</xref>). Underlying these broad, population and community-level responses are individual demographic traits, which also respond to environmental changes in complex ways (<xref rid="PLU061C30" ref-type="bibr">Jongejans <italic>et al.</italic> 2010</xref>; <xref rid="PLU061C27" ref-type="bibr">Hoving <italic>et al.</italic> 2013</xref>). For example, using a meta-analysis, <xref rid="PLU061C5" ref-type="bibr">Chalcraft <italic>et al.</italic> (2008)</xref> showed that larger-scale, across-site responses to nitrogen enrichment were contingent on the smaller scale primary productivity within sites. Top-down effects have also been documented (e.g. <xref rid="PLU061C34" ref-type="bibr">Ludwig <italic>et al.</italic> 2000</xref>), as have complex multidirectional effects across spatial scales (e.g. <xref rid="PLU061C4" ref-type="bibr">Browning <italic>et al.</italic> 2012</xref>). Collectively, these studies suggest that research attempting to identify the more comprehensive implications of climate change requires experiments that can explicitly capture effects across spatial scales which are organized by ecologically relevant biological hierarchies (i.e. from individual plant organs, such as a single leaf, to vegetation canopies) (<xref rid="PLU061C42" ref-type="bibr">Ozinga <italic>et al.</italic> 2013</xref>).</p><p>A relatively recent review found evidence for a dampening effect at increasing spatiotemporal scales in studies of biotic response to global change (<xref rid="PLU061C32" ref-type="bibr">Leuzinger <italic>et al.</italic> 2011</xref>). Specifically, they found that effect size (% deviation from control treatments) shows a negative relationship with the (i) number of treatment factors used in an experiment, (ii) temporal extent of an experiment and (iii) spatial extent of an experiment (Fig. <xref ref-type="fig" rid="PLU061F1">1</xref>A). Effect size is expected to decrease as experimental duration increases, partly due to the widely documented phenomenon of acclimation by the experimental species to the particular treatment simulating global change (e.g. <xref rid="PLU061C45" ref-type="bibr">Pedrol <italic>et al.</italic> 2000</xref>; <xref rid="PLU061C35" ref-type="bibr">Maherali <italic>et al.</italic> 2002</xref>; <xref rid="PLU061C50" ref-type="bibr">Rogers and Ellsworth 2002</xref>; <xref rid="PLU061C61" ref-type="bibr">Wu <italic>et al.</italic> 2012</xref>). Alternatively, an increase in treatment complexity and spatial extent of an experiment can increase the number of factors modifying a response to simulated or natural global change. These additional factors render cause–effect relationships less immediate. This may largely be due to attenuation of effect sizes through antagonistic responses (i.e. <xref rid="PLU061C33" ref-type="bibr">Levin 1993</xref>; <xref rid="PLU061C10" ref-type="bibr">Dieleman <italic>et al.</italic> 2012</xref>). For example, although several factors might be involved in driving a response of a leaf to an experimental treatment (e.g. herbivore presence, light availability, etc.), the effect size of a leaf-level response such as leaf N content is modified primarily by chemical processes occurring inside of a single leaf or stem (e.g. <xref rid="PLU061C49" ref-type="bibr">Reid <italic>et al.</italic> 1998</xref>). As higher spatiotemporal levels are considered, the number of factors that play a role in modifying the effect size of a response must increase. This is because each level of organization will include at least the factors driving the response at lower levels (e.g. <xref rid="PLU061C7" ref-type="bibr">Chesson <italic>et al.</italic> 2005</xref>), in addition to those factors only present at higher levels. For example, the factors that modify effect size of a tree-level response include leaf-level phytochemicals and herbivores, as well as soil properties, plant–plant and plant–atmosphere interactions (e.g. <xref rid="PLU061C52" ref-type="bibr">Saxe <italic>et al.</italic> 1998</xref>). In contrast, factors that modify effect size of a leaf-level response only include those relevant at the leaf level, namely the first two (photochemicals and herbivores). Since an increase in the number and diversity of factors in a system is generally considered to increase ecological complexity (e.g. <xref rid="PLU061C44" ref-type="bibr">Parrot 2010</xref>), this could lead to a dilution of effect size with increasing spatial perspective, as described above. <fig id="PLU061F1" position="float"><label>Figure 1.</label><caption><p>Expectations for the effect of global change treatments (A) and the interaction of global change treatments and invasion (B) on effect size of responses across spatial extents.</p></caption><graphic xlink:href="plu06101"/></fig></p><p>Here, I describe a single experiment in which the effects of two factors associated with global change (nitrate addition and elevated temperature) are assessed at different levels of spatial organization: at the leaf level, at the plant level and at the community level. This experiment allowed me to explore the general relationship between spatial scale and vegetation response to global change treatments. Moreover, an additional treatment, simulated invasion through the introduction of a previously absent plant species into experimental plots, allowed me to assess if increasing ecological complexity serves to extend the distance (as defined above) between treatment and response, thereby dampening the effect size of a global change treatment.</p><p>I expected to find a negative relationship between effect size and the spatial scale at which the treatment response was assessed (<xref rid="PLU061C32" ref-type="bibr">Leuzinger <italic>et al.</italic> 2011</xref>; Fig. <xref ref-type="fig" rid="PLU061F1">1</xref>A). The interaction of experimental invasion with warming and elevated nitrogen, however, was expected to have a less straightforward effect. First, because invasion can cause direct and indirect effects (thereby increasing ecological complexity) across all levels of biological organization (<xref rid="PLU061C59" ref-type="bibr">White <italic>et al.</italic> 2006</xref>), the presence of the invasion treatment was expected to reduce effect size across all spatial extents. Second, I expected the slope of the relationship between spatial extent and effect size to become less steep in the presence of invasion. A meta-analysis by <xref rid="PLU061C57" ref-type="bibr">Vila <italic>et al.</italic> (2011)</xref> suggests that the effects of invaders are larger at lower levels of ecological organization compared with those at higher levels of ecological organization. A larger absolute effect of invasion at these lower levels suggests a bigger disparity in effect size at the leaf and plant level compared with the community and ecosystem level (Fig. <xref ref-type="fig" rid="PLU061F1">1</xref>B).</p></sec><sec sec-type="methods" id="s2"><title>Methods</title><sec id="s2a"><title>Experiment</title><p>This study was conducted between 2011 and 2012 in a 1.6-hectare-old field at Tall Timbers Research Station (30°39′06.37″N, 84°14′58.30″W), just south of the Florida–Georgia border (last used for agriculture ca. 150 years ago). There is a diverse native plant community in the field, dominated by grasses and legumes, and it is surrounded on all sides by a mixed loblolly shortleaf pine forest. The field was disked annually, and the soil type is a slightly acidic sandy loam (pH ranges from 5.2 to 6.0). Precipitation at the site averages 100 cm per year, and the average annual air temperature is 20 °C.</p><p>The experiment was nested within a larger design and is a randomized complete block split-plot design with three main factors: nitrogen addition, warming and experimental invasion, for a total of eight treatment combinations. To minimize leaching of nitrogen between sub-plots, the plots were arranged in a split-plot design, with nitrogen treatments applied to blocks comprising eight plots. Each block of treatments was replicated five times, for a total of 40 plots. Each plot was 4 m<sup>2</sup>, but measurements were only collected from the center 1 m<sup>2</sup> as a precaution against edge effects. Plots were separated by 1 m, and rows between plots were mowed annually.</p><sec id="s2a1"><title>Nitrogen</title><p>Six applications of equal amounts of sodium nitrate (NaNO<sub>3</sub>) were applied during the growing season (April–September) in 2011 and 2012, 5 cm below the soil surface of treatment plots to give a total amount of 4 N g m<sup>−2</sup> per year. This amount was based on projected dry + wet nitrogen deposition rates for northern Florida (<xref rid="PLU061C26" ref-type="bibr">Holland <italic>et al.</italic> 2005</xref>), and exists on the more extreme edge of expected increases in deposition (<xref rid="PLU061C40" ref-type="bibr">NADP 2010</xref>). Each application was followed by the application of 800 mL of water to flush the nitrogen below the soil surface. The nitrogen treatment significantly increased foliar nitrogen of experimental plants (see <xref rid="PLU061C19" ref-type="bibr">Gornish 2014</xref>). Plots not receiving nitrogen received comparable amounts of water.</p></sec><sec id="s2a2"><title>Warming</title><p>Warming was applied to experimental plots by erecting open-top hexagonal chambers constructed of a wooden frame (2.54 × 5 cm boards of pressure treated YellaWood<sup>®</sup>) wrapped with 4 mm clear polyethylene plastic sheeting (<xref rid="PLU061C37" ref-type="bibr">Marion <italic>et al.</italic> 1997</xref>) in August 2011. The base of the chamber was 2.4 × 2 m and the top of the chamber was 1.7 × 0.8 m. Each panel was 0.6 m tall. Due to uneven microtopography, the chambers sat ∼3 cm off the ground, allowing for air circulation beneath the base of the greenhouses (<xref rid="PLU061C22" ref-type="bibr">Havstrom <italic>et al.</italic> 1993</xref>) and the unimpeded movement of ground dwelling insects into and out of the warmed plots. The chambers increased the average ambient temperature by 2.5 °C (<xref rid="PLU061C19" ref-type="bibr">Gornish 2014</xref>), and on average, chambers increased night temperatures 25 % more than they increased day temperatures. The chambers were left in the field for the full year of the experiment.</p></sec><sec id="s2a3"><title>Invasion treatment</title><p>Invasion was simulated by experimentally introducing adult (>1-year old) individuals of the perennial composite <italic>Pityopsis aspera</italic> Shuttlw. Ex Small (Asteraceae) into experimental plots in August 2011. The goldenaster, commonly known as pineland silkgrass, is an herbaceous dicot common in xeric sandhill habitats (<xref rid="PLU061C38" ref-type="bibr">Myers and Ewel 1990</xref>) in northern Florida and south Georgia. It is self-incompatible (<xref rid="PLU061C3" ref-type="bibr">Bowers 1972</xref>), reproducing both vegetatively and sexually. <italic>Pityopsis aspera</italic> was used as an experimental invader because it typically occurs in the understorey of surrounding forests and, therefore, could be reasonably expected to colonize the old field through the range filling as a response to a changing climate. The experimental old field is within the range of <italic>P. aspera</italic>, which occurs in north Florida, but is devoid of <italic>P. aspera</italic> individuals. <italic>Pityopsis aspera</italic> individuals were planted at a density of 20 per plot (10 individuals in the center 1 m<sup>2</sup> of the plot and 10 in the periphery of the 4 m<sup>2</sup> plot). Twenty holes were excavated and refilled in all plots that did not receive transplants, to simulate disturbance due to transplanting.</p></sec></sec><sec id="s2b"><title>Measurements</title><p>Responses to the experimental treatments were assigned to the spatial level at which they are mostly relevant. All leaf- and plant-level measurements were taken from <italic>Ambrosia artemisiifolia</italic> L. (annual ragweed), an abundant native annual composite that was naturally found in all of the experimental plots. This cosmopolitan species emerges in late spring, can grow to a substantial height (∼1 m) and produces copious windborne pollen, contributing to its weedy status outside of the USA (<xref rid="PLU061C18" ref-type="bibr">Gladieux <italic>et al.</italic> 2011</xref>). This species has been shown to respond favorably to nitrogen addition (e.g. <xref rid="PLU061C31" ref-type="bibr">Leskovsek <italic>et al.</italic> 2012</xref>) and warming (e.g. <xref rid="PLU061C13" ref-type="bibr">Essl <italic>et al.</italic> 2009</xref>).</p><p>Response variables were organized from small to large based on three predictions. First, I used common hierarchical organizational approaches where larger scale factors are composed of a collection of smaller scale factors (e.g. <xref rid="PLU061C1" ref-type="bibr">Baldocchi 1993</xref>; <xref rid="PLU061C8" ref-type="bibr">Dent <italic>et al.</italic> 2001</xref>). Second, I assumed that larger scale factors would be involved in more intraspecific and interspecific interactions (<xref rid="PLU061C6" ref-type="bibr">Chesson 1998</xref>). Third, I assumed that changes in larger scale factors would take more time than changes in smaller scale factors (e.g. <xref rid="PLU061C60" ref-type="bibr">Woodmansee 1988</xref>).</p><p>At the leaf level, I measured relative water content (RWC) and leaf toughness were measured. Foliar RWC can be related to both water availability and stomatal function (<xref rid="PLU061C36" ref-type="bibr">Mann <italic>et al.</italic> 2011</xref>), both of which can be modified directly and indirectly by factors associated with climate change. Relative water content was measured using rapid estimate procedures modified from <xref rid="PLU061C54" ref-type="bibr">Smart and Bingham (1974)</xref>. In June 2012, three leaves were collected at random from two randomly chosen <italic>A. artemisiifolia</italic> individuals in each plot. The leaves were wrapped in plastic wrap and placed in a dark container until weighing. Samples were first weighed to determine fresh weight (FW), and were then reweighed to determine turgid weight (TW) after being immersed in deionized water for 3 h in a dark fridge. Finally, the samples were blotted to dryness and placed in an oven at 85 °C for 24 h and then reweighed for dry weight (DW):</p><p><disp-formula><mml:math id="DmEquation1"><mml:mrow><mml:mrow><mml:mi mathvariant="normal">RWC</mml:mi></mml:mrow></mml:mrow><mml:mo>=</mml:mo><mml:mrow><mml:mfrac><mml:mrow><mml:mrow><mml:mrow><mml:mi mathvariant="normal">FW</mml:mi></mml:mrow></mml:mrow><mml:mo>−</mml:mo><mml:mrow><mml:mrow><mml:mi mathvariant="normal">DW</mml:mi></mml:mrow></mml:mrow></mml:mrow><mml:mrow><mml:mrow><mml:mrow><mml:mi mathvariant="normal">TW</mml:mi></mml:mrow></mml:mrow><mml:mo>−</mml:mo><mml:mrow><mml:mrow><mml:mi mathvariant="normal">DW</mml:mi></mml:mrow></mml:mrow></mml:mrow></mml:mfrac></mml:mrow></mml:math></disp-formula> </p><p>Relative water content values for leaves in each plot were averaged for a single plot RWC value.</p><p>I also measured leaf toughness in June 2012 to assess treatment effects at the lowest spatial scale. Leaf toughness can be related to plant defence against biotic and abiotic stresses (<xref rid="PLU061C48" ref-type="bibr">Read and Stokes 2006</xref>; <xref rid="PLU061C11" ref-type="bibr">Dominy <italic>et al.</italic> 2008</xref>) and can play a role in driving plant tolerance to changing environmental factors (<xref rid="PLU061C46" ref-type="bibr">Poorter 2009</xref>). In June 2012, I collected the top two leaves from two randomly chosen <italic>A. artemisiifolia</italic> individuals in each plot. Leaf toughness was calculated by measuring the weight of sand necessary to puncture a hole through the center of a single leaf (<xref rid="PLU061C15" ref-type="bibr">Feeny 1970</xref>). Leaf toughness values were averaged among the four leaves collected per plot.</p><p>For plant-level response, I measured plant height, which is strongly correlated with the above-ground plant biomass and other important traits (<xref rid="PLU061C14" ref-type="bibr">Falster and Westoby 2003</xref>), and is an important component of response to environmental variation. In June 2012, the height of the three largest (generally not yet flowering) individuals of <italic>A. artemisiifolia</italic> was measured to the nearest centimetre in each plot. For community-level response, I measured species diversity and functional diversity of the plant community visually in each plot in August 2012, when most species are at peak biomass. Species diversity was quantified by visually counting the unique number of plant species in each plot. Functional groups were chosen to match the types of plant groups that drive succession in abandoned fields. For example, old fields are generally dominated by graminoids, legumes and annual herbs. As succession progresses, perennial herbs, vines and woody species tend to be dominant (<xref rid="PLU061C25" ref-type="bibr">Hermy and Verheyen 2007</xref>). Plants were therefore divided into functional groups based on a combination of lifespan, nitrogen-fixing capability, amount of woody materials and growth form. Functional groups included in this analysis were perennial and annual herbs, legumes, graminoids, woody plants and vines.</p><p>Due to the breadth of response variables included in this study, variation in measurement precision was likely not similar across the data set. Measurement error was expected to be higher in leaf and plant variables compared with numerical community variables, and these errors could have propagated into effect size estimation (<xref rid="PLU061C16" ref-type="bibr">Garrod <italic>et al.</italic> 2013</xref>; see below). Despite these limitations, the data presented are still useful for exploring concepts related to the role of spatial scale in modifying response to factors associated with global change.</p></sec><sec id="s2c"><title>Analysis</title><p>Using MetaWin (<xref rid="PLU061C51" ref-type="bibr">Rosenberg <italic>et al.</italic> 1999</xref>), I used the log response ratio (ln <italic>R</italic>) as my estimate of effect size for all measured responses, calculated as</p><p><disp-formula><mml:math id="DmEquation2"><mml:mrow><mml:mrow><mml:mi mathvariant="normal">ln</mml:mi></mml:mrow></mml:mrow><mml:mspace width="thinmathspace"/><mml:mi>R</mml:mi><mml:mo>=</mml:mo><mml:mrow><mml:mrow><mml:mi mathvariant="normal">ln</mml:mi></mml:mrow></mml:mrow><mml:mfenced open="(" close=")"><mml:mrow><mml:mrow><mml:mfrac><mml:mrow><mml:msup><mml:mi>X</mml:mi><mml:mrow><mml:mrow><mml:mi mathvariant="normal">E</mml:mi></mml:mrow></mml:mrow></mml:msup></mml:mrow><mml:mrow><mml:msup><mml:mi>X</mml:mi><mml:mrow><mml:mrow><mml:mi mathvariant="normal">C</mml:mi></mml:mrow></mml:mrow></mml:msup></mml:mrow></mml:mfrac></mml:mrow></mml:mrow></mml:mfenced></mml:math></disp-formula> where <italic>X</italic><sup>E</sup> and <italic>X</italic><sup>C</sup> are means of the experimental and control groups, respectively. I used the log response ratio as this metric can reduce the effect of plant size across scales on our response variables (<xref rid="PLU061C23" ref-type="bibr">Hedged <italic>et al.</italic> 1999</xref>).</p><p>I was interested in exploring if spatial scale and the presence of invasion contributed to differences in response variables; however, due to small sample sizes (replicates were ‘taken up’ by calculating effect sizes), this analysis on the effect sizes themselves was descriptive. Additionally, I used analysis of variance (ANOVA) to identify the main and interactive effects of invasion (absence and presence) and spatial scale (leaf, plant and community) on response variables (<italic>n</italic> = 5 for each response) overall, as well as for each of the main treatments (nitrogen addition and warming). Patterns detected in this analysis could suggest dynamics describing the relationship between effect size and the spatial scale of observation and perhaps encourage further investigations. All analyses were conducted in R (version 2.15.1, <xref rid="PLU061C47" ref-type="bibr">R Development Core Team 2012</xref>).</p></sec></sec><sec sec-type="results" id="s3"><title>Results</title><p>Relative water content was mostly unaffected by the treatments (Table <xref ref-type="table" rid="PLU061TB1">1</xref>), although the nitrogen + warming interaction reduced RWC in the leaves of <italic>Ambrosia artemisiifolia</italic> relative to the control plots. Warming and nitrogen as main effects increased leaf toughness in the absence of invasion, but the pattern was reversed in the presence of invasion (Table <xref ref-type="table" rid="PLU061TB1">1</xref>). Height of <italic>A. artemisiifolia</italic> was maintained or reduced in response to all treatment main effects, but was slightly increased in the presence of the nitrogen + warming + invasion treatment. At the community level, both functional and species richness were relatively low across all plots and, unexpectedly, were generally unaffected by all experimental treatments (Table <xref ref-type="table" rid="PLU061TB1">1</xref>). Analysis of variance results suggest that, overall, the effect of global change treatments changed with spatial scale (<italic>F</italic><sub>2, 24</sub> = 7.67, <italic>P</italic> = 0.003). The interaction between experimental invasion and spatial scale also contributed to differences in effect size overall (<italic>F</italic><sub>2, 24</sub> = 4.71, <italic>P</italic> = 0.02). <table-wrap id="PLU061TB1" position="float"><label>Table 1.</label><caption><p>Means and (SD) of all responses, across treatments.</p></caption><table frame="hsides" rules="groups"><colgroup span="1"><col align="left" span="1"/><col align="char" char="." span="1"/><col align="char" char="." span="1"/><col align="char" char="." span="1"/><col align="char" char="." span="1"/><col align="char" char="." span="1"/></colgroup><thead><tr><th rowspan="1" colspan="1"/><th align="left" rowspan="1" colspan="1">RWC (%)</th><th align="left" rowspan="1" colspan="1">Leaf toughness (g)</th><th align="left" rowspan="1" colspan="1">Height (cm)</th><th align="left" rowspan="1" colspan="1">Functional richness (#)</th><th align="left" rowspan="1" colspan="1">Species richness (#)</th></tr></thead><tbody><tr><td colspan="6" rowspan="1">Control</td></tr><tr><td rowspan="1" colspan="1"> Invasion absent</td><td rowspan="1" colspan="1">47.0 (8.5)</td><td rowspan="1" colspan="1">80.6 (9.8)</td><td rowspan="1" colspan="1">99.3 (11.5)</td><td rowspan="1" colspan="1">4.8 (1.0)</td><td rowspan="1" colspan="1">9.4 (2.9)</td></tr><tr><td rowspan="1" colspan="1"> Invasion present</td><td rowspan="1" colspan="1">47.0 (7.7)</td><td rowspan="1" colspan="1">79.9 (10.3)</td><td rowspan="1" colspan="1">103.5 (14.9)</td><td rowspan="1" colspan="1">4.8 (0.7)</td><td rowspan="1" colspan="1">9.1 (2.9)</td></tr><tr><td colspan="6" rowspan="1">Nitrogen</td></tr><tr><td rowspan="1" colspan="1"> Invasion absent</td><td rowspan="1" colspan="1">50.0 (4.9)</td><td rowspan="1" colspan="1">90.6 (30.3)</td><td rowspan="1" colspan="1">99.8 (15.5)</td><td rowspan="1" colspan="1">5.1 (0.9)</td><td rowspan="1" colspan="1">10.8 (4.1)</td></tr><tr><td rowspan="1" colspan="1"> Invasion present</td><td rowspan="1" colspan="1">47.0 (4.9)</td><td rowspan="1" colspan="1">66.8 (10.6)</td><td rowspan="1" colspan="1">96.5 (18.6)</td><td rowspan="1" colspan="1">5.5 (0.5)</td><td rowspan="1" colspan="1">11.1 (2.8)</td></tr><tr><td colspan="6" rowspan="1">Warming</td></tr><tr><td rowspan="1" colspan="1"> Invasion absent</td><td rowspan="1" colspan="1">46.0 (9.1)</td><td rowspan="1" colspan="1">86.6 (12.6)</td><td rowspan="1" colspan="1">97.6 (10.1)</td><td rowspan="1" colspan="1">5.4 (0.5)</td><td rowspan="1" colspan="1">10.3 (2.3)</td></tr><tr><td rowspan="1" colspan="1"> Invasion present</td><td rowspan="1" colspan="1">46.0 (6.9)</td><td rowspan="1" colspan="1">68.5 (19.7)</td><td rowspan="1" colspan="1">102.8 (12.6)</td><td rowspan="1" colspan="1">5.0 (0.8)</td><td rowspan="1" colspan="1">12.2 (2.1)</td></tr><tr><td colspan="6" rowspan="1">Nitrogen + warming</td></tr><tr><td rowspan="1" colspan="1"> Invasion absent</td><td rowspan="1" colspan="1">42.0 (7.6)</td><td rowspan="1" colspan="1">76.8 (24.9)</td><td rowspan="1" colspan="1">102.4 (13.2)</td><td rowspan="1" colspan="1">5.3 (0.7)</td><td rowspan="1" colspan="1">10.5 (2.8)</td></tr><tr><td rowspan="1" colspan="1"> Invasion present</td><td rowspan="1" colspan="1">44.0 (6.8)</td><td rowspan="1" colspan="1">73.2 (6.6)</td><td rowspan="1" colspan="1">99.1 (15.2)</td><td rowspan="1" colspan="1">5.1 (0.7)</td><td rowspan="1" colspan="1">10.3 (2.4)</td></tr></tbody></table></table-wrap></p><sec id="s3a"><title>Nitrogen</title><p>Variance associated with effect size was larger in the absence of invasion (Fig. <xref ref-type="fig" rid="PLU061F2">2</xref>A). Spatial scale contributed to differences in effect size in the presence of nitrogen (<italic>F</italic><sub>2,12</sub> = 6.02, <italic>P</italic> = 0.01). In the presence of invasion, there appeared to be a positive relationship between spatial scale and effect size of nitrogen addition. However, there was no main effect of invasion on effect size (<italic>F</italic><sub>1,12</sub> = 3.48, <italic>P</italic> = 0.08), and no interactive effect of spatial scale and invasion (<italic>F</italic><sub>2,12</sub> = 2.52, <italic>P</italic> = 0.11). <fig id="PLU061F2" position="float"><label>Figure 2.</label><caption><p>Effect sizes and effect size variance for global change treatments in the absence (empty points, solid line) and presence (filled points, dotted line) of the invasion treatment: (A) nitrogen only, (B) warming only and (C) nitrogen and warming. Loess splines are included to highlight relationships. The order of response variables across the <italic>x</italic>-axis is: RWC and leaf toughness (leaf level); height (plant level); species richness and functional richness (community level).</p></caption><graphic xlink:href="plu06102"/></fig></p></sec><sec id="s3b"><title>Warming</title><p>Patterns of effect size across spatial scales in the presence and absence of invasion and warming were almost identical to those identified for the nitrogen treatment (Fig. <xref ref-type="fig" rid="PLU061F2">2</xref>B). Analysis of variance results for plots exposed to warming showed that there was also no significant main effect of spatial scale (<italic>F</italic><sub>2,12</sub> = 3.05, <italic>P</italic> = 0.07) or invasion (<italic>F</italic><sub>1,12</sub> = 1.65, <italic>P</italic> = 0.21) on effect size. There was also no significant interaction between the two factors (<italic>F</italic><sub>2,12</sub> = 2.05, <italic>P</italic> = 0.16).</p></sec><sec id="s3c"><title>Nitrogen + warming</title><p>The nitrogen + warming effect sizes displayed the most similar effect sizes across spatial scales. The shallow, positive relationships between effect size and spatial scale in the absence and presence of invasion were not significant (Fig. <xref ref-type="fig" rid="PLU061F2">2</xref>C).</p></sec></sec><sec sec-type="discussion" id="s4"><title>Discussion</title><p>Although scale effects are common (<xref rid="PLU061C62" ref-type="bibr">Wiens 1989</xref>) and play an important role in driving ecological dynamics (e.g. <xref rid="PLU061C58" ref-type="bibr">Villellas <italic>et al.</italic> 2013</xref>), experiments that attempt to directly assess the relationship between the ecological response to changing environmental factors and spatial scale are uncommon. Understanding the role of spatial scale in driving ecological dynamics is necessary for developing a conceptual framework in which to consider biological response to a changing environment (e.g. <xref rid="PLU061C28" ref-type="bibr">Ibanez <italic>et al.</italic> 2014</xref>). Although I am aware that the interpretation of the data depends on how the treatment responses are defined on the spatial scale, my experimental approach facilitated an exploration of how the spatial scale of response can contribute to different effect sizes of nitrogen addition and warming. Also, over time, response patterns may change, but including the temporal component was beyond the scope of this study. In the following, I concentrate on the effect of spatial scales on plant responses. Further, I look at the role of invasion in modifying scale effects and how effect sizes are impacted by single versus combined treatment effects.</p><p>In the presence of invasion overall, I found a trend of increasing effect size with increasing spatial scale. Although these results correspond with observations recorded in other studies (<xref rid="PLU061C55" ref-type="bibr">Strengbom <italic>et al.</italic> 2006</xref>; <xref rid="PLU061C5" ref-type="bibr">Chalcraft <italic>et al.</italic> 2008</xref>; <xref rid="PLU061C41" ref-type="bibr">Oba <italic>et al.</italic> 2008</xref>), they do not support initial hypotheses (Fig. <xref ref-type="fig" rid="PLU061F1">1</xref>). A possible explanation is that my original hypotheses were partly predicated on the assumption that response rates at small scales are faster than those occurring at larger scales (<xref rid="PLU061C24" ref-type="bibr">Heffernan <italic>et al.</italic> 2014</xref>). A larger effect might then be expected at smaller spatial scales for short-term experiments (like the one described in this paper). However, it is possible that a single year of exposure to experimental treatment was not adequate time for responses at all spatial scales to occur. Moreover, if smaller scale responses occurred immediately after treatment application, then acclimation could have occurred at these smaller scales by the time data collection occurred, dampening the presumed effect of treatments.</p><p>A seeming absence of a contribution from spatial scale or the presence of an invader on responses from plants exposed to the nitrogen + warming treatment was also surprising. The interaction between temperature and nitrogen deposition has been shown to significantly affect plants and plant communities (e.g. <xref rid="PLU061C29" ref-type="bibr">Jones and Power 2011</xref>). Increasing the number of treatments simulates increasing environmental heterogeneity, subsequently affecting resilience across a system through portfolio effects (<xref rid="PLU061C53" ref-type="bibr">Schindler <italic>et al.</italic> 2010</xref>). It is possible that increased resilience reduced the magnitude of response across spatial scales, diluting the effect size–spatial scale relationships. However, the trend of lower effect sizes in the combined treatment plots versus the single treatment plots could be confirmed by this study: generally, effect sizes were larger under warming and nitrogen alone than under its combination.</p><p>Interestingly, I found that invasion played a role in modifying the relationship between spatial scale and effect size overall. I expected that as the number of relevant processes contributing to an ultimate response across spatial scales increases, the ecological ‘distance’ between cause and effect would expand, subsequently modifying the relationship between effect size and spatial scale. My observation could be due to emergent effects (<xref rid="PLU061C9" ref-type="bibr">Didham <italic>et al.</italic> 2007</xref>), which are often responsible for invaders having a larger effect on native plants in the presence of resource addition (e.g. <xref rid="PLU061C20" ref-type="bibr">Green and Galatowitsch 2001</xref>).</p></sec><sec sec-type="conclusions" id="s5"><title>Conclusions</title><p>Studies that explicitly explore scale effects are of primary importance to understanding the underlying ecological processes driving large-scale responses. However, most studies that include spatial scale do so indirectly (e.g. <xref rid="PLU061C56" ref-type="bibr">Takagi and Miyashita 2014</xref>). Results of this study, although exploratory, do suggest that spatial scales play a role in modifying effect sizes of climate change response in plants. Although I found signals of scale effects in response to experimental treatments overall, these signals can be context dependent (<xref rid="PLU061C8" ref-type="bibr">Dent <italic>et al.</italic> 2001</xref>), and perhaps a different type of treatment (elevated CO<sub>2</sub>, for example) may elicit different relationships. Clearly, it is difficult to draw robust conclusions from a single case study, as only a small number of species and treatment effects are involved. The detection of overarching scaling effects often requires a large number of studies in order to obtain a reasonable signal-to-noise ratio. Nevertheless, I argue that it is important to use single case studies to verify the effects of spatial (and temporal) scaling. Such efforts have become more common recently (<xref rid="PLU061C24" ref-type="bibr">Heffernan <italic>et al.</italic> 2014</xref>), and it must become a more regular part of experimental research in order to develop our understanding of the complex relationships driving ecological patterns.</p></sec><sec id="s6"><title>Sources of Funding</title><p>Funding was provided by Tall Timbers Research Station, T. E. Miller and Florida State University.</p></sec><sec id="s7"><title>Contributions by the Authors</title><p>E.S.G. executed the experiment, collected and analysed all data and wrote the manuscript.</p></sec><sec id="s8"><title>Conflicts of Interest Statement</title><p>None declared.</p></sec>
Bifurcations and dynamics of a discrete predator–prey system	<p>In this paper, we study the dynamics behaviour of a stratum of plant–herbivore which is modelled through the following <italic>F</italic>(<italic>x, y</italic>)=(<italic>f</italic>(<italic>x, y</italic>), <italic>g</italic>(<italic>x, y</italic>)) two-dimensional map with four parameters defined by <disp-formula id="UM0001"><graphic xlink:href="tjbd-8-161-u001.jpg" position="float" orientation="portrait"/></disp-formula> where <italic>x</italic>≥0, <italic>y</italic>≥0, and the real parameters <italic>a, b, r, k</italic> are all positive. We will focus on the case <italic>a</italic>≠<italic>b</italic>. We study the stability of fixed points and do the analysis of the period-doubling and the Neimark–Sacker bifurcations in a standard way.</p>	<contrib contrib-type="author"><name><surname>Asheghi</surname><given-names>Rasoul</given-names></name><xref ref-type="aff" rid="AF1"> <sup>a</sup> </xref><xref ref-type="corresp" rid="cor1"> <sup>*</sup> </xref></contrib><aff id="AF1"><label><sup>a</sup></label><institution><named-content content-type="department">Department of Mathematical Sciences</named-content>, <named-content content-type="institution-name">Isfahan University of Technology</named-content></institution>, <named-content content-type="city">Isfahan</named-content><named-content content-type="postal-code">84156</named-content>, <country>Iran</country></aff>	Journal of Biological Dynamics	<sec sec-type="intro" id="S001"><title>Introduction</title><p>The dynamics of a discrete-time mathematical model arises in a variety of applications especially in ecology and mathematical biology and is still a current topic of research. In ecology, predator–prey or plant–herbivore models can be formulated as discrete-time models. A plant–herbivore model (which is also of host–parasite type) applies to study the interaction between a plant species and a herbivore species. Discrete models described by difference equations for interacting populations are of considerable interest to biologists and agricultural ecologists. These models are actually more reasonable than the continuous-time models when populations have non-overlapping generations. This certainly happens for a population that has a one-year life cycle, such as insects. For example, gypsy moth larvae hatch from the egg mass after bud-break and feed on new leaves. At the end of the season, adult gypsy moths lay eggs and die. The gypsy moth is a notorious forest pest in North Central USA whose outbreaks are almost periodic and cause significant damage to the infested forests. A discrete-time model can exhibit more plentiful and hence more complicated dynamical behaviours than a continuous-time model of the same type. Jing and Yang [<xref rid="CIT0016" ref-type="bibr">16</xref>] and Liu and Xiao [<xref rid="CIT0019" ref-type="bibr">19</xref>] remarked that discrete-time prey–predator models have more complicated dynamics than those in continuous models.</p><p>In recent years, a significant number of published papers have been on mathematical models in biology. Mathematical models on prey–predator systems have created a major interest during the last few decades. Study of such systems with discrete- and continuous-time models can be found in [<xref rid="CIT0001" ref-type="bibr">1</xref>,<xref rid="CIT0004" ref-type="bibr">4</xref>,<xref rid="CIT0008" ref-type="bibr">8</xref>,<xref rid="CIT0011" ref-type="bibr">11–15</xref>,<xref rid="CIT0017" ref-type="bibr">17</xref>,<xref rid="CIT0019" ref-type="bibr">19</xref>].</p><p>Summers and colleagues [<xref rid="CIT0007" ref-type="bibr">7</xref>] have analysed four typical discrete-time ecosystem models under the effects of periodic forcing. They observed that periodic forcing can produce a chaotic dynamics. Agiza <italic>et al</italic>. [<xref rid="CIT0002" ref-type="bibr">2</xref>] found chaotic dynamics of a discrete prey–predator model with Holling's Type II response function. They did not consider the natural death rate of the predators. Kang and colleagues [<xref rid="CIT0004" ref-type="bibr">4</xref>] observed quasi-periodicity, period-doubling and chaos in plant–herbivore interaction in the form of a host–parasite model. It is observed that the continuous model of such a system shows global stability of an interior equilibrium point.</p><p>Discrete-generation host–parasite models of the general form <disp-formula-group id="M0001"><disp-formula><graphic xlink:href="tjbd-8-161-e001.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0002"><disp-formula><graphic xlink:href="tjbd-8-161-e002.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> have been used to model the interaction between a host species (a plant) and a parasite species (a herbivore). In such models, <italic>P</italic> <sub><italic>n</italic></sub> and <italic>H</italic> <sub><italic>n</italic></sub> denote the population biomass of the host (a plant) and the parasite (a herbivore) in successive generations <italic>n</italic> and <italic>n</italic>+1, respectively. Here λ is the host's inherent rate of increase in the absence of the parasites, <italic>c</italic> is the biomass conversion constant and the function <italic>f</italic> represents the fraction of hosts surviving parasitism in each generation. Alternatively, <italic>f</italic> can be interpreted as the probability that each individual host escapes the parasites, so that the complementary term 1−<italic>f</italic> in the second equation is the probability of being parasitized.</p><p>The simplest version of this model is that of Nicholson–Bailey [<xref rid="CIT0005" ref-type="bibr">5</xref>]: <disp-formula-group id="M0003"><disp-formula><graphic xlink:href="tjbd-8-161-e003.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0004"><disp-formula><graphic xlink:href="tjbd-8-161-e004.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> in which the proportion of hosts escaping parasitism is given by the Poisson distribution <inline-formula><inline-graphic xlink:href="tjbd-8-161-m001.jpg"/></inline-formula>, where <italic>a</italic> is the mean encounters per host. Thus, <inline-formula><inline-graphic xlink:href="tjbd-8-161-m002.jpg"/></inline-formula> is the probability that a host will be attacked.</p><p>Beddington <italic>et al</italic>. [<xref rid="CIT0006" ref-type="bibr">6</xref>] considered an extension of the Nicholson–Bailey host–parasite model: <disp-formula-group id="M0005"><disp-formula><graphic xlink:href="tjbd-8-161-e005.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0006"><disp-formula><graphic xlink:href="tjbd-8-161-e006.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <italic>P</italic> <sub>max</sub> is the so-called environment-imposed ‘carrying capacity’ for the host in the absence of the parasite. The host density dependence is of the form <inline-formula><inline-graphic xlink:href="tjbd-8-161-m003.jpg"/></inline-formula>. From Equations (5) and (6) we see that the host density dependence acts at a particular time in their life cycle in relation to the stage attacked by the parasites. The <italic>H</italic> <sub><italic>n</italic></sub> herbivores search for <italic>P</italic> <sub><italic>n</italic></sub> hosts before the density-dependent growth regulation takes effect. Hence, the next generation of herbivores depends on <italic>P</italic> <sub><italic>n</italic></sub>, the initial host population prior to parasitism.</p><p>In order to have a more realistic model, one can consider the system <disp-formula-group id="M0007"><disp-formula><graphic xlink:href="tjbd-8-161-e007.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0008"><disp-formula><graphic xlink:href="tjbd-8-161-e008.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>In the case <italic>b</italic>=<italic>a</italic>, this is the density-dependent predator–prey model studied by Beddington <italic>et al</italic>. [<xref rid="CIT0006" ref-type="bibr">6</xref>]. By choosing an appropriate family chart (a parameter-dependent family of coordinate changes), one can assume <italic>b</italic>=<italic>r</italic>, and then the behaviour of the two populations just depends on the three positive parameters <italic>a, r</italic> and <italic>k</italic>. This system has the advantage that the dynamics restricted to <inline-formula><inline-graphic xlink:href="tjbd-8-161-m004.jpg"/></inline-formula> is given by the Ricker difference equation <inline-formula><inline-graphic xlink:href="tjbd-8-161-m005.jpg"/></inline-formula>, so the growth of the prey is limited and does not become unbounded.</p><p>Density-dependent models of the general form <disp-formula-group id="M0009"><disp-formula><graphic xlink:href="tjbd-8-161-e009.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0010"><disp-formula><graphic xlink:href="tjbd-8-161-e010.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> are considered in [<xref rid="CIT0003" ref-type="bibr">3</xref>], where <italic>f</italic> denotes the fraction of prey surviving predation in each generation. For <italic>b</italic>≠<italic>a</italic>, system (7) and (8) is not of this type. We prove that this system can include both the features of the Neimark–Sacker bifurcation appearing in the system <disp-formula-group id="M0011"><disp-formula><graphic xlink:href="tjbd-8-161-e011.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0012"><disp-formula><graphic xlink:href="tjbd-8-161-e012.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> and the period-doubling bifurcations that are inherited from the Ricker map.</p><p>The existence of a Neimark–Sacker bifurcation in the model implies that both the host and parasite populations can oscillate around some mean values, and that these oscillations are stable and will continue indefinitely under suitable conditions.</p><p>It is proved in [<xref rid="CIT0012" ref-type="bibr">12</xref>] that system (11) and (12) has a non-trivial steady-state solution which is stable for a certain range of parameter values, which is explicitly determined, and also undergoes a Neimark–Sacker bifurcation that produces an attracting invariant curve in some areas of the parameter space and a repelling one in others.</p><p>A general plant–herbivore model of the form <disp-formula-group id="M0013"><disp-formula><graphic xlink:href="tjbd-8-161-e013.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0014"><disp-formula><graphic xlink:href="tjbd-8-161-e014.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> was considered in [<xref rid="CIT0004" ref-type="bibr">4</xref>], making the following assumptions:</p><p> <italic>Assumption 1</italic> <italic>P</italic> <sub><italic>n</italic></sub> represents the biomass of the plant population (nutritious) after the attacks by the herbivore but before its defoliation. <italic>H</italic> <sub><italic>n</italic></sub> represents the biomass of the herbivore population before they die at the end of the season <italic>n</italic>.</p><p> <italic>Assumption 2</italic> Without the herbivore, the biomass of the plant population follows the dynamics of the Ricker model [<xref rid="CIT0021" ref-type="bibr">21</xref>], namely <disp-formula-group id="M0015"><disp-formula><graphic xlink:href="tjbd-8-161-e015.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> with a constant growth rate <italic>r</italic> and plant carrying capacity <italic>P</italic> <sub>max</sub>. The Ricker dynamics determines the amount of new leaves available for consumption for the herbivore.</p><p> <italic>Assumption 3</italic> It is supposed that the herbivores search for food randomly. The leaf area consumed is measured by the parameter <italic>a</italic>, i.e. <italic>a</italic> is a constant that correlates the total amount of the biomass that an herbivore consumes. The herbivore has a one-year life cycle, the larger the <italic>a</italic>, the faster is the feeding rate.</p><p> <italic>Assumption 4</italic> After attacks by herbivores, the biomass in the plant population is reduced to a fraction <inline-formula><inline-graphic xlink:href="tjbd-8-161-m006.jpg"/></inline-formula> of that present in the absence of herbivores. Hence, <disp-formula id="UM0002"><graphic xlink:href="tjbd-8-161-u002.jpg" position="float" orientation="portrait"/></disp-formula> </p><p> <italic>Assumption 5</italic> The amount of decreased biomass in the plants is converted to the biomass of the herbivore. It is assumed that the biomass conversion constant <italic>c</italic> is 1. Therefore, at the end of the season <italic>n</italic>, one has that <disp-formula id="UM0003"><graphic xlink:href="tjbd-8-161-u003.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>The bifurcation diagram in the parameter space of Equations (13) and (14) was presented by Armbruster <italic>et al</italic>. [<xref rid="CIT0004" ref-type="bibr">4</xref>]. Two control strategies are suggested in [<xref rid="CIT0004" ref-type="bibr">4</xref>]: reducing the population of the herbivore under some threshold and increasing the growth rate of the plant leaves.</p><p>Our paper is organized as follows. In Section 2, we give a brief discussion on the model under consideration. Fixed points and their stability are discussed in Section 3. Bifurcations analysis of the model is performed in Section 4.</p></sec><sec id="S002"><label>2. </label><title>Model</title><p>Consider the following <inline-formula><inline-graphic xlink:href="tjbd-8-161-m007.jpg"/></inline-formula> two-dimensional map with four parameters defined by <disp-formula-group id="M0016"><disp-formula><graphic xlink:href="tjbd-8-161-e016.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <italic>x</italic>≥0, <italic>y</italic>≥0, and the real parameters <italic>a, b, r, k</italic> are all positive. Since we are interested in the case <italic>a</italic>≠<italic>b</italic>, we assume that <italic>a</italic>≠<italic>b</italic>. The corresponding recurrence equations are written as <disp-formula-group id="M0017"><disp-formula><graphic xlink:href="tjbd-8-161-e017.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <italic>x</italic> <sub><italic>n</italic></sub>≥0 and <italic>y</italic> <sub><italic>n</italic></sub>≥0. This is a generalized Beddington host–parasitoid model to study the interaction of certain plants and herbivores, where <italic>x</italic> <sub><italic>n</italic></sub> and <italic>y</italic> <sub><italic>n</italic></sub> stand for the density of two populations at time <italic>n</italic>, and <italic>r</italic> is the growth rate. In [<xref rid="CIT0010" ref-type="bibr">10</xref>], Elaydi <italic>et al</italic>. investigate the stability and invariant manifolds of this model and also the stability of the coexistence fixed point. Elaydi <italic>et al</italic>. [<xref rid="CIT0010" ref-type="bibr">10</xref>] obtain the stability region for positive fixed point in parameter space by using a numerical method.</p><p>In this paper, we study the stability of fixed points and do the analysis of the period-doubling and the Neimark–Sacker bifurcations in a standard way. Using an appropriate rescaling <inline-formula><inline-graphic xlink:href="tjbd-8-161-m008.jpg"/></inline-formula>, we can without loss of generality suppose that <italic>b</italic>=<italic>r</italic>>0 and <italic>a</italic>≠<italic>r</italic>. One can, of course, choose another family chart which permits to take <italic>b</italic>=1 and <italic>a</italic>≠1. For convenience, we prefer here to take <italic>b</italic>=<italic>r</italic> and <italic>a</italic>≠<italic>r</italic>. In that case, we can have at most three fixed points at <inline-formula><inline-graphic xlink:href="tjbd-8-161-m009.jpg"/></inline-formula>, where <inline-formula><inline-graphic xlink:href="tjbd-8-161-m010.jpg"/></inline-formula> and <italic>y</italic>* is the unique positive solution of <disp-formula-group id="M0018"><disp-formula><graphic xlink:href="tjbd-8-161-e018.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>Also we have the following estimates on <italic>x</italic>, <italic>y</italic>: <disp-formula id="UM0004"><graphic xlink:href="tjbd-8-161-u004.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Let <disp-formula id="UM0005"><graphic xlink:href="tjbd-8-161-u005.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Then the interior equilibria are the intersection points of the functions <italic>f</italic> <sub>1</sub> and <italic>f</italic> <sub>2</sub> in the first quadrant. In addition, we have the following observations: <italic>f</italic> <sub>1</sub>(0)=0, <inline-formula><inline-graphic xlink:href="tjbd-8-161-m011.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-161-m012.jpg"/></inline-formula>; <italic>f</italic> <sub>2</sub>(0)=0, <inline-formula><inline-graphic xlink:href="tjbd-8-161-m013.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-161-m014.jpg"/></inline-formula>; <inline-formula><inline-graphic xlink:href="tjbd-8-161-m015.jpg"/></inline-formula> for <italic>y</italic>∈(0, 1) and <italic>a</italic>≤1/<italic>k</italic> while it has a unique zero at some <italic>y</italic>∈(0, 1) when <italic>a</italic>>1/<italic>k</italic>; and that <disp-formula id="UM0006"><graphic xlink:href="tjbd-8-161-u006.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Thus if <italic>ak</italic>>1, then <italic>F</italic>(<italic>y</italic>)>0 for 0<<italic>y</italic>≪1 and if <italic>ak</italic>≤1, then <italic>F</italic>(<italic>y</italic>)<0 for 0<<italic>y</italic>≪1. Hence, the intersection of <italic>f</italic> <sub>1</sub>(<italic>y</italic>) and <italic>f</italic> <sub>2</sub>(<italic>y</italic>) in the first quadrant has 1 (<xref rid="F0001" ref-type="fig">Figure 1</xref>) or 0 (<xref rid="F0002" ref-type="fig">Figure 2</xref>) positive fixed point depending on the sign of <italic>ak</italic>−1. Therefore, we have the following propositions: <fig id="F0001" orientation="portrait" position="float"><label>Figure 1. </label><caption><p>Graph of the functions <italic>f</italic> <sub>1</sub> and <italic>f</italic> <sub>2</sub> in the case 0<<italic>a</italic>≤1/<italic>k</italic>.</p></caption><graphic xlink:href="tjbd-8-161-g001"/></fig> <fig id="F0002" orientation="portrait" position="float"><label>Figure 2. </label><caption><p>Graph of the functions <italic>f</italic> <sub>1</sub> and <italic>f</italic> <sub>2</sub> in the case <italic>a</italic>>1/<italic>k</italic>.</p></caption><graphic xlink:href="tjbd-8-161-g002"/></fig> </p><statement id="E0001"><label>Proposition 2.1 </label><p>For any <italic>r</italic>>0 and 0<<italic>a</italic>≤1/<italic>k</italic>, system (17) has no positive fixed points (<xref rid="F0001" ref-type="fig">Figure 1</xref>).</p></statement><statement id="E0002"><label>Proposition 2.2 </label><p>For any <italic>r</italic>>0 and <italic>a</italic>>1/<italic>k</italic>, system (17) has a unique positive fixed point (<xref rid="F0002" ref-type="fig">Figure 2</xref>).</p></statement><p>Now we turn to investigate the stability of system (17) by taking <italic>b</italic>=<italic>r</italic> and <italic>a</italic>≠<italic>r</italic>.</p></sec><sec id="S003"><label>3. </label><title>Stability of fixed points</title><p>In this section, we study the stability of fixed points via linearization, namely using the linear part of Equation (16) evaluated at each fixed point. To this end, we need the partial derivatives of <italic>f</italic> and <italic>g</italic>, which are as follows: <disp-formula id="UM0007"><graphic xlink:href="tjbd-8-161-u007.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Since <inline-formula><inline-graphic xlink:href="tjbd-8-161-m016.jpg"/></inline-formula>, then the Jacobian matrix at the origin is <disp-formula id="UM0008"><graphic xlink:href="tjbd-8-161-u008.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Hence, the extinction fixed point (0, 0) is a saddle point. This implies that plants cannot die out. It is noted that the lines <inline-formula><inline-graphic xlink:href="tjbd-8-161-m017.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-161-m018.jpg"/></inline-formula> are invariant under the dynamics of Equation (16) such that <inline-formula><inline-graphic xlink:href="tjbd-8-161-m019.jpg"/></inline-formula> is the stable manifold of the origin (0, 0) and <inline-formula><inline-graphic xlink:href="tjbd-8-161-m020.jpg"/></inline-formula> denotes the unstable manifold. The map restricted to <inline-formula><inline-graphic xlink:href="tjbd-8-161-m021.jpg"/></inline-formula> is the Ricker map given by <disp-formula-group id="M0019"><disp-formula><graphic xlink:href="tjbd-8-161-e019.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>It is noted that the Schwartz derivative of the Ricker map <italic>R</italic>(<italic>x</italic>) defined in Equation (19) at <italic>x</italic>=<italic>k</italic> is calculated as follows: <disp-formula-group id="M0020"><disp-formula><graphic xlink:href="tjbd-8-161-e020.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>This derivative is negative provided that <italic>r</italic>≠1. Since for <italic>r</italic>=2 we have <inline-formula><inline-graphic xlink:href="tjbd-8-161-m022.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-161-m023.jpg"/></inline-formula>, in this case, by the well-known arguments, any trajectory with the initial condition (<italic>x</italic> <sub>0</sub>, 0) on the <italic>x</italic>-axis with <italic>x</italic> <sub>0</sub>>0 goes towards the exclusion fixed point (<italic>k</italic>, 0). To obtain the Jacobian matrix at the exclusion fixed point (<italic>k</italic>, 0), we have that <disp-formula id="UM0009"><graphic xlink:href="tjbd-8-161-u009.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>The eigenvalues of the above matrix are given by <inline-formula><inline-graphic xlink:href="tjbd-8-161-m024.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-161-m025.jpg"/></inline-formula>. Depending on the location of the eigenvalues in the complex plane w.r.t. the unit circle, we have the following statements on (<italic>k</italic>, 0):</p><statement id="E0003"><label>Proposition 3.1 </label><p>The following hold. <list list-type="order"><list-item><p>If <italic>ak</italic><1 and 0<<italic>r</italic><2, then it is an attracting node.</p></list-item><list-item><p>If <italic>ak</italic><1 and <italic>r</italic>>2, then it is a saddle point.</p></list-item><list-item><p>If <italic>ak</italic>>1 and <italic>r</italic>>2, then it is a repelling node.</p></list-item><list-item><p>If <italic>ak</italic>>1 and 0<<italic>r</italic><2, then it is a saddle point.</p></list-item></list> </p></statement><p> <italic>Proof</italic> The proof is straightforward. For example, if <italic>ak</italic><1 and 0<<italic>r</italic><2, then <inline-formula><inline-graphic xlink:href="tjbd-8-161-m026.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-161-m027.jpg"/></inline-formula>; hence, we have a stable node at (<italic>k</italic>, 0).</p><statement id="E0004"><label>Proposition 3.2 </label><p>When <italic>ak</italic>=1 and 0<<italic>r</italic><2, we have a non-hyperbolic fixed point at (<italic>k</italic>, 0), which is unstable.</p></statement><p> <italic>Proof</italic> In the case <italic>ak</italic>=1 and 0<<italic>r</italic><2, we compute below the centre manifold at (<italic>k</italic>, 0) and the dynamics restricted to the centre manifold to determine the orbit structure near (<italic>k</italic>, 0) in the first quadrant <inline-formula><inline-graphic xlink:href="tjbd-8-161-m028.jpg"/></inline-formula>. First, we bring the fixed point (<italic>k</italic>, 0) to the origin by using a linear translation, i.e. <inline-formula><inline-graphic xlink:href="tjbd-8-161-m029.jpg"/></inline-formula>. This yields the following <inline-formula><inline-graphic xlink:href="tjbd-8-161-m030.jpg"/></inline-formula> two-dimensional map with two parameters defined by <disp-formula-group id="M0021"><disp-formula><graphic xlink:href="tjbd-8-161-e021.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>Next, we put the linear part of Equation (21) in Jordan normal form by using the linear transformation <disp-formula id="UM0010"><graphic xlink:href="tjbd-8-161-u010.jpg" position="float" orientation="portrait"/></disp-formula> This induces the following two-dimensional map: <disp-formula id="UM0011"><graphic xlink:href="tjbd-8-161-u011.jpg" position="float" orientation="portrait"/></disp-formula> Thus, the centre manifold is given by the graph of <disp-formula id="UM0012"><graphic xlink:href="tjbd-8-161-u012.jpg" position="float" orientation="portrait"/></disp-formula> Then the dynamics on the centre manifold is given by the following scalar map: <disp-formula id="UM0013"><graphic xlink:href="tjbd-8-161-u013.jpg" position="float" orientation="portrait"/></disp-formula> which shows that <italic>v</italic>=0 is unstable. If we return to the original coordinates, then the centre manifold at (<italic>k</italic>, 0) takes the form <disp-formula id="UM0014"><graphic xlink:href="tjbd-8-161-u014.jpg" position="float" orientation="portrait"/></disp-formula> This centre manifold looks like a line near (<italic>k</italic>, 0).</p><statement id="E0005"><label>Proposition 3.3 </label><p>When <italic>r</italic>=2 and <italic>ak</italic><1, the exclusion fixed point (<italic>k</italic>, 0) is asymptotically stable.</p></statement><p> <italic>Proof</italic> Similar to the proof of Proposition 3.2, for the case <italic>r</italic>=2 and <italic>ak</italic><1, we can get <inline-formula><inline-graphic xlink:href="tjbd-8-161-m031.jpg"/></inline-formula> and <disp-formula id="UM0015"><graphic xlink:href="tjbd-8-161-u015.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>This shows that <italic>u</italic>=0 is asymptotically stable. This fact is consistent with Equation (20) for <italic>r</italic>=2.</p><p> <italic>Remark 3.4</italic> If either <italic>ak</italic>>1, <italic>r</italic>=2 or <italic>ak</italic>=1, <italic>r</italic>>2, then we have either <inline-formula><inline-graphic xlink:href="tjbd-8-161-m032.jpg"/></inline-formula> or <inline-formula><inline-graphic xlink:href="tjbd-8-161-m033.jpg"/></inline-formula>, respectively. In consequence, there is a one-dimensional unstable direction and a one-dimensional centre direction at the point (<italic>k</italic>, 0). Therefore, this fixed point is unstable. Using the centre manifold theory, one can compute the centre manifold at this point.</p><p> <italic>Remark 3.5</italic> In the case <italic>b</italic>=<italic>r</italic>=2 and <italic>a</italic>=1/<italic>k</italic>, system (17) reduces to <disp-formula-group id="M0022"><disp-formula><graphic xlink:href="tjbd-8-161-e022.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> which has no positive fixed point. The extinction fixed point (0, 0) is a hyperbolic saddle with eigenvalues <inline-formula><inline-graphic xlink:href="tjbd-8-161-m034.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-161-m035.jpg"/></inline-formula>. For this fixed point, the positive <italic>y</italic>-axis is the stable invariant manifold and the segment <inline-formula><inline-graphic xlink:href="tjbd-8-161-m036.jpg"/></inline-formula> on the <italic>x</italic>-axis is the unstable invariant manifold. The exclusion fixed point (<italic>k</italic>, 0) is a non-hyperbolic fixed point with eigenvalues <inline-formula><inline-graphic xlink:href="tjbd-8-161-m037.jpg"/></inline-formula>. Numerical evidences show that any trajectory of Equation (22) with starting point contained in the first quadrant converges to the fixed point (<italic>k</italic>, 0) on the <italic>x</italic>-axis. But we were not able to prove analytically this claim. Consider now an arbitrary orbit <inline-formula><inline-graphic xlink:href="tjbd-8-161-m038.jpg"/></inline-formula> with initial condition (<italic>x</italic> <sub>0</sub>, <italic>y</italic> <sub>0</sub>) located in the first quadrant, namely with <italic>x</italic> <sub>0</sub>>0 and <italic>y</italic> <sub>0</sub>>0. First we observe that near the origin we have <inline-formula><inline-graphic xlink:href="tjbd-8-161-m039.jpg"/></inline-formula>, therefore <inline-formula><inline-graphic xlink:href="tjbd-8-161-m040.jpg"/></inline-formula>, and hence <inline-formula><inline-graphic xlink:href="tjbd-8-161-m041.jpg"/></inline-formula>. Now let us consider the functions <italic>f</italic>(<italic>x</italic>) and <italic>g</italic>(<italic>y</italic>) defined by <disp-formula id="UM0016"><graphic xlink:href="tjbd-8-161-u016.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Because of having the estimates <inline-formula><inline-graphic xlink:href="tjbd-8-161-m042.jpg"/></inline-formula> and <italic>g</italic>(<italic>y</italic>)≤1/<italic>k</italic>, we see that <disp-formula id="UM0017"><graphic xlink:href="tjbd-8-161-u017.jpg" position="float" orientation="portrait"/></disp-formula> Here we write <italic>e</italic> for exp(1). As a result, after two iterations, our orbit lies entirely in the square <inline-formula><inline-graphic xlink:href="tjbd-8-161-m043.jpg"/></inline-formula> defined as <inline-formula><inline-graphic xlink:href="tjbd-8-161-m044.jpg"/></inline-formula>.</p><p>In this way, we are finished with (<italic>k</italic>, 0). Now, we pay our attention to the positive fixed point (<italic>x</italic>, <italic>y</italic>) that exists for <italic>ak</italic>>1. The Jacobian matrix of Equation (16) at this fixed point is given by the following: <disp-formula id="UM0018"><graphic xlink:href="tjbd-8-161-u018.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>A direct computation implies that <disp-formula id="UM0019"><graphic xlink:href="tjbd-8-161-u019.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>By the Jury test [<xref rid="CIT0009" ref-type="bibr">9</xref>], we see that the positive fixed point is locally asymptotically stable if <inline-formula><inline-graphic xlink:href="tjbd-8-161-m045.jpg"/></inline-formula>. The biological implication of this inequality is simple: if it holds, then the plant–herbivore interaction exhibits simple stable steady-state dynamics. In domain <inline-formula><inline-graphic xlink:href="tjbd-8-161-m046.jpg"/></inline-formula> of the parameter space, we have that (<italic>x</italic>, <italic>y</italic>) is locally attractive. It follows from Equation (18) that <disp-formula-group id="M0023"><disp-formula><graphic xlink:href="tjbd-8-161-e023.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>Since the function <disp-formula id="UM0020"><graphic xlink:href="tjbd-8-161-u020.jpg" position="float" orientation="portrait"/></disp-formula> is strictly increasing on (0, <italic>k</italic>/(<italic>k</italic>+1)) and <italic>E</italic>(0)=0, then <italic>E</italic>(<italic>y</italic>)><italic>E</italic>(0)=0 and hence <disp-formula-group id="M0024"><disp-formula><graphic xlink:href="tjbd-8-161-e024.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>Under the assumption det(<italic>J</italic>)<1 and using Equation (23), we get <disp-formula-group id="M0025"><disp-formula><graphic xlink:href="tjbd-8-161-e025.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula-group id="M0026"><disp-formula><graphic xlink:href="tjbd-8-161-e026.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>It is easily seen that <disp-formula id="UM0021"><graphic xlink:href="tjbd-8-161-u021.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Let us define the following functions: <disp-formula-group id="M0027"><disp-formula><graphic xlink:href="tjbd-8-161-e027.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0028"><disp-formula><graphic xlink:href="tjbd-8-161-e028.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0029"><disp-formula><graphic xlink:href="tjbd-8-161-e029.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>Then <inline-formula><inline-graphic xlink:href="tjbd-8-161-m047.jpg"/></inline-formula> and for <italic>k</italic>>1 we have the following properties: <inline-formula><inline-graphic xlink:href="tjbd-8-161-m048.jpg"/></inline-formula>; <inline-formula><inline-graphic xlink:href="tjbd-8-161-m049.jpg"/></inline-formula>; <inline-formula><inline-graphic xlink:href="tjbd-8-161-m050.jpg"/></inline-formula>. This implies that for <italic>k</italic>>1 and <italic>Y</italic>∈(0, 1), each one of the two functions <italic>g</italic> <sub>1</sub> and <italic>g</italic> <sub>2</sub> has a unique zero (<xref rid="F0003" ref-type="fig">Figure 3</xref>). <fig id="F0003" orientation="portrait" position="float"><label>Figure 3. </label><caption><p>Graph of the functions <italic>g</italic> <sub>1</sub> and <italic>g</italic> <sub>2</sub> in the case <italic>k</italic>>1.</p></caption><graphic xlink:href="tjbd-8-161-g003"/></fig> </p><p>For <italic>k</italic>>1, let <inline-formula><inline-graphic xlink:href="tjbd-8-161-m051.jpg"/></inline-formula> be the unique zeros of <italic>g</italic> <sub>1</sub>(<italic>Y</italic>) and <italic>g</italic> <sub>2</sub>(<italic>Y</italic>), respectively. Next we define <disp-formula-group id="M0030"><disp-formula><graphic xlink:href="tjbd-8-161-e030.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0031"><disp-formula><graphic xlink:href="tjbd-8-161-e031.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>Then, the unique solution <inline-formula><inline-graphic xlink:href="tjbd-8-161-m052.jpg"/></inline-formula> of the equation <inline-formula><inline-graphic xlink:href="tjbd-8-161-m053.jpg"/></inline-formula> that exists for <inline-formula><inline-graphic xlink:href="tjbd-8-161-m054.jpg"/></inline-formula>, in which det(<italic>J</italic>)=1 and tr(<italic>J</italic>)=−2, lies in the interval <inline-formula><inline-graphic xlink:href="tjbd-8-161-m055.jpg"/></inline-formula> (<xref rid="F0004" ref-type="fig">Figure 4</xref>). Now, we can determine the basin of attraction Ω of (<italic>x</italic>, <italic>y</italic>), keeping <italic>a</italic>>1/<italic>k</italic> fixed, as <disp-formula id="UM0022"><graphic xlink:href="tjbd-8-161-u022.jpg" position="float" orientation="portrait"/></disp-formula> <fig id="F0004" orientation="portrait" position="float"><label>Figure 4. </label><caption><p>Position of <italic>Ŷ</italic> <sub>3</sub>(<italic>k</italic>) for <italic>k</italic>∈(⅓, 1].</p></caption><graphic xlink:href="tjbd-8-161-g004"/></fig> </p><p>The shape of the region Ω in (<italic>Y, r</italic>)-plane is shown in <xref rid="F0005" ref-type="fig">Figure 5</xref>. It is noted that Elaydi <italic>et al</italic>. [<xref rid="CIT0010" ref-type="bibr">10</xref>] obtain the stability region for positive fixed point in parameter space by using a numerical method. <fig id="F0005" orientation="portrait" position="float"><label>Figure 5. </label><caption><p>Shape of Ω in the case <italic>k</italic>>1.</p></caption><graphic xlink:href="tjbd-8-161-g005"/></fig> </p><sec id="S003-S2001"><label>3.1. </label><title>Case 0<<italic>k</italic>≤1</title><p>For the functions <italic>g</italic> <sub>1</sub>(<italic>Y</italic>) and <italic>g</italic> <sub>2</sub>(<italic>Y</italic>) defined in Equations (27) and (28), in this case, we have the following properties: <inline-formula><inline-graphic xlink:href="tjbd-8-161-m056.jpg"/></inline-formula> for <italic>Y</italic><1; <inline-formula><inline-graphic xlink:href="tjbd-8-161-m057.jpg"/></inline-formula>; <inline-formula><inline-graphic xlink:href="tjbd-8-161-m058.jpg"/></inline-formula>; <inline-formula><inline-graphic xlink:href="tjbd-8-161-m059.jpg"/></inline-formula>. This implies that for 0<<italic>k</italic>≤1 and <italic>Y</italic>∈(0, 1) the function <italic>g</italic> <sub>1</sub> has no zero, while <italic>g</italic> <sub>2</sub> has a unique zero (<xref rid="F0006" ref-type="fig">Figure 6</xref>). For 0<<italic>k</italic>≤1, suppose that <inline-formula><inline-graphic xlink:href="tjbd-8-161-m060.jpg"/></inline-formula> be the unique zero of <italic>g</italic> <sub>2</sub>(<italic>Y</italic>) and let <italic>Ŷ</italic> <sub>3</sub>(<italic>k</italic>) be as before, i.e. the unique solution of <inline-formula><inline-graphic xlink:href="tjbd-8-161-m061.jpg"/></inline-formula> where <inline-formula><inline-graphic xlink:href="tjbd-8-161-m062.jpg"/></inline-formula> and the functions <italic>h</italic> <sub>1</sub> and <italic>h</italic> <sub>2</sub> are given by Equations (30) and (31). Then we have <inline-formula><inline-graphic xlink:href="tjbd-8-161-m063.jpg"/></inline-formula>. Therefore, the stability region Ω for the positive fixed point, in this case, is also determined by <disp-formula id="UM0023"><graphic xlink:href="tjbd-8-161-u023.jpg" position="float" orientation="portrait"/></disp-formula> <fig id="F0006" orientation="portrait" position="float"><label>Figure 6. </label><caption><p>Graph of the functions <italic>g</italic> <sub>1</sub> and <italic>g</italic> <sub>2</sub> in the case 0<<italic>k</italic>≤1.</p></caption><graphic xlink:href="tjbd-8-161-g006"/></fig> </p><p>For a picture of Ω in this case, see <xref rid="F0007" ref-type="fig">Figure 7</xref>. <fig id="F0007" orientation="portrait" position="float"><label>Figure 7. </label><caption><p>Shape of Ω in the case 0<<italic>k</italic>≤1.</p></caption><graphic xlink:href="tjbd-8-161-g007"/></fig> </p><p>It is worth noting that for a given <italic>k</italic>>0, when one chooses a pair (<italic>Y</italic>, <italic>r</italic>) of Ω, the value of the parameter <italic>a</italic>>1/<italic>k</italic> is immediately specified by Equation (23).</p><p>Here, the local stability analysis of the fixed points is finished. Now we are in a position that our attention goes to treat bifurcations in the biological model under consideration. This will be done shortly in the next section.</p></sec></sec><sec id="S004"><label>4. </label><title>Bifurcations of fixed points</title><p>In this section, we are planning to do the analysis of the local bifurcations of the fixed points of the system (16), including Neimark–Sacker and period-doubling bifurcations.</p><sec id="S004-S2001"><label>4.1. </label><title>The Neimark–Sacker bifurcation</title><p>In the discrete setting, the Neimark–Sacker bifurcation is the analogue of the Hopf bifurcation that occurs in the continuous systems. It was discovered by Neimark [<xref rid="CIT0020" ref-type="bibr">20</xref>], and independently by Sacker [<xref rid="CIT0022" ref-type="bibr">22</xref>], who originally studied it in (connection) line with the stability of periodic solutions of ordinary differential equations, where it arises from the map obtained by taking a Poincare section transverse to the periodic flow. Hopf bifurcations create limit cycles in the phase plane of continuous models. On the other hand, Neimark–Sacker bifurcations generate dynamically invariant circles. As a result, we may find isolated periodic orbits as well as trajectories that cover the invariant circle densely. We seek conditions for Equation (16) to have a non-hyperbolic fixed point with a pair of complex conjugate eigenvalues of modulus 1. This happens surely at the interior fixed point (<italic>x</italic>, <italic>y</italic>). The associated Jacobian matrix <inline-formula><inline-graphic xlink:href="tjbd-8-161-m064.jpg"/></inline-formula> has two complex conjugate eigenvalues with modulus 1 in the case det(<italic>J</italic>)=1 and <inline-formula><inline-graphic xlink:href="tjbd-8-161-m065.jpg"/></inline-formula>. Hence, the candidate for the bifurcation curve is the curve <disp-formula-group id="M0032"><disp-formula><graphic xlink:href="tjbd-8-161-e032.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>We consider (<italic>k, a</italic>) as fixed and take <italic>r</italic>>0 as a parameter and write it as <italic>r</italic>=<italic>r</italic>+μ, where <disp-formula-group id="M0033"><disp-formula><graphic xlink:href="tjbd-8-161-e033.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>Following the standard way, we first must do some preliminary (linear or affine) transformations in order to put the linear part of the two-dimensional map (16) into normal form. This will be done shortly. Utilizing the linear transformation <disp-formula id="UM0024"><graphic xlink:href="tjbd-8-161-u024.jpg" position="float" orientation="portrait"/></disp-formula> we get the following <inline-formula><inline-graphic xlink:href="tjbd-8-161-m066.jpg"/></inline-formula> two-dimensional map defined by <disp-formula-group id="M0034"><disp-formula><graphic xlink:href="tjbd-8-161-e034.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>In the expanded form, we obtain that <disp-formula-group id="M0035"><disp-formula><graphic xlink:href="tjbd-8-161-e035.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>Now, the eigenvalues of the linear part of Equation (35) are given by <disp-formula-group id="M0036"><disp-formula><graphic xlink:href="tjbd-8-161-e036.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <inline-formula><inline-graphic xlink:href="tjbd-8-161-m067.jpg"/></inline-formula>. At μ=0 we have <disp-formula id="UM0025"><graphic xlink:href="tjbd-8-161-u025.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>More precisely, it follows from Equation (35) that <disp-formula id="UM0026"><graphic xlink:href="tjbd-8-161-u026.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>With a linear change of coordinates, we can put the two-dimensional map <inline-formula><inline-graphic xlink:href="tjbd-8-161-m068.jpg"/></inline-formula> defined by Equation (34) in the following form: <disp-formula id="UM0027"><graphic xlink:href="tjbd-8-161-u027.jpg" position="float" orientation="portrait"/></disp-formula> where <inline-formula><inline-graphic xlink:href="tjbd-8-161-m069.jpg"/></inline-formula>, <italic>i</italic>=1, 2, are nonlinear in <inline-formula><inline-graphic xlink:href="tjbd-8-161-m070.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-161-m071.jpg"/></inline-formula>. To simplify the notation, let us define <disp-formula-group id="M0037"><disp-formula><graphic xlink:href="tjbd-8-161-e037.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>We make the linear transformation <disp-formula id="UM0028"><graphic xlink:href="tjbd-8-161-u028.jpg" position="float" orientation="portrait"/></disp-formula> to obtain the following two-dimensional map defined by <disp-formula-group id="M0038"><disp-formula><graphic xlink:href="tjbd-8-161-e038.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula id="UM0029"><graphic xlink:href="tjbd-8-161-u029.jpg" position="float" orientation="portrait"/></disp-formula> We can simplify Equation (38) further and arrive at <disp-formula-group id="M0039"><disp-formula><graphic xlink:href="tjbd-8-161-e039.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula id="UM0030"><graphic xlink:href="tjbd-8-161-u030.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>To analyse the corresponding bifurcation, we introduce the complex variables <disp-formula id="UM0031"><graphic xlink:href="tjbd-8-161-u031.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>By Equation (39), the equation for <italic>w</italic> reads <disp-formula-group id="M0040"><disp-formula><graphic xlink:href="tjbd-8-161-e040.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula id="UM0032"><graphic xlink:href="tjbd-8-161-u032.jpg" position="float" orientation="portrait"/></disp-formula> with <disp-formula id="UM0033"><graphic xlink:href="tjbd-8-161-u033.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Now it follows from the normal form theorem for the Neimark–Sacker bifurcation that the one-dimensional map defined by Equation (40) can be transformed by an invertible parameter-dependent change of complex coordinate, which is smoothly dependent on the parameter, <disp-formula id="UM0034"><graphic xlink:href="tjbd-8-161-u034.jpg" position="float" orientation="portrait"/></disp-formula> for μ near zero, into a map with only the resonant cubic term: <disp-formula-group id="M0041"><disp-formula><graphic xlink:href="tjbd-8-161-e041.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula id="UM0035"><graphic xlink:href="tjbd-8-161-u035.jpg" position="float" orientation="portrait"/></disp-formula> </p><p> <italic>Remark 4.1</italic> Any such map (39) has the normal form (41) near μ=0 provided that <inline-formula><inline-graphic xlink:href="tjbd-8-161-m072.jpg"/></inline-formula> for <italic>q</italic>=1, 2, 3, 4. Subject to the further condition that <inline-formula><inline-graphic xlink:href="tjbd-8-161-m073.jpg"/></inline-formula>, for sufficiently small μ the map has an invariant closed curve enclosing the origin when <inline-formula><inline-graphic xlink:href="tjbd-8-161-m074.jpg"/></inline-formula>. In the case that <inline-formula><inline-graphic xlink:href="tjbd-8-161-m075.jpg"/></inline-formula>, the bifurcation is said to be supercritical, and there is a stable attracting invariant curve for small enough μ>0, while a subcritical bifurcation arises for <inline-formula><inline-graphic xlink:href="tjbd-8-161-m076.jpg"/></inline-formula>, when there is a repelling invariant curve for small μ<0 (see [<xref rid="CIT0018" ref-type="bibr">18</xref>] for more details).</p><p>Writing <inline-formula><inline-graphic xlink:href="tjbd-8-161-m077.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-161-m078.jpg"/></inline-formula>, expression (41) can be written as <disp-formula-group id="M0042"><disp-formula><graphic xlink:href="tjbd-8-161-e042.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula id="UM0036"><graphic xlink:href="tjbd-8-161-u036.jpg" position="float" orientation="portrait"/></disp-formula> Using the representation <inline-formula><inline-graphic xlink:href="tjbd-8-161-m079.jpg"/></inline-formula>, we obtain the following polar form of Equation (42): <disp-formula-group id="M0043"><disp-formula><graphic xlink:href="tjbd-8-161-e043.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> In the polar form (43), the ρ-map will depend on ϕ. If we neglect the higher order terms in Equation (43), then we will have the truncated polar form <disp-formula-group id="M0044"><disp-formula><graphic xlink:href="tjbd-8-161-e044.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Bifurcations of the phase portrait of Equation (42) as <italic>p</italic> passes through zero can easily be analysed using the latter form, since the mapping for ρ is independent of ϕ. The first equation in Equation (44) defines a one-dimensional discrete dynamical system that has the fixed point ρ=0 for all values of <italic>p</italic>. The point is linearly stable if <italic>p</italic><0; for <italic>p</italic>>0 the point becomes linearly unstable. The stability of the fixed point at <italic>p</italic>=0 is determined by the sign of the coefficient <italic>a</italic>(0). Suppose that <italic>a</italic>(0)<0; then the origin (ρ=0) is nonlinearly stable at <italic>p</italic>=0. Moreover, the ρ-map of Equation (44) has an additional stable fixed point <disp-formula id="UM0037"><graphic xlink:href="tjbd-8-161-u037.jpg" position="float" orientation="portrait"/></disp-formula> The ϕ-map of Equation (44) describes a rotation by an angle θ(<italic>p</italic>). Thus, by superposition of the mappings defined by Equation (44), we obtain the bifurcation diagram. The system always has a fixed point at the origin. This point is stable for <italic>p</italic><0 and unstable for <italic>p</italic>>0. The invariant curves of the system near the origin look like the orbits near the stable focus of a continuous-time system for <italic>p</italic><0 and like orbits near the unstable focus for <italic>p</italic>>0. At the critical parameter value <italic>p</italic>=0 the point is nonlinearly stable. The fixed point is surrounded for <italic>p</italic>>0 by an isolated closed invariant curve that is unique and stable. The curve is a circle of radius <inline-formula><inline-graphic xlink:href="tjbd-8-161-m080.jpg"/></inline-formula>. All orbits starting outside or inside the closed invariant curve, except at the origin, tend to the curve under iterations of Equation (44). This is a Neimark–Sacker bifurcation. The case <italic>a</italic>(0)>0 can be analysed in the same way. The system undergoes the Neimark–Sacker bifurcation at <italic>p</italic>=0. Contrary to the considered case, there is an unstable closed invariant curve that disappears when <italic>p</italic> crosses zero from negative to positive values. By Lemma 4.3 on page 128 (or Appendix 2) of [<xref rid="CIT0018" ref-type="bibr">18</xref>], the higher order terms in Equation (43) do not affect the bifurcation of the closed invariant curve and in fact a locally unique invariant curve bifurcates from the origin in the same direction and with the same stability as in system (44). The coefficient <italic>a</italic>(0), which determines the direction of the appearance of the invariant curve in a generic system exhibiting the Neimark–Sacker bifurcation, can be computed via <disp-formula-group id="M0045"><disp-formula><graphic xlink:href="tjbd-8-161-e045.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <inline-formula><inline-graphic xlink:href="tjbd-8-161-m081.jpg"/></inline-formula>. Since at μ=0 we have <inline-formula><inline-graphic xlink:href="tjbd-8-161-m082.jpg"/></inline-formula>, with <disp-formula id="UM0038"><graphic xlink:href="tjbd-8-161-u038.jpg" position="float" orientation="portrait"/></disp-formula> therefore <disp-formula-group id="M0046"><disp-formula><graphic xlink:href="tjbd-8-161-e046.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> with <disp-formula id="UM0039"><graphic xlink:href="tjbd-8-161-u039.jpg" position="float" orientation="portrait"/></disp-formula> The formulae for computing the coefficients of <italic>N</italic> <sub>1</sub> and <italic>N</italic> <sub>2</sub> are omitted due to very long expressions. It follows from Equation (46) that <disp-formula-group id="M0047"><disp-formula><graphic xlink:href="tjbd-8-161-e047.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> with <disp-formula id="UM0040"><graphic xlink:href="tjbd-8-161-u040.jpg" position="float" orientation="portrait"/></disp-formula> The formulas for specifying the coefficients of <italic>N</italic> are omitted due to very long expressions. Since <disp-formula id="UM0041"><graphic xlink:href="tjbd-8-161-u041.jpg" position="float" orientation="portrait"/></disp-formula> a unique closed invariant curve bifurcates from the positive fixed point when <inline-formula><inline-graphic xlink:href="tjbd-8-161-m083.jpg"/></inline-formula>. The sign of <italic>a</italic>(0) determines the stability of the invariant curve.</p><p> <italic>Remark 4.2</italic> Due to the technical nature of the coefficients (depending on the parameter) appeared for the expressions of <italic>N</italic> <sub>1</sub>, <italic>N</italic> <sub>2</sub>, <italic>D</italic> <sub>1</sub>, <italic>D</italic> <sub>2</sub> in Equation (46) and for the expressions of <italic>N, D</italic> in Equation (47), we decided to choose some values for the parameters (<italic>a, k</italic>) and then follow the whole procedure constructed along the paper. By this choice of parameters, we can obtain the exact value of the positive fixed point (<italic>x</italic>, <italic>y</italic>).</p><p>By following computations presented in the paper, we can finally find the value of <italic>a</italic>(0). For example, if we choose <italic>a</italic>=10 and <italic>k</italic>=0.5, then we find that <inline-formula><inline-graphic xlink:href="tjbd-8-161-m084.jpg"/></inline-formula> and then <inline-formula><inline-graphic xlink:href="tjbd-8-161-m085.jpg"/></inline-formula>. Thus, for <italic>r</italic>><italic>r</italic>* we have a closed invariant curve which is stable (<xref rid="F0008" ref-type="fig">Figure 8</xref>). On this bases, <xref rid="T0001" ref-type="table">Table 1</xref> is constructed by using numerical computations. In order to support the results, phase diagrams for some particular parameter values would be presented for the Neimark–Sacker bifurcation in <xref rid="F0008 F0009 F0010 F0011 F0012 F0013" ref-type="fig">Figures 8–13</xref>. <fig id="F0008" orientation="portrait" position="float"><label>Figure 8. </label><caption><p>Phase diagram when <italic>r</italic>=3.7, <italic>a</italic>=10, <italic>k</italic>=0.5, (<italic>x</italic> <sub>0</sub>, <italic>y</italic> <sub>0</sub>)=(0.3, 0.3) and <italic>n</italic>=25, 000.</p></caption><graphic xlink:href="tjbd-8-161-g008"/></fig> <fig id="F0009" orientation="portrait" position="float"><label>Figure 9. </label><caption><p>Phase diagram when <italic>r</italic>=3.6, <italic>a</italic>=10, <italic>k</italic>=0.5, (<italic>x</italic> <sub>0</sub>, <italic>y</italic> <sub>0</sub>)=(0.3, 0.3) and <italic>n</italic>=25, 000.</p></caption><graphic xlink:href="tjbd-8-161-g009"/></fig> <fig id="F0010" orientation="portrait" position="float"><label>Figure 10. </label><caption><p>Phase diagram when <italic>r</italic>=1.6, <italic>a</italic>=27, <italic>k</italic>=2, (<italic>x</italic> <sub>0</sub>, <italic>y</italic> <sub>0</sub>)=(0.6, 0.6) and <italic>n</italic>=25, 000.</p></caption><graphic xlink:href="tjbd-8-161-g010"/></fig> <fig id="F0011" orientation="portrait" position="float"><label>Figure 11. </label><caption><p>Phase diagram when <italic>r</italic>=1.49, <italic>a</italic>=27, <italic>k</italic>=2, (<italic>x</italic> <sub>0</sub>, <italic>y</italic> <sub>0</sub>)=(0.6, 0.6) and <italic>n</italic>=25, 000.</p></caption><graphic xlink:href="tjbd-8-161-g011"/></fig> <fig id="F0012" orientation="portrait" position="float"><label>Figure 12. </label><caption><p>Phase diagram when <italic>r</italic>=2.2, <italic>a</italic>=50, <italic>k</italic>=0.9, (<italic>x</italic> <sub>0</sub>, <italic>y</italic> <sub>0</sub>)=(0.5, 0.5) and <italic>n</italic>=25, 000.</p></caption><graphic xlink:href="tjbd-8-161-g012"/></fig> <fig id="F0013" orientation="portrait" position="float"><label>Figure 13. </label><caption><p>Phase diagram when <italic>r</italic>=2.0, <italic>a</italic>=50, <italic>k</italic>=0.9, (<italic>x</italic> <sub>0</sub>, <italic>y</italic> <sub>0</sub>)=(0.5, 0.5) and <italic>n</italic>=25, 000.</p></caption><graphic xlink:href="tjbd-8-161-g013"/></fig> <table-wrap id="T0001" orientation="portrait" position="float"><label>Table 1. </label><caption><title> The numerical exact values for the positive fixed point (<italic>x</italic>, <italic>y</italic>) and for the coefficient <italic>a</italic>(0) corresponding to the chosen values of the parameters (<italic>a, k</italic>).</title></caption><!--OASIS TABLE HERE--><table frame="hsides" rules="groups"><colgroup><col width="1"/><col width="1"/><col width="1"/><col width="1"/><col width="1"/><col width="1"/><col width="1"/><col width="1"/></colgroup><thead valign="bottom"><tr><th align="left"><italic>k</italic></th><th align="center"><italic>a</italic></th><th align="center"><italic>x</italic></th><th align="center"><italic>y</italic></th><th align="center"><italic>r</italic></th><th align="center">θ<sub>0</sub></th><th align="center"><italic>a</italic>(0)</th><th align="center"><italic>Ŷ</italic><sub>3</sub>(<italic>k</italic>)</th></tr></thead><tbody><tr><td align="left">0.5</td><td align="char" char=".">10</td><td align="center">0.337717</td><td align="center">0.324566</td><td align="char" char=".">3.68416</td><td align="char" char=".">2.31644</td><td align="char" char=".">−2.10532</td><td align="char" char=".">0.0787963</td></tr><tr><td align="left">0.5</td><td align="char" char=".">15</td><td align="center">0.334101</td><td align="center">0.331798</td><td align="char" char=".">3.12737</td><td align="char" char=".">2.12655</td><td align="char" char=".">−0.572171</td><td align="char" char=".">0.0787963</td></tr><tr><td align="left">0.5</td><td align="char" char=".">30</td><td align="center">0.333338</td><td align="center">0.333323</td><td align="char" char=".">3.00145</td><td align="char" char=".">2.09471</td><td align="char" char=".">−0.156109</td><td align="char" char=".">0.0787963</td></tr><tr><td align="left">0.75</td><td align="char" char=".">8</td><td align="center">0.435051</td><td align="center">0.419931</td><td align="char" char=".">2.51322</td><td align="char" char=".">1.74004</td><td align="char" char=".">−0.341566</td><td align="char" char=".">0.214038</td></tr><tr><td align="left">0.75</td><td align="char" char=".">5</td><td align="center">0.456131</td><td align="center">0.391825</td><td align="char" char=".">3.45667</td><td align="char" char=".">1.97182</td><td align="char" char=".">−2.16554</td><td align="char" char=".">0.214038</td></tr><tr><td align="left">0.75</td><td align="char" char=".">40</td><td align="center">0.428571</td><td align="center">0.428571</td><td align="char" char=".">2.33333</td><td align="char" char=".">1.73824</td><td align="char" char=".">−0.0850811</td><td align="char" char=".">0.214038</td></tr><tr><td align="left">0.9</td><td align="char" char=".">4</td><td align="center">0.518854</td><td align="center">0.423496</td><td align="char" char=".">3.03818</td><td align="char" char=".">1.75691</td><td align="char" char=".">−1.29779</td><td align="char" char=".">0.285252</td></tr><tr><td align="left">0.9</td><td align="char" char=".">7</td><td align="center">0.482578</td><td align="center">0.463803</td><td align="char" char=".">2.20824</td><td align="char" char=".">1.59711</td><td align="char" char=".">−0.176965</td><td align="char" char=".">0.285252</td></tr><tr><td align="left">0.9</td><td align="char" char=".">50</td><td align="center">0.473684</td><td align="center">0.473684</td><td align="char" char=".">2.11111</td><td align="char" char=".">1.59711</td><td align="char" char=".">−0.0533712</td><td align="char" char=".">0.285252</td></tr><tr><td align="left">1</td><td align="char" char=".">3</td><td align="center">0.583714</td><td align="center">0.416286</td><td align="char" char=".">4.04332</td><td align="char" char=".">2.0141</td><td align="char" char=".">−4.87002</td><td align="char" char=".">0.327559</td></tr><tr><td align="left">1</td><td align="char" char=".">13</td><td align="center">0.500378</td><td align="center">0.499622</td><td align="char" char=".">2.00154</td><td align="char" char=".">1.56665</td><td align="char" char=".">−0.0657914</td><td align="char" char=".">0.327559</td></tr><tr><td align="left">1</td><td align="char" char=".">23</td><td align="center">0.500003</td><td align="center">0.499997</td><td align="char" char=".">2.00001</td><td align="char" char=".">1.57074</td><td align="char" char=".">−0.0625699</td><td align="char" char=".">0.327559</td></tr><tr><td align="left">2</td><td align="char" char=".">3</td><td align="center">0.741519</td><td align="center">0.629241</td><td align="char" char=".">1.31488</td><td align="char" char=".">1.1322</td><td align="char" char=".">0.000485813</td><td align="char" char=".">0.588677</td></tr><tr><td align="left">2</td><td align="char" char=".">7</td><td align="center">0.670936</td><td align="center">0.664532</td><td align="char" char=".">1.47064</td><td align="char" char=".">1.29144</td><td align="char" char=".">−0.00438717</td><td align="char" char=".">0.588677</td></tr><tr><td align="left">2</td><td align="char" char=".">27</td><td align="center">0.666667</td><td align="center">0.666667</td><td align="char" char=".">1.5</td><td align="char" char=".">1.31812</td><td align="char" char=".">−0.0351532</td><td align="char" char=".">0.588677</td></tr><tr><td align="left">20</td><td align="char" char=".">10</td><td align="center">0.952447</td><td align="center">0.952378</td><td align="char" char=".">1.04931</td><td align="char" char=".">1.07542</td><td align="char" char=".">−0.0152574</td><td align="char" char=".">0.950933</td></tr></tbody></table></table-wrap> </p></sec><sec id="S004-S2002"><label>4.2. </label><title>Period-doubling bifurcation</title><p>In this subsection, we are planning to do the analysis of the period-doubling bifurcation. In the first step, let <italic>k</italic>>0 be fixed and consider the function <disp-formula-group id="M0048"><disp-formula><graphic xlink:href="tjbd-8-161-e048.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>Then <disp-formula id="UM0042"><graphic xlink:href="tjbd-8-161-u042.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Moreover, <disp-formula id="UM0043"><graphic xlink:href="tjbd-8-161-u043.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>For <italic>k</italic>>1 we have <italic>E</italic> <sub>1</sub>(<italic>Y</italic>)>0 when <inline-formula><inline-graphic xlink:href="tjbd-8-161-m086.jpg"/></inline-formula>, where <italic>Ŷ</italic> <sub>1</sub> is the unique solution of <italic>g</italic> <sub>1</sub>(<italic>Y</italic>)=0. Clearly the values of <italic>Ŷ</italic> <sub>1</sub> depend on 1<<italic>k</italic><∞ adding that <inline-formula><inline-graphic xlink:href="tjbd-8-161-m087.jpg"/></inline-formula> as <inline-formula><inline-graphic xlink:href="tjbd-8-161-m088.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-161-m089.jpg"/></inline-formula> as <inline-formula><inline-graphic xlink:href="tjbd-8-161-m090.jpg"/></inline-formula>, and <italic>Ŷ</italic> <sub>1</sub>=0.5 when <italic>k</italic>=3.38630. For 0<<italic>k</italic>≤1 the function <italic>E</italic> <sub>1</sub>(<italic>Y</italic>) is strictly decreasing and positive in the interval (0, 1). When <italic>k</italic>=1 this function is strictly decreasing from +∞ to +2.</p><p>We note that λ=1 is an eigenvalue of the Jacobian matrix <italic>J</italic> of the system (17) at the fixed point (<italic>x</italic>, <italic>y</italic>) if and only if <inline-formula><inline-graphic xlink:href="tjbd-8-161-m091.jpg"/></inline-formula> which yields <italic>r</italic>=0. Note that <italic>r</italic>=0 is not biologically of interest. On the other hand, λ=−1 is an eigenvalue of the Jacobian matrix <italic>J</italic> of the system (17) at the fixed point (<italic>x</italic>, <italic>y</italic>) if and only if <inline-formula><inline-graphic xlink:href="tjbd-8-161-m092.jpg"/></inline-formula> which yields <italic>r</italic>=<italic>r</italic> <sub>1</sub>, where <inline-formula><inline-graphic xlink:href="tjbd-8-161-m093.jpg"/></inline-formula> defined by Equation (48). The other eigenvalue of <italic>J</italic> for <italic>r</italic>=<italic>r</italic> <sub>1</sub> is given by <inline-formula><inline-graphic xlink:href="tjbd-8-161-m094.jpg"/></inline-formula>, where <disp-formula-group id="M0049"><disp-formula><graphic xlink:href="tjbd-8-161-e049.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>We have the following properties of the function <italic>E</italic> <sub>2</sub>(<italic>Y</italic>) defined by Equation (49): <disp-formula id="UM0044"><graphic xlink:href="tjbd-8-161-u044.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>For 0<<italic>k</italic><1, the function <italic>E</italic> <sub>2</sub>(<italic>Y</italic>) is strictly increasing on the interval (0, 1) and its values change from 2<italic>k</italic>/(<italic>k</italic>−1) to 1 when <italic>Y</italic> increases. When <italic>k</italic>>1, let <inline-formula><inline-graphic xlink:href="tjbd-8-161-m095.jpg"/></inline-formula> be the unique zero of <italic>g</italic> <sub>1</sub>(<italic>Y</italic>). Then one has for <inline-formula><inline-graphic xlink:href="tjbd-8-161-m096.jpg"/></inline-formula> that <inline-formula><inline-graphic xlink:href="tjbd-8-161-m097.jpg"/></inline-formula>. Now, we treat the dynamics on the centre manifold of the fixed point (<italic>x</italic>, <italic>y</italic>) in the case <italic>r</italic>=<italic>r</italic> <sub>1</sub> with <inline-formula><inline-graphic xlink:href="tjbd-8-161-m098.jpg"/></inline-formula>. To do this, we have to place the fixed point at the origin. Let <disp-formula id="UM0045"><graphic xlink:href="tjbd-8-161-u045.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Then according to Equation (17) and using Equation (37), we get the following <inline-formula><inline-graphic xlink:href="tjbd-8-161-m099.jpg"/></inline-formula> two-dimensional map defined by <disp-formula id="UM0046"><graphic xlink:href="tjbd-8-161-u046.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>For <inline-formula><inline-graphic xlink:href="tjbd-8-161-m100.jpg"/></inline-formula> we have the eigenvalues <inline-formula><inline-graphic xlink:href="tjbd-8-161-m101.jpg"/></inline-formula> with eigenvectors <disp-formula id="UM0047"><graphic xlink:href="tjbd-8-161-u047.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Introducing the change of variables <disp-formula id="UM0048"><graphic xlink:href="tjbd-8-161-u048.jpg" position="float" orientation="portrait"/></disp-formula> and applying to the previous map, we obtain the following two-dimensional map defined by <disp-formula-group id="M0050"><disp-formula><graphic xlink:href="tjbd-8-161-e050.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> with <disp-formula id="UM0049"><graphic xlink:href="tjbd-8-161-u049.jpg" position="float" orientation="portrait"/></disp-formula> where the dots denote higher order terms, and with <disp-formula id="UM0050"><graphic xlink:href="tjbd-8-161-u050.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Here, we do not give the explicit expressions for the other coefficients of Equation (50), because they are out of use. In fact, the centre manifold will be given as a graph over <italic>u</italic> and <inline-formula><inline-graphic xlink:href="tjbd-8-161-m102.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-161-m103.jpg"/></inline-formula>, and be at least <italic>O</italic>(2). Thus, by computing the centre manifold as <disp-formula id="UM0051"><graphic xlink:href="tjbd-8-161-u051.jpg" position="float" orientation="portrait"/></disp-formula> we can immediately see that the reduced map <inline-formula><inline-graphic xlink:href="tjbd-8-161-m104.jpg"/></inline-formula> is given by <disp-formula-group id="M0051"><disp-formula><graphic xlink:href="tjbd-8-161-e051.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>A direct calculation yields that <disp-formula id="UM0052"><graphic xlink:href="tjbd-8-161-u052.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Therefore, the one-dimensional map <inline-formula><inline-graphic xlink:href="tjbd-8-161-m105.jpg"/></inline-formula> defined by Equation (51) undergoes a period-doubling bifurcation at <inline-formula><inline-graphic xlink:href="tjbd-8-161-m106.jpg"/></inline-formula> provided that <disp-formula-group id="M0052"><disp-formula><graphic xlink:href="tjbd-8-161-e052.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>By an easy computation, we can find that <disp-formula id="UM0053"><graphic xlink:href="tjbd-8-161-u053.jpg" position="float" orientation="portrait"/></disp-formula> </p><p> <italic>Remark 4.3</italic> Due to the complicated nature of the expressions (depending on the parameter) in Equations (50)–(52), we decided to choose some particular values for the parameters (<italic>a, k</italic>) and then follow the whole procedure made along this section. By this choice of parameters, we can determine the exact value of <inline-formula><inline-graphic xlink:href="tjbd-8-161-m107.jpg"/></inline-formula>. By following computations presented in this section, we can compute the value of <italic>F</italic> given by Equation (52). For this reason, <xref rid="T0002" ref-type="table">Table 2</xref> is constructed by using numerical computations. <table-wrap id="T0002" orientation="portrait" position="float"><label>Table 2. </label><caption><title> The numerical exact values for (<italic>r</italic> <sub>1</sub>, λ<sub>2</sub>, <italic>F</italic>) corresponding to the chosen parameter values for (<italic>a, k</italic>).</title></caption><!--OASIS TABLE HERE--><table frame="hsides" rules="groups"><colgroup><col width="1"/><col width="1"/><col width="1"/><col width="1"/><col width="1"/><col width="1"/><col width="1"/><col width="1"/><col width="1"/></colgroup><thead valign="bottom"><tr><th align="left"><italic>k</italic></th><th align="center"><italic>a</italic></th><th align="center"><italic>r</italic><sub>1</sub></th><th align="center">λ<sub>2</sub></th><th align="center"><italic>F</italic><sub>1200</sub></th><th align="center"><italic>F</italic><sub>1300</sub></th><th align="center"><italic>F</italic><sub>1110</sub></th><th align="center"><italic>F</italic><sub>1101</sub></th><th align="center"><italic>F</italic></th></tr></thead><tbody><tr><td align="left">5</td><td align="center">0.5</td><td align="char" char=".">12.0238</td><td align="char" char=".">−2.62186</td><td align="char" char=".">6.92079</td><td align="char" char=".">−7.42265</td><td align="char" char=".">0.191821</td><td align="char" char=".">7.41292</td><td align="char" char=".">6.30868</td></tr><tr><td align="left">0.5</td><td align="center">10</td><td align="char" char=".">5.14677</td><td align="char" char=".">−1.34479</td><td align="char" char=".">4.77887</td><td align="char" char=".">−7.38963</td><td align="char" char=".">1.27529</td><td align="char" char=".">5.00919</td><td align="char" char=".">9.72169</td></tr><tr><td align="left">0.75</td><td align="center">5</td><td align="char" char=".">6.4168</td><td align="char" char=".">−1.58101</td><td align="char" char=".">6.18323</td><td align="char" char=".">−3.98477</td><td align="char" char=".">0.708933</td><td align="char" char=".">7.51087</td><td align="char" char=".">15.4186</td></tr><tr><td align="left">0.75</td><td align="center">40</td><td align="char" char=".">14</td><td align="char" char=".">−5.99998</td><td align="char" char=".">0.00001649</td><td align="char" char=".">−0.725698</td><td align="char" char=".">0.0285716</td><td align="char" char=".">−13.9999</td><td align="char" char=".">−0.020734</td></tr><tr><td align="left">0.9</td><td align="center">4</td><td align="char" char=".">7.40901</td><td align="char" char=".">−1.8899</td><td align="char" char=".">5.1228</td><td align="char" char=".">−2.06433</td><td align="char" char=".">0.419045</td><td align="char" char=".">4.88231</td><td align="char" char=".">7.9262</td></tr><tr><td align="left">0.9</td><td align="center">30</td><td align="char" char=".">37.9965</td><td align="char" char=".">−17.9981</td><td align="char" char=".">0.00017345</td><td align="char" char=".">−0.172959</td><td align="char" char=".">0.0030966</td><td align="char" char=".">−37.9958</td><td align="char" char=".">−0.0005355</td></tr><tr><td align="left">1</td><td align="center">3</td><td align="char" char=".">6.52291</td><td align="char" char=".">−1.30523</td><td align="char" char=".">15.1958</td><td align="char" char=".">−22.1696</td><td align="char" char=".">1.5091</td><td align="char" char=".">26.0496</td><td align="char" char=".">99.6364</td></tr><tr><td align="left">1</td><td align="center">13</td><td align="char" char=".">355.97</td><td align="char" char=".">−176.11</td><td align="char" char=".">0.0842284</td><td align="char" char=".">0.392571</td><td align="char" char=".">0.0000324</td><td align="char" char=".">−352.337</td><td align="char" char=".">0.000013</td></tr><tr><td align="left">2</td><td align="center">1</td><td align="char" char=".">5.15839</td><td align="char" char=".">−0.306219</td><td align="char" char=".">−6.72308</td><td align="char" char=".">33.6391</td><td align="char" char=".">−1.01462</td><td align="char" char=".">−16.2874</td><td align="char" char=".">57.6695</td></tr><tr><td align="left">5</td><td align="center">0.4</td><td align="char" char=".">6.45939</td><td align="char" char=".">−0.619264</td><td align="char" char=".">−16.2076</td><td align="char" char=".">56.3775</td><td align="char" char=".">−1.55496</td><td align="char" char=".">−35.7933</td><td align="char" char=".">199.463</td></tr></tbody></table></table-wrap> </p></sec><sec id="S004-S2003"><label>4.3. </label><title>Conclusions</title><p>In this paper, we studied a discrete-time predator–prey model which was a generalized Beddington–Nicholson–Bailey model. It is also a generalization of the system studied by Hone, Irle and Thurura. We investigated the stability and bifurcations of a generalized Beddington host–parasitoid model. We were able to compute the normal form coefficients of the Neimark–Sacker bifurcation without having the explicit form of the positive fixed point. The coefficients were very long and involved so that we decided to give numerical results along with phase diagrams (<xref rid="T0001" ref-type="table">Tables 1</xref> and <xref rid="T0002" ref-type="table">2</xref> and <xref rid="F0008 F0009 F0010 F0011 F0012 F0013" ref-type="fig">Figures 8–13</xref>) to verify our findings. The presence of the Neimark–Sacker bifurcation was shown. The same process was done for the period-doubling bifurcation in standard way without knowing the exact value of the positive fixed point. For the positive fixed point, the stability region was obtained in (<italic>Y, r</italic>)-space. Moreover, the stability condition of the extinction fixed point (0, 0) and the exclusion fixed point (<italic>k</italic>, 0) was investigated in a standard way. Our system for values <italic>b</italic>=<italic>r</italic>=2 and <italic>a</italic>=1/<italic>k</italic> seems to have simple dynamics, but it needs further study to obtain the global dynamics. Future research will focus on the global stability of the system and on the study of the other types of bifurcations and dynamics phenomena. This includes a deeper discussion and a further study.</p></sec></sec>
Global asymptotic behaviour of positive solutions to a non-autonomous Nicholson's blowflies model with delays	<p>This paper addresses the global existence and global asymptotic behaviour of positive solutions to a non-autonomous Nicholson's blowflies model with delays. By using a novel approach, sufficient conditions are derived for the existence and global exponential convergence of positive solutions of the model without any restriction on uniform positiveness of the per capita dead rate. Numerical examples are provided to illustrate the effectiveness of the obtained results.</p>	<contrib contrib-type="author"><name><surname>Van Hien</surname><given-names>Le</given-names></name><xref ref-type="aff" rid="AF1"> <sup>a</sup> </xref><xref ref-type="corresp" rid="cor1"> <sup>*</sup> </xref></contrib><aff id="AF1"><label><sup>a</sup></label><institution><named-content content-type="department">Department of Mathematics</named-content>, <named-content content-type="institution-name">Hanoi National University of Education</named-content></institution>, 136 Xuan Thuy Road, <named-content content-type="city">Hanoi</named-content>, <country>Vietnam</country></aff>	Journal of Biological Dynamics	<sec sec-type="intro" id="S001"><label>1. </label><title>Introduction</title><p>Nicholson [<xref rid="CIT0014" ref-type="bibr">14</xref>] used the following delay differential equation: <disp-formula-group id="M0001"><disp-formula><graphic xlink:href="tjbd-8-135-e001.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <italic>a</italic>, δ, <italic>P</italic> and τ are positive constants, to model laboratory population of the Australian sheep-blowfly. Biologically, <italic>x</italic>(<italic>t</italic>) is the size of the population at time <italic>t, P</italic> is the maximum per capita daily egg production rate, 1/<italic>a</italic> is the size at which the population reproduces at its maximum rate, δ is the per capita daily adult mortality rate and τ is the generation time, or the time taken from birth to maturity. The dynamics of Equation (1) was later studied in [<xref rid="CIT0005" ref-type="bibr">5</xref>,<xref rid="CIT0015" ref-type="bibr">15</xref>], where this model was referred to as the Nicholson's blowflies equation.</p><p>The theory of the Nicholson's blowflies equation has made a remarkable progress in the past 40 years and attracted extensive attention from researchers (see, for example, [<xref rid="CIT0001" ref-type="bibr">1</xref>] and the references therein). Many important results on the qualitative properties of the model such as existence of positive solutions, positive periodic/almost periodic solutions, persistence, permanence, oscillation and stability for the classical Nicholson's model and its generalizations (in particular, to variable coefficients, time-varying delays and impulsive equations) have been established in the literature [<xref rid="CIT0002" ref-type="bibr">2–4</xref>,<xref rid="CIT0007" ref-type="bibr">7–13</xref>,<xref rid="CIT0016" ref-type="bibr">16–22</xref>].</p><p>However, it should be noted that, in most of the aforementioned works, the per capita daily adult mortality terms have been restricted to be uniformly positive in order to use the coincidence degree method [<xref rid="CIT0003" ref-type="bibr">3</xref>,<xref rid="CIT0018" ref-type="bibr">18</xref>,<xref rid="CIT0021" ref-type="bibr">21</xref>], fixed point theorems [<xref rid="CIT0007" ref-type="bibr">7</xref>,<xref rid="CIT0012" ref-type="bibr">12</xref>,<xref rid="CIT0013" ref-type="bibr">13</xref>] or comparison principles [<xref rid="CIT0009" ref-type="bibr">9–11</xref>,<xref rid="CIT0017" ref-type="bibr">17</xref>,<xref rid="CIT0019" ref-type="bibr">19</xref>,<xref rid="CIT0020" ref-type="bibr">20</xref>,<xref rid="CIT0022" ref-type="bibr">22</xref>]. Furthermore, it is difficult to study the global asymptotic behaviour of the Nicholson's blowflies model with variable coefficients and time-varying delays. So far, there has been no result in the literature considering the global existence and global exponential convergence to the zero equilibrium point of positive solutions of nonautonomous Nicholson's blowflies model without the assumption on the uniform positiveness of the per capita daily mortality term.</p><p>Motivated by the above discussions, in this paper, we first consider the problem of global existence of positive solutions for a non-autonomous Nicholson's blowflies model of the following form: <disp-formula-group id="M0002"><disp-formula><graphic xlink:href="tjbd-8-135-e002.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <italic>m</italic> is a given positive integer, <inline-formula><inline-graphic xlink:href="tjbd-8-135-m001.jpg"/></inline-formula>, are continuous functions on ℝ<sup>+</sup>, <inline-formula><inline-graphic xlink:href="tjbd-8-135-m002.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-135-m003.jpg"/></inline-formula> for all <inline-formula><inline-graphic xlink:href="tjbd-8-135-m004.jpg"/></inline-formula>. We then employ a novel proof to establish conditions for the global exponential convergence to the zero equilibrium point of model (2). It is worth noting that, the restriction on the uniform positiveness of α(<italic>t</italic>) (that means, there is a positive constant α<sup>−</sup> such that <inline-formula><inline-graphic xlink:href="tjbd-8-135-m005.jpg"/></inline-formula> for all <italic>t</italic>≥0) as well as the upper and the lower bounds of <inline-formula><inline-graphic xlink:href="tjbd-8-135-m006.jpg"/></inline-formula>, will be removed.</p><p>We assume that <inline-formula><inline-graphic xlink:href="tjbd-8-135-m007.jpg"/></inline-formula> and let <inline-formula><inline-graphic xlink:href="tjbd-8-135-m008.jpg"/></inline-formula>. Throughout this paper, let <inline-formula><inline-graphic xlink:href="tjbd-8-135-m009.jpg"/></inline-formula> be the set of nonnegative continuous functions with the usual <italic>supremum</italic> norm \|.\| and <inline-formula><inline-graphic xlink:href="tjbd-8-135-m010.jpg"/></inline-formula>. In the biological interpretation of model (2), only positive solutions are meaningful and admissible. Thus we consider only the admissible initial conditions for Equation (2) as follows: <disp-formula-group id="M0003"><disp-formula><graphic xlink:href="tjbd-8-135-e003.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <inline-formula><inline-graphic xlink:href="tjbd-8-135-m011.jpg"/></inline-formula> is defined as <inline-formula><inline-graphic xlink:href="tjbd-8-135-m012.jpg"/></inline-formula> for all <inline-formula><inline-graphic xlink:href="tjbd-8-135-m013.jpg"/></inline-formula>.</p><p>Note that, the function <inline-formula><inline-graphic xlink:href="tjbd-8-135-m014.jpg"/></inline-formula> defined as <disp-formula id="UM0001"><graphic xlink:href="tjbd-8-135-u001.jpg" position="float" orientation="portrait"/></disp-formula> is continuous and locally Lipschitz with respect to <inline-formula><inline-graphic xlink:href="tjbd-8-135-m015.jpg"/></inline-formula>. Thus, for each <inline-formula><inline-graphic xlink:href="tjbd-8-135-m016.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-135-m017.jpg"/></inline-formula>, there exists a unique locally solution <inline-formula><inline-graphic xlink:href="tjbd-8-135-m018.jpg"/></inline-formula> of Equations (2) and (3) (for more details, see [<xref rid="CIT0006" ref-type="bibr">6</xref>]). Let <inline-formula><inline-graphic xlink:href="tjbd-8-135-m019.jpg"/></inline-formula> be the right maximal interval of existence of <inline-formula><inline-graphic xlink:href="tjbd-8-135-m020.jpg"/></inline-formula>.</p></sec><sec id="S002"><label>2. </label><title>Main results</title><sec id="S002-S2001"><label>2.1. </label><title>Global existence of positive solutions</title><p>In this section we will prove the global existence of positive solutions of Equation (2) for admissible initial conditions (3).</p><statement id="E0001"><label>Theorem 2.1 </label><p>For any <inline-formula><inline-graphic xlink:href="tjbd-8-135-m021.jpg"/></inline-formula> <inline-formula><inline-graphic xlink:href="tjbd-8-135-m022.jpg"/></inline-formula> the solution <inline-formula><inline-graphic xlink:href="tjbd-8-135-m023.jpg"/></inline-formula> of Equation (2) satisfies <disp-formula id="UM0002"><graphic xlink:href="tjbd-8-135-u002.jpg" position="float" orientation="portrait"/></disp-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-135-m024.jpg"/></inline-formula>.</p></statement><p> <italic>Proof</italic> Let <inline-formula><inline-graphic xlink:href="tjbd-8-135-m025.jpg"/></inline-formula> be a solution of Equations (2) and (3). For convenience, let us denote <inline-formula><inline-graphic xlink:href="tjbd-8-135-m026.jpg"/></inline-formula> if it does not make any confusion. We will show that <disp-formula-group id="M0004"><disp-formula><graphic xlink:href="tjbd-8-135-e004.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>Suppose in contrary that Equation (4) does not hold. Then, there exists <inline-formula><inline-graphic xlink:href="tjbd-8-135-m027.jpg"/></inline-formula> such that <italic>x</italic>(<italic>t</italic> <sub></sub>)=0 and <italic>x</italic>(<italic>t</italic>)>0 for all <italic>t</italic>∈[<italic>t</italic> <sub>0</sub>, <italic>t</italic> <sub></sub>). Thus, <italic>x</italic>(<italic>t</italic>)≥0 for all <inline-formula><inline-graphic xlink:href="tjbd-8-135-m028.jpg"/></inline-formula>. Observing that <inline-formula><inline-graphic xlink:href="tjbd-8-135-m029.jpg"/></inline-formula> for all <inline-formula><inline-graphic xlink:href="tjbd-8-135-m030.jpg"/></inline-formula>, from Equation (2), we have <disp-formula id="UM0003"><graphic xlink:href="tjbd-8-135-u003.jpg" position="float" orientation="portrait"/></disp-formula> which yields <disp-formula-group id="M0005"><disp-formula><graphic xlink:href="tjbd-8-135-e005.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Let <inline-formula><inline-graphic xlink:href="tjbd-8-135-m031.jpg"/></inline-formula>, it follows from Equation (5) that, <inline-formula><inline-graphic xlink:href="tjbd-8-135-m032.jpg"/></inline-formula>, which contradicts with <inline-formula><inline-graphic xlink:href="tjbd-8-135-m033.jpg"/></inline-formula>. This shows that Equation (4) holds. Consequently, <inline-formula><inline-graphic xlink:href="tjbd-8-135-m034.jpg"/></inline-formula> for all <inline-formula><inline-graphic xlink:href="tjbd-8-135-m035.jpg"/></inline-formula>. Next, we will prove the global existence of <inline-formula><inline-graphic xlink:href="tjbd-8-135-m036.jpg"/></inline-formula>, that means <inline-formula><inline-graphic xlink:href="tjbd-8-135-m037.jpg"/></inline-formula>. Note that <inline-formula><inline-graphic xlink:href="tjbd-8-135-m038.jpg"/></inline-formula> for all <inline-formula><inline-graphic xlink:href="tjbd-8-135-m039.jpg"/></inline-formula>. Using the fact <inline-formula><inline-graphic xlink:href="tjbd-8-135-m040.jpg"/></inline-formula>, we have <disp-formula id="UM0004"><graphic xlink:href="tjbd-8-135-u004.jpg" position="float" orientation="portrait"/></disp-formula> Therefore <disp-formula-group id="M0006"><disp-formula><graphic xlink:href="tjbd-8-135-e006.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> It follows from Equation (6) <disp-formula-group id="M0007"><disp-formula><graphic xlink:href="tjbd-8-135-e007.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> for all <inline-formula><inline-graphic xlink:href="tjbd-8-135-m041.jpg"/></inline-formula>. Suppose in contrary that <inline-formula><inline-graphic xlink:href="tjbd-8-135-m042.jpg"/></inline-formula>, then <disp-formula-group id="M0008"><disp-formula><graphic xlink:href="tjbd-8-135-e008.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> On the other hand, from Equation (7), we have <disp-formula id="UM0005"><graphic xlink:href="tjbd-8-135-u005.jpg" position="float" orientation="portrait"/></disp-formula> which yields a contradiction with Equation (8). Therefore, <inline-formula><inline-graphic xlink:href="tjbd-8-135-m043.jpg"/></inline-formula>. The proof is completed.</p><p> <italic>Remark 1</italic> In the proof of Theorem 2.1, we do not require the upper and the lower boundedness of <inline-formula><inline-graphic xlink:href="tjbd-8-135-m044.jpg"/></inline-formula>, by positive constants as considered in many other results (see, e.g., [<xref rid="CIT0003" ref-type="bibr">3</xref>,<xref rid="CIT0004" ref-type="bibr">4</xref>,<xref rid="CIT0007" ref-type="bibr">7</xref>,<xref rid="CIT0009" ref-type="bibr">9–13</xref>,<xref rid="CIT0017" ref-type="bibr">17–22</xref>]). Furthermore, if <inline-formula><inline-graphic xlink:href="tjbd-8-135-m045.jpg"/></inline-formula> are assumed to be upper and lower bounded by positive constants then the boundedness of any solution <inline-formula><inline-graphic xlink:href="tjbd-8-135-m046.jpg"/></inline-formula> on <inline-formula><inline-graphic xlink:href="tjbd-8-135-m047.jpg"/></inline-formula> can be proved easily as follows.</p><p>For a given bounded continuous function <italic>g</italic>(<italic>t</italic>) on ℝ<sup>+</sup>, let us denote <disp-formula id="UM0006"><graphic xlink:href="tjbd-8-135-u006.jpg" position="float" orientation="portrait"/></disp-formula> From Equation (7), we obtain <disp-formula id="UM0007"><graphic xlink:href="tjbd-8-135-u007.jpg" position="float" orientation="portrait"/></disp-formula> This shows that <inline-formula><inline-graphic xlink:href="tjbd-8-135-m048.jpg"/></inline-formula> is bounded on <inline-formula><inline-graphic xlink:href="tjbd-8-135-m049.jpg"/></inline-formula> and thus <inline-formula><inline-graphic xlink:href="tjbd-8-135-m050.jpg"/></inline-formula> as concluded in Theorem 2.1. As shown in the following example, if α(<italic>t</italic>) is not uniformly positive then positive solutions of Equation (2) may not be bounded.</p><statement id="E0002"><label>ExampleConsider the following equation: </label><p> <disp-formula-group id="M0009"><disp-formula><graphic xlink:href="tjbd-8-135-e009.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>It can be seen that <inline-formula><inline-graphic xlink:href="tjbd-8-135-m051.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-135-m052.jpg"/></inline-formula> are continuous functions, <inline-formula><inline-graphic xlink:href="tjbd-8-135-m053.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-135-m054.jpg"/></inline-formula> for all <inline-formula><inline-graphic xlink:href="tjbd-8-135-m055.jpg"/></inline-formula> but α(<italic>t</italic>) and γ(<italic>t</italic>) are not uniformly positive. For any <inline-formula><inline-graphic xlink:href="tjbd-8-135-m056.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-135-m057.jpg"/></inline-formula>, by Theorem 2.1, Equation (9) has a unique positive solution <inline-formula><inline-graphic xlink:href="tjbd-8-135-m058.jpg"/></inline-formula> on [<italic>t</italic> <sub>0</sub>,+∞). For illustrative purpose, in the following numerical simulation, we take <inline-formula><inline-graphic xlink:href="tjbd-8-135-m059.jpg"/></inline-formula> for <inline-formula><inline-graphic xlink:href="tjbd-8-135-m060.jpg"/></inline-formula>. To visualize the effect of initial condition, we simulate the state trajectory from <italic>t</italic>=−1. It is shown in <xref rid="F0001" ref-type="fig">Figure 1</xref> that the corresponding solution <inline-formula><inline-graphic xlink:href="tjbd-8-135-m061.jpg"/></inline-formula> of Equation (9) is unbounded. <fig id="F0001" orientation="portrait" position="float"><label>Fig. 1. </label><caption><p>Unbounded state trajectory of (9) with <inline-formula><inline-graphic xlink:href="tjbd-8-135-m062.jpg"/></inline-formula>.</p></caption><graphic xlink:href="tjbd-8-135-g001"/></fig> </p></statement></sec><sec id="S002-S2002"><label>2.2. </label><title>Global exponential convergence to the zero equilibrium point</title><p>In this section, we will establish conditions for the global exponential convergence to the zero equilibrium point of positive solutions of model (2).</p><p>Let us consider the following assumptions: <list list-type="simple"><list-item><p>(A1) There exists <italic>m</italic>[α]>0 such that <inline-formula><inline-graphic xlink:href="tjbd-8-135-m063.jpg"/></inline-formula> </p></list-item><list-item><p>(A2) <inline-formula><inline-graphic xlink:href="tjbd-8-135-m064.jpg"/></inline-formula>.</p></list-item><list-item><p>(A3) <inline-formula><inline-graphic xlink:href="tjbd-8-135-m065.jpg"/></inline-formula> </p></list-item></list> </p><p> <italic>Remark 2</italic> If α(<italic>t</italic>) is assumed to be upper- and lower-bounded by positive constants then assumptions (A1) and (A2) are obviously removed.</p><statement id="E0003"><label>Proposition 2.3 </label><p>Let assumption (A3) holds. Then there exists a positive constant δ such that any solution <inline-formula><inline-graphic xlink:href="tjbd-8-135-m066.jpg"/></inline-formula> of Equation (2) satisfies <disp-formula-group id="M0010"><disp-formula><graphic xlink:href="tjbd-8-135-e010.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p></statement><p> <italic>Proof</italic> By (A3), there exists <italic>T</italic>>0 such that <disp-formula-group id="M0011"><disp-formula><graphic xlink:href="tjbd-8-135-e011.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>Let us define <disp-formula id="UM0008"><graphic xlink:href="tjbd-8-135-u008.jpg" position="float" orientation="portrait"/></disp-formula> then δ is a positive constant. Suppose <inline-formula><inline-graphic xlink:href="tjbd-8-135-m067.jpg"/></inline-formula> be a solution of Equation (2). Without loss of generality, we assume that <italic>t</italic> <sub>0</sub>≤<italic>T</italic>. From Equation (7), we have <disp-formula-group id="M0012"><disp-formula><graphic xlink:href="tjbd-8-135-e012.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> We will prove that Equation (12) holds for all <inline-formula><inline-graphic xlink:href="tjbd-8-135-m068.jpg"/></inline-formula>. For given ε>0, assume that there exists <inline-formula><inline-graphic xlink:href="tjbd-8-135-m069.jpg"/></inline-formula> such that <disp-formula id="UM0009"><graphic xlink:href="tjbd-8-135-u009.jpg" position="float" orientation="portrait"/></disp-formula> Then, for <inline-formula><inline-graphic xlink:href="tjbd-8-135-m070.jpg"/></inline-formula>, from Equation (2) we have <disp-formula id="UM0010"><graphic xlink:href="tjbd-8-135-u010.jpg" position="float" orientation="portrait"/></disp-formula> Therefore, <disp-formula-group id="M0013"><disp-formula><graphic xlink:href="tjbd-8-135-e013.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Let <inline-formula><inline-graphic xlink:href="tjbd-8-135-m071.jpg"/></inline-formula>, from Equation (13) we obtain <disp-formula id="UM0011"><graphic xlink:href="tjbd-8-135-u011.jpg" position="float" orientation="portrait"/></disp-formula> which yields a contradiction. Thus, <disp-formula id="UM0012"><graphic xlink:href="tjbd-8-135-u012.jpg" position="float" orientation="portrait"/></disp-formula> Let <inline-formula><inline-graphic xlink:href="tjbd-8-135-m072.jpg"/></inline-formula> we finally obtain <disp-formula id="UM0013"><graphic xlink:href="tjbd-8-135-u013.jpg" position="float" orientation="portrait"/></disp-formula> This completes the proof.</p><p> <italic>Remark 3</italic> It can be seen from the proof of Proposition 2.3 that, if Equation (11) holds for all <inline-formula><inline-graphic xlink:href="tjbd-8-135-m073.jpg"/></inline-formula> then every solution <inline-formula><inline-graphic xlink:href="tjbd-8-135-m074.jpg"/></inline-formula> of Equation (2) satisfies <inline-formula><inline-graphic xlink:href="tjbd-8-135-m075.jpg"/></inline-formula> for all <italic>t</italic>≥<italic>t</italic> <sub>0</sub>.</p><p>We now prove the global exponential convergence to the zero equilibrium point of positive solutions of Equation (2) as given in the following theorem.</p><statement id="E0004"><label>Theorem 2.4 </label><p>Under assumptions (A1)–(A3), all positive solutions of Equation (2) converge exponentially to the zero equilibrium point of Equation (2). More precisely, there exist positive constants <inline-formula><inline-graphic xlink:href="tjbd-8-135-m076.jpg"/></inline-formula> such that every solution <inline-formula><inline-graphic xlink:href="tjbd-8-135-m077.jpg"/></inline-formula> of Equations (2) and (3), with <inline-formula><inline-graphic xlink:href="tjbd-8-135-m078.jpg"/></inline-formula> satisfies <disp-formula-group id="M0014"><disp-formula><graphic xlink:href="tjbd-8-135-e014.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p></statement><p> <italic>Remark 4</italic> It is worth noting that, estimation of (14) and (A2) guarantee the global exponential convergence to zero of all positive solutions of (2) which we will refer to <italic>generalized exponential convergence</italic>.</p><p> <italic>Proof</italic> Let <inline-formula><inline-graphic xlink:href="tjbd-8-135-m079.jpg"/></inline-formula> be a solution of Equations (2) and (3). By Proposition 2.3, there exists a constant δ>0 such that <disp-formula id="UM0014"><graphic xlink:href="tjbd-8-135-u014.jpg" position="float" orientation="portrait"/></disp-formula> Also, by (A3), <disp-formula id="UM0015"><graphic xlink:href="tjbd-8-135-u015.jpg" position="float" orientation="portrait"/></disp-formula> and hence, <inline-formula><inline-graphic xlink:href="tjbd-8-135-m080.jpg"/></inline-formula>. Therefore, there exists <italic>T</italic> <sub></sub>≥<italic>T</italic> (defined in Equation (11)) such that <disp-formula-group id="M0015"><disp-formula><graphic xlink:href="tjbd-8-135-e015.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Furthermore, we can assume that <inline-formula><inline-graphic xlink:href="tjbd-8-135-m081.jpg"/></inline-formula> for all <inline-formula><inline-graphic xlink:href="tjbd-8-135-m082.jpg"/></inline-formula>. Then, by (A1), we have <disp-formula id="UM0016"><graphic xlink:href="tjbd-8-135-u016.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Now, we consider the following scalar equation: <disp-formula-group id="M0016"><disp-formula><graphic xlink:href="tjbd-8-135-e016.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Note that, <italic>H</italic>(λ) is a continuous function, <inline-formula><inline-graphic xlink:href="tjbd-8-135-m083.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-135-m084.jpg"/></inline-formula> as λ tends to infinity and its derivative <inline-formula><inline-graphic xlink:href="tjbd-8-135-m085.jpg"/></inline-formula> for all <inline-formula><inline-graphic xlink:href="tjbd-8-135-m086.jpg"/></inline-formula>. Therefore, Equation (16) has a unique positive solution λ<sub></sub>. Moreover, <inline-formula><inline-graphic xlink:href="tjbd-8-135-m087.jpg"/></inline-formula> for all <inline-formula><inline-graphic xlink:href="tjbd-8-135-m088.jpg"/></inline-formula>. Then, it can be verified that <disp-formula id="UM0017"><graphic xlink:href="tjbd-8-135-u017.jpg" position="float" orientation="portrait"/></disp-formula> for all <italic>t</italic>≥<italic>T</italic> <sub></sub>, <inline-formula><inline-graphic xlink:href="tjbd-8-135-m089.jpg"/></inline-formula>, from which we obtain <disp-formula-group id="M0017"><disp-formula><graphic xlink:href="tjbd-8-135-e017.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Let us consider the following function: <disp-formula id="UM0018"><graphic xlink:href="tjbd-8-135-u018.jpg" position="float" orientation="portrait"/></disp-formula> where <inline-formula><inline-graphic xlink:href="tjbd-8-135-m090.jpg"/></inline-formula>. Note that <disp-formula id="UM0019"><graphic xlink:href="tjbd-8-135-u019.jpg" position="float" orientation="portrait"/></disp-formula> and thus, by Equation (17), <disp-formula-group id="M0018"><disp-formula><graphic xlink:href="tjbd-8-135-e018.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> We will show that <disp-formula-group id="M0019"><disp-formula><graphic xlink:href="tjbd-8-135-e019.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> For given ε>0, note that <inline-formula><inline-graphic xlink:href="tjbd-8-135-m091.jpg"/></inline-formula> for all <italic>t</italic>∈[<italic>t</italic> <sub>0</sub>, <italic>T</italic> <sub></sub>], we have <inline-formula><inline-graphic xlink:href="tjbd-8-135-m092.jpg"/></inline-formula>. Assume that there exists a <italic>t˜</italic>><italic>T</italic> <sub>*</sub> satisfying <inline-formula><inline-graphic xlink:href="tjbd-8-135-m093.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-135-m094.jpg"/></inline-formula> for all <inline-formula><inline-graphic xlink:href="tjbd-8-135-m095.jpg"/></inline-formula>. Then, <inline-formula><inline-graphic xlink:href="tjbd-8-135-m096.jpg"/></inline-formula>. From Equations (2) and (18), we have <disp-formula id="UM0020"><graphic xlink:href="tjbd-8-135-u020.jpg" position="float" orientation="portrait"/></disp-formula> Taking integral on both sides of the above inequality we obtain <disp-formula-group id="M0020"><disp-formula><graphic xlink:href="tjbd-8-135-e020.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Let <inline-formula><inline-graphic xlink:href="tjbd-8-135-m097.jpg"/></inline-formula> we obtain <inline-formula><inline-graphic xlink:href="tjbd-8-135-m098.jpg"/></inline-formula>, which yields a contradiction. This shows that <inline-formula><inline-graphic xlink:href="tjbd-8-135-m099.jpg"/></inline-formula> for all <italic>t</italic>≥<italic>t</italic> <sub>0</sub>. Consequently, Equation (19) holds for all <italic>t</italic>≥<italic>t</italic> <sub>0</sub>, and thus, Equation (14) holds for any <inline-formula><inline-graphic xlink:href="tjbd-8-135-m100.jpg"/></inline-formula>. The proof is now completed.</p><p>As a special case, if α(<italic>t</italic>) is upper- and lower-bounded by positive constants as considered in many other works in the literature (e.g., [<xref rid="CIT0003" ref-type="bibr">3</xref>,<xref rid="CIT0011" ref-type="bibr">11</xref>,<xref rid="CIT0012" ref-type="bibr">12</xref>,<xref rid="CIT0017" ref-type="bibr">17</xref>,<xref rid="CIT0020" ref-type="bibr">20</xref>,<xref rid="CIT0022" ref-type="bibr">22</xref>]) then we obtain the following corollary.</p><statement id="E0005"><label>Corollary 2.5 </label><p>Assume that (A3) holds and there exist constants <inline-formula><inline-graphic xlink:href="tjbd-8-135-m101.jpg"/></inline-formula> such that <inline-formula><inline-graphic xlink:href="tjbd-8-135-m102.jpg"/></inline-formula>. Then every solution <inline-formula><inline-graphic xlink:href="tjbd-8-135-m103.jpg"/></inline-formula> of Equations (2) and (3) satisfies <disp-formula id="UM0021"><graphic xlink:href="tjbd-8-135-u021.jpg" position="float" orientation="portrait"/></disp-formula> where <inline-formula><inline-graphic xlink:href="tjbd-8-135-m104.jpg"/></inline-formula> are defined as in the proof of Theorem 2.1 and η is the unique positive solution of the following scalar equation: <disp-formula id="UM0022"><graphic xlink:href="tjbd-8-135-u022.jpg" position="float" orientation="portrait"/></disp-formula> </p></statement></sec><sec id="S002-S2003"><label>2.3. </label><title>An example</title><p>In this section, we give a numerical example to illustrate the effectiveness of our results.</p><statement id="E0006"><label>Example 2.6 </label><p>Consider the following Nicholson's blowflies model with time-varying delay <disp-formula-group id="M0021"><disp-formula><graphic xlink:href="tjbd-8-135-e021.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula id="UM0023"><graphic xlink:href="tjbd-8-135-u023.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>It should be noted that, for this model, the obtained results in the literature cannot be applied to conclude the convergence of positive solutions of Equation (21). In this case we have <disp-formula id="UM0024"><graphic xlink:href="tjbd-8-135-u024.jpg" position="float" orientation="portrait"/></disp-formula> Therefore, assumptions (A1)–(A3) hold. The scalar equation <inline-formula><inline-graphic xlink:href="tjbd-8-135-m105.jpg"/></inline-formula> has a unique positive solution <inline-formula><inline-graphic xlink:href="tjbd-8-135-m106.jpg"/></inline-formula>. By Remark 3 and Theorem 2.4, every solution <inline-formula><inline-graphic xlink:href="tjbd-8-135-m107.jpg"/></inline-formula> of Equation (21) with <inline-formula><inline-graphic xlink:href="tjbd-8-135-m108.jpg"/></inline-formula>, satisfies <disp-formula id="UM0025"><graphic xlink:href="tjbd-8-135-u025.jpg" position="float" orientation="portrait"/></disp-formula> where <inline-formula><inline-graphic xlink:href="tjbd-8-135-m109.jpg"/></inline-formula>. Furthermore, it can be seen that, there is no positive constant η such that <inline-formula><inline-graphic xlink:href="tjbd-8-135-m110.jpg"/></inline-formula>, for all <italic>t</italic>≥0. And thus, a classical exponential estimation, that is, <inline-formula><inline-graphic xlink:href="tjbd-8-135-m111.jpg"/></inline-formula>, does not exist. Therefore, the exponential estimation proposed in this paper is less conservative and is expected to relax conditions for the exponential convergence of the model. In the following simulation, we take initial function <inline-formula><inline-graphic xlink:href="tjbd-8-135-m112.jpg"/></inline-formula>. As shown in <xref rid="F0002" ref-type="fig">Figure 2</xref>, the corresponding state trajectory of Equation (21) satisfies a generalized exponential estimation <inline-formula><inline-graphic xlink:href="tjbd-8-135-m113.jpg"/></inline-formula>. Furthermore, a classical exponential estimation does not exist. For illustrative purpose, we take η=0.2 then it can be seen in <xref rid="F0002" ref-type="fig">Figure 2</xref> that <inline-formula><inline-graphic xlink:href="tjbd-8-135-m114.jpg"/></inline-formula> for all <italic>t</italic>≥20. <fig id="F0002" orientation="portrait" position="float"><label>Fig. 2. </label><caption><p>A state trajectory of Equation (21) with <inline-formula><inline-graphic xlink:href="tjbd-8-135-m115.jpg"/></inline-formula>.</p></caption><graphic xlink:href="tjbd-8-135-g002"/></fig> </p></statement></sec></sec><sec sec-type="conclusions" id="S003"><label>3. </label><title>Conclusion</title><p>This paper has dealt with the global existence and global asymptotic behaviour of positive solutions to a non-autonomous Nicholson's blowflies model with time-varying delays. By using a new approach, we have derived sufficient conditions for the global generalized exponential convergence of positive solutions of the model without any restriction on the uniform positiveness of the per capita dead rate term. Numerical examples have been provided to illustrate the effectiveness of the obtained results.</p></sec><sec id="S004"><title>Funding</title><p>This work was supported by the Ministry of Education and Training of Vietnam (B2013.17.42).</p></sec>
A study of tumour growth based on stoichiometric principles: a continuous model and its discrete analogue	<p>In this paper, we consider a continuous mathematically tractable model and its discrete analogue for the tumour growth. The model formulation is based on stoichiometric principles considering tumour-immune cell interactions in potassium (K <sup>+</sup>)-limited environment. Our both continuous and discrete models illustrate ‘cancer immunoediting’ as a dynamic process having all three phases namely elimination, equilibrium and escape. The stoichiometric principles introduced into the model allow us to study its dynamics with the variation in the total potassium in the surrounding of the tumour region. It is found that an increase in the total potassium may help the patient fight the disease for a longer period of time. This result seems to be in line with the protective role of the potassium against the risk of pancreatic cancer as has been reported by Bravi <italic>et al</italic>. [<italic>Dietary intake of selected micronutrients and risk of pancreatic cancer: An Italian case-control study</italic>, Ann. Oncol. 22 (2011), pp. 202–206].</p>	<contrib contrib-type="author"><name><surname>Saleem</surname><given-names>M.</given-names></name><xref ref-type="aff" rid="AF1"> <sup>a</sup> </xref><xref ref-type="corresp" rid="cor1"> <sup>*</sup> </xref></contrib><contrib contrib-type="author"><name><surname>Agrawal</surname><given-names>Tanuja</given-names></name><xref ref-type="aff" rid="AF1"> <sup>a</sup> </xref><xref rid="AN1" ref-type="author-notes"/></contrib><contrib contrib-type="author"><name><surname>Anees</surname><given-names>Afzal</given-names></name><xref ref-type="aff" rid="AF2"> <sup>b</sup> </xref><xref rid="AN2" ref-type="author-notes"/></contrib><aff id="AF1"><label><sup>a</sup></label><institution><named-content content-type="department">Department of Applied Mathematics</named-content>, <named-content content-type="institution-name">Z.H. College of Engineering and Technology, A.M.U</named-content></institution>., <named-content content-type="city">Aligarh</named-content><named-content content-type="postal-code">202002</named-content>, <country>India</country></aff><aff id="AF2"><label><sup>b</sup></label><institution><named-content content-type="department">Department of Surgery</named-content>, <named-content content-type="institution-name">J.N. Medical College, A.M.U</named-content></institution>., <named-content content-type="city">Aligarh</named-content><named-content content-type="postal-code">202002</named-content>, <country>India</country></aff>	Journal of Biological Dynamics	<sec sec-type="intro" id="S001"><label>1. </label><title>Introduction</title><p>The evidence accumulated in the past decade indicates that the immune system can recognize and eliminate malignant tumours in a process termed ‘cancer immunosurveillance’ [<xref rid="CIT0007" ref-type="bibr">7</xref>,<xref rid="CIT0022" ref-type="bibr">22</xref>,<xref rid="CIT0034" ref-type="bibr">34</xref>]. The work from many laboratories has validated the concept of cancer immunosurveillance, demonstrating that immune system can indeed protect mice from outgrowth of tumours [<xref rid="CIT0011" ref-type="bibr">11</xref>,<xref rid="CIT0015" ref-type="bibr">15</xref>,<xref rid="CIT0030" ref-type="bibr">30</xref>,<xref rid="CIT0032" ref-type="bibr">32</xref>,<xref rid="CIT0033" ref-type="bibr">33</xref>,<xref rid="CIT0039" ref-type="bibr">39</xref>]. The current view on the host immune system is conflicted by evidence for antitumour effects as well as evidence for tumour-favouring actions [<xref rid="CIT0039" ref-type="bibr">39–41</xref>]. ‘Cancer immunoediting’ is the term used in the literature for this dual action of the host immune system [<xref rid="CIT0039" ref-type="bibr">39</xref>].</p><p>The immune system mostly consists of white blood cells, especially T lymphocytes (usually CD8<sup>+</sup> and CD4<sup>+</sup> T cells along with their characteristically produced cytokine IFN-γ), natural killer cells and macrophages. The lymphocytes are used to detect any foreign or non-self cells in the body known as antigens. It has been shown experimentally that these immune cells can lyse tumour cells very effectively [<xref rid="CIT0007" ref-type="bibr">7</xref>,<xref rid="CIT0026" ref-type="bibr">26</xref>]. Tumour associated macrophages or myeloid-derived suppressive cells, CD4<sup>+</sup> Foxp3<sup>+</sup> Treg cells and Th17 cells and their associated cytokines Il-6, TNF, IL-1β, IL-23 and TGF-β are generally recognized as dominant tumour-promoting forces [<xref rid="CIT0041" ref-type="bibr">41</xref>].</p><p>Evidence indicates that a healthy immune system is necessary for control of malignant disease. One of the common factors that have been associated with pronounced abnormalities in the immune system is poor nutrition [<xref rid="CIT0023" ref-type="bibr">23</xref>,<xref rid="CIT0025" ref-type="bibr">25</xref>]. The convincing evidence exists that individuals who have been on immune-suppressive medications for longer periods of time, or have autoimmune disease or chronic infection (such as AIDS) are particularly at risk of malignancy [<xref rid="CIT0024" ref-type="bibr">24</xref>,<xref rid="CIT0025" ref-type="bibr">25</xref>]. The chronic inflammation, previous viral infections such as EBV, Hepatitis B and C, herpes virus or HIV are also significant factors that hamper proper functioning of immune system and lead to development of various cancers [<xref rid="CIT0012" ref-type="bibr">12</xref>,<xref rid="CIT0024" ref-type="bibr">24</xref>,<xref rid="CIT0027" ref-type="bibr">27</xref>,<xref rid="CIT0031" ref-type="bibr">31</xref>]. Failure of intact immune responses, such as immunosurveillance or immunoediting, has also been associated with evasion or immune-suppression activities of cancer [<xref rid="CIT0028" ref-type="bibr">28</xref>,<xref rid="CIT0029" ref-type="bibr">29</xref>]. Tumours often avoid detection by Killer T cells by having a reduced number of MHC class I molecules on their surface [<xref rid="CIT0010" ref-type="bibr">10</xref>]. Some tumours release products, such as cytokine TGF-β, which suppress the activity of macrophages and lymphocytes [<xref rid="CIT0010" ref-type="bibr">10</xref>].</p><p>In a recently reported study, Bravi <italic>et al</italic>. [<xref rid="CIT0002" ref-type="bibr">2</xref>] considered the role of 15 selected vitamins and carotenoids and 6 minerals including potassium in the protection against pancreatic cancer. Analysing separate role of different minerals, they found significant inverse trends in pancreatic cancer risk for increasing intake of potassium. Prior to Bravi <italic>et al</italic>. ’s [<xref rid="CIT0002" ref-type="bibr">2</xref>] work, two studies of Jansson [<xref rid="CIT0013" ref-type="bibr">13</xref>,<xref rid="CIT0014" ref-type="bibr">14</xref>] regarding the colorectal cancer risk in the USA also hinted at the possible role of electrolytes sodium and potassium in cancer etiology. In its simpler form Jansson studies attempted to suggest that intracellular and dietary potassium (K <sup>+</sup>) protects against cancer and intracellular and dietary sodium (Na) increases the risk of cancer. It can be mentioned here that all studies such as mentioned above have been of the suggestive nature for the positive role of electrolyte potassium against the risk of cancer. Unfortunately there are neither direct cell-biology-based investigations nor biological data available in the literature that support or negate the role of potassium against cancer risk.</p><p>We focus in this paper on the suggestive prediction of the above studies especially the work of Bravi <italic>et al</italic>. [<xref rid="CIT0002" ref-type="bibr">2</xref>] regarding the positive role of potassium against the cancer risk. To this end, we formulate a tractable mathematical model based on the principles of stoichiometry to represent interactions of cancer and immune cells in (K <sup>+</sup>)-limited environment using the structure of the Kuang–Huisman–Elser (KHE) model [<xref rid="CIT0016" ref-type="bibr">16</xref>]. We modify this model and incorporate in it the coercing of the surrounding immune cells by cancer cells and a medical treatment strategy that may help add immune cells in the body by a constant rate.</p><p>We structure the paper as follows. The continuous model formulation is given in Section 2. The results for boundedness of the continuous model solution and local stability of its equilibriums are given in Section 3. The discrete analogue of the continuous model and the local stability results of its equilibriums are discussed in Section 4. Numerically drawn bifurcation diagrams illustrating the positive role of potassium against the cancer are given in Section 5. This section also shows the possibility of chaotic dynamics in the discrete model. Section 6 contains discussion and conclusions.</p></sec><sec id="S002"><label>2. </label><title>Model formulation based on stoichiometric principles</title><p>For the theory of ecological stoichiometry, one may consult the masterpiece from Sterner and Elser [<xref rid="CIT0035" ref-type="bibr">35</xref>]. According to Sui <italic>et al</italic>. [<xref rid="CIT0037" ref-type="bibr">37</xref>], ‘ecological stoichiometry is the study of the balance of energy and multiple chemical resources (usually elements) in ecological interactions’. In this section, we formulate a mathematically tractable model that specifically deals with the dynamics of cancer-immune cell interactions in closed potassium (K <sup>+</sup>)-limited environment. The above definition of ecological stoichiometry when applied to our problem can be restated as ‘cellular stoichiometry is the study of the balance of energy and multiple chemical resources (such as carbon (C) and potassium (K <sup>+</sup>)) in cancer-immune cell interactions’. The greatest advantage of applying stoichiometry principles to our problem is that it allows us to employ the variability in the K <sup>+</sup> content of the cancer cell using the Droop equation for its growth.</p><p>The concept of immune-surveillance hypothesis that immune system is capable of inhibiting the growth of very small tumours and eliminating them before they become clinically evident motivates the derivation of our mathematical model of the interactions between tumour cells and immune cells. Our immune cell population may represent any of the cytotoxic immune cells (also called effector cells) such as CD8<sup>+</sup> or CD4<sup>+</sup> T Cells [<xref rid="CIT0001" ref-type="bibr">1</xref>] of the adaptive immune system. Since adaptive immune system's effector cells proliferate in response to antigenic stimulation and kill the tumour cells, we assume that our effector cell population interacts with tumour cells in a predator–prey relationship [<xref rid="CIT0009" ref-type="bibr">9</xref>,<xref rid="CIT0020" ref-type="bibr">20</xref>] where immune cells play the role of the predator and the tumour cells that of prey. Without stoichiometric considerations, such a model, in a general setting, can be expressed as <disp-formula-group id="M0001"><disp-formula><graphic xlink:href="tjbd-8-117-e001.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0002"><disp-formula><graphic xlink:href="tjbd-8-117-e002.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <italic>x</italic> represents the size or density of the tumour cell population and <italic>y</italic> that of the effector cell population. The parameter <italic>b</italic> is the intrinsic growth rate of the tumour cell; parameter <italic>L</italic> represents the carrying capacity of the tumour cells; the negative term, −<italic>f</italic>(<italic>x</italic>)<italic>y</italic> represents the rate of killing of tumour cells by the immune cells with <italic>f</italic>(<italic>x</italic>) denoting the functional response of the immune cells; the positive term, <italic>ef</italic>(<italic>x</italic>)<italic>y</italic> with <italic>e</italic> being a positive parameter denotes the addition of immune cells in the system being activated by the presence of tumour cells; the negative term, −<italic>lxy</italic> represents the coercion of the immune cells by the tumour cells to their advantage; and the negative term, −<italic>dy</italic> depicts the apoptosis of the immune cells. The time dependent function <italic>g</italic>(<italic>t</italic>) represents the treatment term or the external antitumour activity. When <italic>g</italic>(<italic>t</italic>)=<italic>u</italic> (constant), the term describes continuous production of immune cells, even in the absence of cancer cells. It may be noted that we have mentioned model (1) here just for the purpose of its comparison with the model to be formulated in the following based on stoichiometric principles.</p><p>While incorporating stoichiometric reality into model (1), we concentrate on two important substances, carbon and potassium. We assume that all other substances required for proliferation of both tumour and immune cells are abundant in the system. Since the bulk of dry weight of most organisms is carbon, we express biomass of populations in carbon terms. Our model formulation approach in introducing stoichiometric considerations in model (1) follows same steps as for the KHE model in [<xref rid="CIT0016" ref-type="bibr">16</xref>,<xref rid="CIT0037" ref-type="bibr">37</xref>] but for suitably modifying the main assumptions of the KHE model to suit our requirements. We begin with the following assumptions: <list list-type="simple"><list-item><p>A0. All cells are assumed to be made of carbon (C)</p></list-item><list-item><p>The total mass of the potassium <italic>K</italic> <sub>t</sub> in the entire system is fixed; i.e. the system is closed for potassium.</p></list-item><list-item><p>Stoichiometry of the immune cells is relatively stable compared with stoichiometry of the tumour cells. Thus it is assumed, that the potassium to carbon ratio <inline-formula><inline-graphic xlink:href="tjbd-8-117-m001.jpg"/></inline-formula> in the tumour cells varies, but it never falls below a minimum <inline-formula><inline-graphic xlink:href="tjbd-8-117-m002.jpg"/></inline-formula>; the immune cells maintain a constant <inline-formula><inline-graphic xlink:href="tjbd-8-117-m003.jpg"/></inline-formula> ratio, denoted by <inline-formula><inline-graphic xlink:href="tjbd-8-117-m004.jpg"/></inline-formula>.</p></list-item><list-item><p>All potassium in the system is divided into three pools: potassium in the tumour cells, potassium in the immune cells, and free potassium in the blood stream in the surrounding.</p></list-item></list> </p><p>If <italic>K</italic> <sub>c</sub>, <italic>K</italic> <sub><italic>i</italic></sub><0 and <italic>K</italic> <sub>f</sub> denote the potassium in cancer cells, potassium in immune cells, and the free potassium in the blood stream in the surrounding, then <inline-formula><inline-graphic xlink:href="tjbd-8-117-m005.jpg"/></inline-formula>. Let <italic>Q</italic>=<italic>Q</italic>(<italic>t</italic>) be the tumour cell's quota for K <sup>+</sup>, then <italic>K</italic> <sub>c</sub>=<italic>Qx, K</italic> <sub><italic>i</italic></sub>=θ <italic>y</italic>. Hence <disp-formula-group id="M0003"><disp-formula><graphic xlink:href="tjbd-8-117-e003.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>As mentioned above, denoting the tumour cell's minimal quota for potassium by <italic>q</italic>, the tumour's true maximal growth rate by μ<sub>m</sub>, its death rate by <italic>D</italic>, and the tumour cell's rate of killing by immune system by <italic>f</italic>(<italic>x</italic>); and using the variable-internal-stores model based on the Droop [<xref rid="CIT0003" ref-type="bibr">3</xref>,<xref rid="CIT0004" ref-type="bibr">4</xref>] equation that relates growth rate to the internal cell quota (also see [<xref rid="CIT0005" ref-type="bibr">5</xref>,<xref rid="CIT0016" ref-type="bibr">16</xref>,<xref rid="CIT0037" ref-type="bibr">37</xref>]), the growth rate of the tumour cells is assumed to be governed by <disp-formula-group id="M0004"><disp-formula><graphic xlink:href="tjbd-8-117-e004.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>As for the dynamics of <italic>Q</italic> (the tumour's cell quota for K <sup>+</sup>), it is assumed that <italic>Q</italic>’s recruitment comes proportionally from the free potassium (<inline-formula><inline-graphic xlink:href="tjbd-8-117-m006.jpg"/></inline-formula>, α being the proportionality constant) and it is depleted by <inline-formula><inline-graphic xlink:href="tjbd-8-117-m007.jpg"/></inline-formula> because of cell growth. This results in the following simple equation <inline-formula><inline-graphic xlink:href="tjbd-8-117-m008.jpg"/></inline-formula>. It can be verified that <italic>Q</italic>(<italic>t</italic>)≥<italic>q</italic> for all <italic>t</italic>>0 since <italic>Q</italic>(0)≥<italic>q</italic>.</p><p>Since the cell metabolic process operates in a much faster pace than the growth of total biomass of either cell species, the quasi-steady-state argument allows us to approximate <italic>Q</italic>(<italic>t</italic>) by the solution of <inline-formula><inline-graphic xlink:href="tjbd-8-117-m009.jpg"/></inline-formula>. It gives <disp-formula-group id="M0005"><disp-formula><graphic xlink:href="tjbd-8-117-e005.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Using Equation (2), Equation (4) becomes <disp-formula-group id="M0006"><disp-formula><graphic xlink:href="tjbd-8-117-e006.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Substituting <italic>Q</italic> from Equation (5) into Equation (3), the equation for the growth dynamics of tumour cells can be written as <disp-formula-group id="M0007"><disp-formula><graphic xlink:href="tjbd-8-117-e007.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>We let <italic>e</italic> measure the addition rate of immune cells into the system when the tumour cells are K <sup>+</sup>-rich (when <italic>Q</italic>≥θ). If the tumour cells are K <sup>+</sup>-poor (when <italic>Q</italic><θ), then we assume that the addition rate suffers a reduction, and it becomes <italic>eQ</italic>/θ. This approach follows Liebig's [<xref rid="CIT0017" ref-type="bibr">17</xref>] minimum principle and has been used in [<xref rid="CIT0018" ref-type="bibr">18</xref>] model formulation. Thus, we have the following growth equation for immune cells: <disp-formula-group id="M0008"><disp-formula><graphic xlink:href="tjbd-8-117-e008.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>In Equation (7), parameter <italic>d</italic> represents the natural death rate of the immune cells.</p><p>Up till now, we considered potassium only. Now we include the possibility that carbon may also be a potentially limiting factor. It can be simply incorporated by assuming that if carbon acquisition limits the growth of the cancer cells then its population dynamics is governed by the classical logistic equation. Applying Liebig's minimum principle to potassium versus carbon limitation of the tumour cells and accordingly modifying Equation (6) and then combining it with Equation (7), we obtain the following tumour-immune cell growth model: <disp-formula-group id="M0009"><disp-formula><graphic xlink:href="tjbd-8-117-e009.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0010"><disp-formula><graphic xlink:href="tjbd-8-117-e010.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <inline-formula><inline-graphic xlink:href="tjbd-8-117-m010.jpg"/></inline-formula> and <italic>L</italic> is the carrying capacity of the tumour cells.</p><p> <italic>Remark 2.1</italic> Model (8) is indeed the model that has been referred to and studied as the KHE model in the literature (see [<xref rid="CIT0016" ref-type="bibr">16</xref>,<xref rid="CIT0037" ref-type="bibr">37</xref>]) for the growth dynamics of plant and herbivore under ecological stoichiometric principles. We have simply reinterpreted it for the tumour and immune cell interactions in potassium (K <sup>+</sup>)-limited environment.</p><p>We modify model (8) to consider the following model for the growth dynamics of tumour and immune cell system as our main model. <disp-formula-group id="M0011"><disp-formula><graphic xlink:href="tjbd-8-117-e011.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0012"><disp-formula><graphic xlink:href="tjbd-8-117-e012.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>In Equation (9b), the term (−<italic>lxy</italic>) represents the coercing of surrounding immune cells by tumour cells into a servile role in the tumour stroma and the term (<italic>u</italic>) may denote a medical treatment term or any antitumour activity that may help add immune cells in the body at a constant rate.</p><p>Now the two models (1) and (9) may be compared easily. It can be seen that incorporation of stoichiometry concepts in model (1) brings in two significant changes as given below: <list list-type="simple"><list-item><p>(a) Unlike Equation (1a) where the carrying capacity of the tumour cells is <italic>L</italic> (constant), the carrying capacity of the tumour cells in Equation (9a) based on stoichiometry principles depends on total potassium <italic>K</italic> <sub>t</sub> as well as on the biomass or density of the immune cells.</p></list-item><list-item><p>(b) The production efficiency of the immune cells in Equation (1b) is considered a constant <italic>e</italic> whereas in Equation (9b) it depends on the ratio of the cell quota of potassium of cancer cells and immune cells.</p></list-item></list> </p><p>Likewise [<xref rid="CIT0018" ref-type="bibr">18</xref>], we assume that the function <italic>f</italic>(<italic>x</italic>) (in Equations (9)) that denotes the rate of killing of tumour cells by immune cells is a bounded smooth function such that <disp-formula id="UM0001"><graphic xlink:href="tjbd-8-117-u001.jpg" position="float" orientation="portrait"/></disp-formula> It follows from the appendix in [<xref rid="CIT0018" ref-type="bibr">18</xref>] that the function <italic>P</italic>(<italic>x</italic>)=<italic>f</italic>(<italic>x</italic>)/<italic>x</italic> has the following properties: <disp-formula id="UM0002"><graphic xlink:href="tjbd-8-117-u002.jpg" position="float" orientation="portrait"/></disp-formula> For facilitation of analysis, another version of model (9) will also be used in this paper given as <disp-formula-group id="M0013"><disp-formula><graphic xlink:href="tjbd-8-117-e013.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0014"><disp-formula><graphic xlink:href="tjbd-8-117-e014.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Model equations (10) can be obtained from Equations (9) by substituting <inline-formula><inline-graphic xlink:href="tjbd-8-117-m011.jpg"/></inline-formula>, <italic>s</italic>=<italic>q</italic>/θ.</p><p>Here <italic>p</italic> denotes the maximal immune cells’ density allowed by the total potassium in the system and <italic>s</italic> is a dimensionless constant equal to the tumour cells minimal <inline-formula><inline-graphic xlink:href="tjbd-8-117-m012.jpg"/></inline-formula> divided by the constant immune cell's <inline-formula><inline-graphic xlink:href="tjbd-8-117-m013.jpg"/></inline-formula>. Model (10) is different from model (9) in that in Equations (10), θ is scaled out while all other parameters are retained.</p><p>The following theorem gives sufficient conditions that ensure that the solution of the model (10) (or model (9)) remains bounded.</p><statement id="E0001"><label>Theorem 2.1 </label><p>Let <inline-formula><inline-graphic xlink:href="tjbd-8-117-m014.jpg"/></inline-formula>. Solutions with initial conditions in the open trapezoid (or triangle if <inline-formula><inline-graphic xlink:href="tjbd-8-117-m015.jpg"/></inline-formula> <disp-formula id="UM0003"><graphic xlink:href="tjbd-8-117-u003.jpg" position="float" orientation="portrait"/></disp-formula> remain there for all forward times provided <inline-formula><inline-graphic xlink:href="tjbd-8-117-m016.jpg"/></inline-formula>.</p></statement><p>We relegate the proof of this theorem to the appendix. This result becomes important in that it gives bounds on <italic>u</italic> such that if <italic>u</italic> is chosen within these bounds then it guarantees that the solution of model (10) (or model (9)) remains bounded. As remarked above, the term (<italic>u</italic>) may easily be treated as the control parameter for the tumour and directly be related to the dose of the medicine to the patient during the course of his/her treatment.</p><p> <italic>Remark 2.2</italic> Model (9) is a generalization of the KHE model (2.1) in [<xref rid="CIT0037" ref-type="bibr">37</xref>] as it reduces to it when <italic>l</italic>=0 and <italic>u</italic>=0. The noticeable difference between the KHE model and model (9) is that they have different boundary equilibriums. While boundary equilibriums for the KHE model (or model (8) of this paper) are <italic>E</italic> <sub>0</sub>=(0, 0) and <italic>E</italic> <sub>1</sub>=(<italic>k</italic>, 0), model (9) has single boundary equilibrium <italic>E</italic> <sub>1</sub>=(0, <italic>u</italic>/<italic>d</italic>). For the dynamics of the KHE model and the stability of its equilibriums one may refer to various results (theorems) reported in [<xref rid="CIT0037" ref-type="bibr">37</xref>].</p></sec><sec id="S003"><label>3. </label><title>Local stability of model (9)</title><p>To study the equilibrium solutions of model (9) and their local stability, we rewrite this model in the following form: <disp-formula id="UM0004"><graphic xlink:href="tjbd-8-117-u004.jpg" position="float" orientation="portrait"/></disp-formula> where <disp-formula id="UM0005"><graphic xlink:href="tjbd-8-117-u005.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>For an equilibrium solution (<italic>x</italic> <sup>+</sup>, <italic>y</italic> <sup>+</sup>) of model (9) satisfying the equations <italic>F</italic>(<italic>x, y</italic>)=0 and <italic>G</italic>(<italic>x, y</italic>)=0, the Jacobian matrix of the system at this equilibrium can be written as <disp-formula-group id="M0015"><disp-formula><graphic xlink:href="tjbd-8-117-e015.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>The partial derivatives of <italic>F</italic> and <italic>G</italic>, after using the notations, <disp-formula-group id="M0016"><disp-formula><graphic xlink:href="tjbd-8-117-e016.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> can be written as <disp-formula id="UM0006"><graphic xlink:href="tjbd-8-117-u006.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>It can be seen that the model (9) has unique boundary equilibrium <italic>E</italic> <sub>1</sub>=(0, <italic>u</italic>/<italic>d</italic>). The Jacobian matrix (11) at <italic>E</italic> <sub>1</sub> turns out to be <disp-formula-group id="M0017"><disp-formula><graphic xlink:href="tjbd-8-117-e017.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula id="UM0007"><graphic xlink:href="tjbd-8-117-u007.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>The local asymptotic stability results for the boundary equilibrium <italic>E</italic> <sub>1</sub> can be easily obtained by studying the eigenvalues of the matrix (13). We state these results as follows: <disp-formula-group id="M0018"><disp-formula><graphic xlink:href="tjbd-8-117-e018.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>These results reveal: (i) that strengthening of immune system at larger rate may help eradicate the disease (result (a)) (ii) if the immune system is helped by the medical treatment at relatively smaller rates then while for an averaged value of <italic>K</italic> <sub>t</sub> the disease may be eradicated (result (b)) but a large value of <italic>K</italic> <sub>t</sub> may not eradicate the disease though in some cases (when internal equilibrium will turn out to be stable) it may help prolong the life of the patient with the disease (result (c)).</p><p>We now assume that an internal equilibrium <inline-formula><inline-graphic xlink:href="tjbd-8-117-m017.jpg"/></inline-formula> of model (9) exists. Note that −<italic>F</italic> <sub><italic>x</italic></sub>/<italic>F</italic> <sub><italic>y</italic></sub> and −<italic>G</italic> <sub><italic>x</italic></sub>/<italic>G</italic> <sub><italic>y</italic></sub> denote the slopes of the tumour cell and immune cell nullclines at (<italic>x, y</italic>) respectively. The determinant and the trace of the Jacobian matrix (11) at <italic>E</italic> <sub>2</sub> are <disp-formula id="UM0008"><graphic xlink:href="tjbd-8-117-u008.jpg" position="float" orientation="portrait"/></disp-formula> Let <disp-formula-group id="M0019"><disp-formula><graphic xlink:href="tjbd-8-117-e019.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> be the regions in the positive cone separated by the line <inline-formula><inline-graphic xlink:href="tjbd-8-117-m018.jpg"/></inline-formula>. We now state and prove the (local) stability results for the internal equilibrium <italic>E</italic> <sub>2</sub>.</p><statement id="E0002"><label>Theorem 3.1 </label><p> <list list-type="simple"><list-item><p>(a) Let <inline-formula><inline-graphic xlink:href="tjbd-8-117-m019.jpg"/></inline-formula> <list list-type="simple"><list-item><p>(i) Let <inline-formula><inline-graphic xlink:href="tjbd-8-117-m020.jpg"/></inline-formula>. If the slope of the immune cell's nullcline at <italic>E</italic> <sub>2</sub> is smaller than the tumour cell's (i.e. <inline-formula><inline-graphic xlink:href="tjbd-8-117-m021.jpg"/></inline-formula>, then <italic>E</italic> <sub>2</sub> is a saddle. If the slope of the immune cell's nullcline at <italic>E</italic> <sub>2</sub> is greater than the tumour cell's (i.e. <inline-formula><inline-graphic xlink:href="tjbd-8-117-m022.jpg"/></inline-formula> and if <inline-formula><inline-graphic xlink:href="tjbd-8-117-m023.jpg"/></inline-formula> then <italic>E</italic> <sub>2</sub> is locally asymptotically stable (LAS).</p></list-item><list-item><p>(ii) Let <inline-formula><inline-graphic xlink:href="tjbd-8-117-m024.jpg"/></inline-formula>. If the slope of the immune cell's nullcline at <italic>E</italic> <sub>2</sub> is greater than the tumour cell's (i.e. <inline-formula><inline-graphic xlink:href="tjbd-8-117-m025.jpg"/></inline-formula>) then <italic>E</italic> <sub>2</sub> is a saddle. If the slope of the immune cell's nullcline at <italic>E</italic> <sub>2</sub> is smaller than the tumour cell's (i.e. <inline-formula><inline-graphic xlink:href="tjbd-8-117-m026.jpg"/></inline-formula> and if <inline-formula><inline-graphic xlink:href="tjbd-8-117-m027.jpg"/></inline-formula> then <italic>E</italic> <sub>2</sub> is unstable.</p></list-item></list> </p></list-item><list-item><p>Let <inline-formula><inline-graphic xlink:href="tjbd-8-117-m028.jpg"/></inline-formula> <list list-type="simple"><list-item><p>(i) Let <inline-formula><inline-graphic xlink:href="tjbd-8-117-m029.jpg"/></inline-formula>. If the slope of the immune cell's nullcline at <italic>E</italic> <sub>2</sub> is smaller than the tumour cell's (i.e. <inline-formula><inline-graphic xlink:href="tjbd-8-117-m030.jpg"/></inline-formula>, then <italic>E</italic> <sub>2</sub> is a saddle. If the slope of the immune cell's nullcline at <italic>E</italic> <sub>2</sub> is greater than the tumour cell's (i.e. <inline-formula><inline-graphic xlink:href="tjbd-8-117-m031.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-117-m032.jpg"/></inline-formula> then <italic>E</italic> <sub>2</sub> is LAS.</p></list-item><list-item><p>(ii) Let <inline-formula><inline-graphic xlink:href="tjbd-8-117-m033.jpg"/></inline-formula>. If the slope of the immune cell's nullcline at <italic>E</italic> <sub>2</sub> is greater than the tumour cell's (i.e. <inline-formula><inline-graphic xlink:href="tjbd-8-117-m034.jpg"/></inline-formula> then <italic>E</italic> <sub>2</sub> is a saddle. If the slope of the immune cell's nullcline at <italic>E</italic> <sub>2</sub> is smaller than the tumour cell's (i.e. <inline-formula><inline-graphic xlink:href="tjbd-8-117-m035.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-117-m036.jpg"/></inline-formula> then <italic>E</italic> <sub>2</sub> is unstable.</p></list-item></list> </p></list-item></list> </p></statement><p> <italic>Proof</italic> Part (a) <list list-type="simple"><list-item><p>(i) Obviously <italic>F</italic> <sub><italic>y</italic></sub><0 and <italic>G</italic> <sub><italic>y</italic></sub><0 at <italic>E</italic> <sub>2</sub>. Since <disp-formula id="UM0009"><graphic xlink:href="tjbd-8-117-u009.jpg" position="float" orientation="portrait"/></disp-formula> it follows that <inline-formula><inline-graphic xlink:href="tjbd-8-117-m037.jpg"/></inline-formula> and <italic>E</italic> <sub>2</sub> is a saddle if <inline-formula><inline-graphic xlink:href="tjbd-8-117-m038.jpg"/></inline-formula>. Now if <inline-formula><inline-graphic xlink:href="tjbd-8-117-m039.jpg"/></inline-formula> then <inline-formula><inline-graphic xlink:href="tjbd-8-117-m040.jpg"/></inline-formula>. The condition <inline-formula><inline-graphic xlink:href="tjbd-8-117-m041.jpg"/></inline-formula> yields <italic>G</italic> <sub><italic>x</italic></sub><0. Then <inline-formula><inline-graphic xlink:href="tjbd-8-117-m042.jpg"/></inline-formula> gives <italic>F</italic> <sub><italic>x</italic></sub><0 and hence <inline-formula><inline-graphic xlink:href="tjbd-8-117-m043.jpg"/></inline-formula>. Thus <italic>E</italic> <sub>2</sub> is LAS.</p></list-item><list-item><p>(ii) In this case, <italic>F</italic> <sub><italic>y</italic></sub><0 and <italic>G</italic> <sub><italic>y</italic></sub>>0. Since <disp-formula id="UM0010"><graphic xlink:href="tjbd-8-117-u010.jpg" position="float" orientation="portrait"/></disp-formula> it follows that <inline-formula><inline-graphic xlink:href="tjbd-8-117-m044.jpg"/></inline-formula> and <italic>E</italic> <sub>2</sub> is a saddle if <inline-formula><inline-graphic xlink:href="tjbd-8-117-m045.jpg"/></inline-formula>. Now if <inline-formula><inline-graphic xlink:href="tjbd-8-117-m046.jpg"/></inline-formula>, then <inline-formula><inline-graphic xlink:href="tjbd-8-117-m047.jpg"/></inline-formula>. The condition <inline-formula><inline-graphic xlink:href="tjbd-8-117-m048.jpg"/></inline-formula> yields <italic>G</italic> <sub><italic>x</italic></sub><0. Then <inline-formula><inline-graphic xlink:href="tjbd-8-117-m049.jpg"/></inline-formula> implies <italic>F</italic> <sub><italic>x</italic></sub>>0 and thus <inline-formula><inline-graphic xlink:href="tjbd-8-117-m050.jpg"/></inline-formula>. Hence <italic>E</italic> <sub>2</sub> is unstable. Results of Part (b) can be proved similarly.</p></list-item></list> </p><p> <italic>Remark 3.1</italic> Local stability results of Theorem 3.1 are only partial results. Results for other situations could not be discussed as in those situations signs of the partial derivatives of <italic>F</italic> and <italic>G</italic> could not be determined.</p></sec><sec id="S004"><label>4. </label><title>A discrete analogue of model (9)</title><p>In this section, we consider a discrete analogue of the continuous model (9). We do so for three main reasons: (i) discrete time models may be more appropriate for application in experiments where data are collected on discrete time intervals or periodically, (ii) a comparison of the results of continuous and discrete models would give better idea about the robustness of the results of the continuous model on discrete time scale and (iii) by knowing the dynamics of both versions (continuous and discrete) of the models by theoretic analysis, we can be in a better position to justify which model type would fit well to the experimental investigations. Unfortunately, there have been very few instances, for example [<xref rid="CIT0021" ref-type="bibr">21</xref>], where experimental or clinical results have been compared with predictions of mathematical models. There can be several ways of deriving discrete time versions of dynamical systems corresponding to a continuous time formulation. Here we follow the method used in [<xref rid="CIT0008" ref-type="bibr">8</xref>]. Assuming that the per capita growth rates in Equation (9) change only at <italic>t</italic>=0, 1, 2 … , then model (9) can be written as <disp-formula-group id="M0020"><disp-formula><graphic xlink:href="tjbd-8-117-e020.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>Here [<italic>t</italic>] denotes the integral part of <inline-formula><inline-graphic xlink:href="tjbd-8-117-m051.jpg"/></inline-formula>. On any interval <inline-formula><inline-graphic xlink:href="tjbd-8-117-m052.jpg"/></inline-formula>, we can integrate Equation (16) and obtain the following equations for <italic>n</italic>≤<italic>t</italic><<italic>n</italic>+1: <disp-formula id="UM0011"><graphic xlink:href="tjbd-8-117-u011.jpg" position="float" orientation="portrait"/></disp-formula> Letting <italic>t</italic>→<italic>n</italic>+1 gives <disp-formula-group id="M0021"><disp-formula><graphic xlink:href="tjbd-8-117-e021.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>Equation (17) represent a discrete time analogue of model (9). In the following, we focus our attention on the study of equilibrium solutions of model (17) and their local asymptotic stability. To facilitate this analysis, we rewrite this model as <disp-formula-group id="M0022"><disp-formula><graphic xlink:href="tjbd-8-117-e022.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula id="UM0012"><graphic xlink:href="tjbd-8-117-u012.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>The fact that the model (9) and its discrete analogue (17) have the same equilibrium solutions implies that model (17) has unique boundary equilibrium <italic>E</italic> <sub>1</sub>=(0, <italic>u</italic>/<italic>d</italic>) and one (possibly multiple) internal equilibrium (equilibriums). We have denoted one such internal equilibrium by <italic>E</italic> <sub>2</sub> in Section 3.</p><p>The Jacobian of Equations (18) is <disp-formula-group id="M0023"><disp-formula><graphic xlink:href="tjbd-8-117-e023.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where the partial derivatives of <italic>R</italic> and <italic>S</italic> read as <disp-formula-group id="M0024"><disp-formula><graphic xlink:href="tjbd-8-117-e024.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0025"><disp-formula><graphic xlink:href="tjbd-8-117-e025.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0026"><disp-formula><graphic xlink:href="tjbd-8-117-e026.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0027"><disp-formula><graphic xlink:href="tjbd-8-117-e027.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>Here functions <italic>A</italic>(<italic>x</italic>) and <italic>B</italic>(<italic>x</italic>) are same as given in Equations (12).</p><p>The Jacobian matrix (19) at <italic>E</italic> <sub>1</sub> turns out to be <disp-formula-group id="M0028"><disp-formula><graphic xlink:href="tjbd-8-117-e028.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>Following results can be easily verified by studying the characteristic roots of the matrix (21). <disp-formula-group id="M0029"><disp-formula><graphic xlink:href="tjbd-8-117-e029.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0030"><disp-formula><graphic xlink:href="tjbd-8-117-e030.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>Now we discuss the local stability of internal equilibrium <italic>E</italic> <sub>2</sub>. The Jacobian matrix (19) at <italic>E</italic> <sub>2</sub> becomes <disp-formula-group id="M0031"><disp-formula><graphic xlink:href="tjbd-8-117-e031.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <disp-formula id="UM0013"><graphic xlink:href="tjbd-8-117-u013.jpg" position="float" orientation="portrait"/></disp-formula> The trace and the determinant of the matrix (23) are <disp-formula-group id="M0032"><disp-formula><graphic xlink:href="tjbd-8-117-e032.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>We will be using the following standard Jury test [<xref rid="CIT0006" ref-type="bibr">6</xref>] to prove the local asymptotic stability results for <italic>E</italic> <sub>2</sub>.</p><statement id="E0003"><label>lemma 4.1 </label><p>Let A be a 2×2 constant matrix. Both characteristic roots of A have magnitude less than 1 if and only if <disp-formula id="UM0014"><graphic xlink:href="tjbd-8-117-u014.jpg" position="float" orientation="portrait"/></disp-formula> </p></statement><p>Our main local asymptotic stability results are given in the following theorem.</p><statement id="E0004"><label>Theorem 4.1 </label><p>If the slope of the immune cell's nullcline at <italic>E</italic> <sub>2</sub> is smaller than the tumour cell's (i.e. <inline-formula><inline-graphic xlink:href="tjbd-8-117-m053.jpg"/></inline-formula>, then <italic>E</italic> <sub>2</sub> is unstable. If the slope of the immune cell's nullcline at <italic>E</italic> <sub>2</sub> is greater than the tumour cell's (i.e. <inline-formula><inline-graphic xlink:href="tjbd-8-117-m054.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-117-m055.jpg"/></inline-formula>, then <italic>E</italic> <sub>2</sub> is LAS.</p></statement><p> <italic>Proof</italic> In both regions Ω<sub>1</sub> and Ω<sub>2</sub>, <italic>R</italic> <sub><italic>y</italic></sub><0 and <italic>S</italic> <sub><italic>y</italic></sub><0. Thus <disp-formula id="UM0015"><graphic xlink:href="tjbd-8-117-u015.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Now if <inline-formula><inline-graphic xlink:href="tjbd-8-117-m056.jpg"/></inline-formula> at <italic>E</italic> <sub>2</sub>, then it follows that <inline-formula><inline-graphic xlink:href="tjbd-8-117-m057.jpg"/></inline-formula> and hence the desired result of Theorem 4.1 holds true because the inequality of Lemma 4.1 is violated. On the other hand, if <inline-formula><inline-graphic xlink:href="tjbd-8-117-m058.jpg"/></inline-formula> at <italic>E</italic> <sub>2</sub>, then it can be shown that whenever the inequality <inline-formula><inline-graphic xlink:href="tjbd-8-117-m059.jpg"/></inline-formula> holds true then the inequality of Lemma 4.1 holds and hence the desired result of Theorem 4.1 follows.</p><p> <italic>Remark 4.1</italic>  <list list-type="simple"><list-item><p>(i) It can be seen that a substitution <italic>l</italic>=0 and <italic>u</italic>=0 in the discrete model (17) leads us to the discrete KHE model (3.2) of Sui <italic>et al</italic>. [<xref rid="CIT0037" ref-type="bibr">37</xref>]. It can also be noticed by comparing Equations (20) of this paper with Equation (3.9) of Sui <italic>et al</italic>. [<xref rid="CIT0037" ref-type="bibr">37</xref>] that the nullcline functions <italic>R</italic> and <italic>S</italic> used in this paper retain same properties as displayed by corresponding functions of the KHE model. More specifically, the function <italic>S</italic> <sub><italic>x</italic></sub> (<italic>x, y</italic>) (<italic>G</italic> <sub><italic>x</italic></sub> (<italic>x, y</italic>) in [<xref rid="CIT0037" ref-type="bibr">37</xref>]) changes its sign from (+ve) in region Ω<sub>1</sub> to (−ve) in region Ω<sub>2</sub> and the function <italic>S</italic> <sub><italic>y</italic></sub> (<italic>x, y</italic>) becomes zero in region Ω<sub>1</sub>.</p></list-item><list-item><p>(ii) It is interesting to note that parameters <italic>l</italic> and <italic>u</italic> may produce some new situations. For example, a simple manoeuvring in <italic>l</italic> in Equation (20c) can change the sign of <italic>S</italic> <sub><italic>x</italic></sub> in either region thus affecting the dynamics. Choosing suitable value of <italic>l</italic> and considering <inline-formula><inline-graphic xlink:href="tjbd-8-117-m060.jpg"/></inline-formula> may even change the sign of <italic>S</italic> <sub><italic>x</italic></sub> from (−ve) in region Ω<sub>1</sub> to (+ve) in region Ω<sub>2</sub>. But we have not studied these situations in this paper.</p></list-item></list> </p></sec><sec id="S005"><label>5. </label><title>Numerical simulations</title><p>In this section, we present some numerical simulations for the continuous model (9) and the discrete model (17). We choose the Monod type function <inline-formula><inline-graphic xlink:href="tjbd-8-117-m061.jpg"/></inline-formula> as the functional response function of the immune system. All the numerical simulations are done with MATLAB. We will consider parameter values adapted by Loladze <italic>et al</italic>. [<xref rid="CIT0018" ref-type="bibr">18</xref>], Kuang <italic>et al</italic>. [<xref rid="CIT0016" ref-type="bibr">16</xref>], and Sui <italic>et al</italic>. [<xref rid="CIT0037" ref-type="bibr">37</xref>] as our reference data set. We reproduce this data set in <xref rid="T0001" ref-type="table">Table 1</xref>. <table-wrap id="T0001" orientation="portrait" position="float"><label>Table 1. </label><caption><title> Reference data set for model parameters.</title></caption><!--OASIS TABLE HERE--><table frame="hsides" rules="groups"><colgroup><col width="1"/><col width="1"/><col width="1*"/></colgroup><thead valign="bottom"><tr><th align="left">Parameter</th><th align="center">Value</th><th align="center">Unit</th></tr></thead><tbody><tr><td align="left"><italic>K</italic><sub>t</sub></td><td align="char" char=".">0.025</td><td align="left"><inline-formula><inline-graphic xlink:href="tjbd-8-117-m073.gif"/></inline-formula></td></tr><tr><td align="left"><italic>e</italic></td><td align="char" char=".">0.8</td><td align="left"> </td></tr><tr><td align="left"><italic>b</italic></td><td align="char" char=".">1.2</td><td align="left">day <sup>−1</sup></td></tr><tr><td align="left"><italic>d</italic></td><td align="char" char=".">0.25</td><td align="left">day <sup>−1</sup></td></tr><tr><td align="left">θ</td><td align="char" char=".">0.03</td><td align="left"><inline-formula><inline-graphic xlink:href="tjbd-8-117-m074.gif"/></inline-formula></td></tr><tr><td align="left"><italic>q</italic></td><td align="char" char=".">0.0038</td><td align="left"><inline-formula><inline-graphic xlink:href="tjbd-8-117-m075.gif"/></inline-formula></td></tr><tr><td align="left"><italic>c</italic></td><td align="char" char=".">0.81</td><td align="left">day <sup>−1</sup></td></tr><tr><td align="left"><italic>a</italic></td><td align="char" char=".">0.25</td><td align="left"><inline-formula><inline-graphic xlink:href="tjbd-8-117-m076.gif"/></inline-formula></td></tr><tr><td align="left">μ<sub>m</sub></td><td align="char" char=".">1.2</td><td align="left">day <sup>−1</sup></td></tr><tr><td align="left">α</td><td align="char" char=".">10</td><td align="left">day <sup>−1</sup></td></tr><tr><td align="left"><italic>L</italic></td><td align="char" char=".">0.25–2.0</td><td align="left"><inline-formula><inline-graphic xlink:href="tjbd-8-117-m077.gif"/></inline-formula></td></tr></tbody></table></table-wrap> </p><p>As pointed out in [<xref rid="CIT0037" ref-type="bibr">37</xref>], the condition <inline-formula><inline-graphic xlink:href="tjbd-8-117-m062.jpg"/></inline-formula> is satisfied with the initial conditions <inline-formula><inline-graphic xlink:href="tjbd-8-117-m063.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-117-m064.jpg"/></inline-formula> when <italic>l</italic>=0, <italic>u</italic>=0. Naturally this condition will remain satisfied when <italic>l</italic> and <italic>u</italic> are different from zero but small. We do not claim that the data of <xref rid="T0001" ref-type="table">Table 1</xref> represent any clinical situation. We have picked up these data because they provide desired negative signs of partial derivatives <italic>G</italic> <sub><italic>x</italic></sub> and <italic>S</italic> <sub><italic>x</italic></sub> giving stability changes for equilibriums and ensuring interesting dynamics. Moreover, same data closely mimic laboratory experiments investigating stoichiometric aspects of phytoplankton–zooplankton interactions [<xref rid="CIT0036" ref-type="bibr">36</xref>,<xref rid="CIT0038" ref-type="bibr">38</xref>] and thus in a sense represent biologically realistic values.</p><p>The main purpose of this section is to numerically investigate the dynamics of both the continuous model (9) and the discrete model (17) with the variation in the total potassium in the surrounding of the tumour and see whether increasing potassium K <sup>+</sup> has any protective role against the cancer as has been suggested by Jansson studies [<xref rid="CIT0013" ref-type="bibr">13</xref>,<xref rid="CIT0014" ref-type="bibr">14</xref>] and very recently by Bravi <italic>et al</italic>. [<xref rid="CIT0002" ref-type="bibr">2</xref>]. It can be seen through the bifurcation diagrams (<xref rid="F0001" ref-type="fig">Figure 1</xref>) that our continuous model (9) and its discrete analogue model (17) both support the protective role of potassium. In these bifurcation diagrams, the carrying capacity <italic>L</italic> of the tumour cells is taken as the bifurcation parameter. It is observed that when <inline-formula><inline-graphic xlink:href="tjbd-8-117-m065.jpg"/></inline-formula>, <italic>l</italic>=0.2, <italic>u</italic>=0.05 are fixed and other parameter values are chosen from <xref rid="T0001" ref-type="table">Table 1</xref>, then immune system fights well against averaged size tumours but it can fail for large size tumours that threat to win ultimately (<xref rid="F0001" ref-type="fig">Figure 1</xref>(A)–(D)). But when the value of <italic>K</italic> <sub>t</sub> is raised from 0.045 to 0.06 keeping all other parameters values fixed, then immune system shows strength to fight even larger size tumours. Here, the disease is not eradicated but it seems likely that the patient can live with disease all through his life (<xref rid="F0001" ref-type="fig">Figure 1</xref>(E)–(H)). <fig id="F0001" orientation="portrait" position="float"><label>Fig. 1. </label><caption><p>In this figure <italic>l</italic>=0.2, <italic>u</italic>=0.05 are fixed and values of rest of the parameters are chosen from <xref rid="T0001" ref-type="table">Table 1</xref>. (A), (C), (E) and (G) are bifurcation diagrams of the continuous model (9) while (B), (D) (F) and (H) are corresponding bifurcation diagrams of the discrete model (17). The carrying capacity of the tumour cell population <italic>L</italic> is considered as bifurcation parameter.</p></caption><graphic xlink:href="tjbd-8-117-g001"/></fig> </p><sec id="S005-S2001"><label>5.1. </label><title>Chaotic dynamics</title><p>In this section, we present one situation just for illustration that the dynamics of the continuous model (9) and the discrete model (17) may differ at times. The following bifurcation diagrams (<xref rid="F0002" ref-type="fig">Figure 2</xref>(A) and 2(B)) with <italic>b</italic>, the intrinsic growth rate of tumour cells, as bifurcation parameter show that the dynamics of the two models almost match for small and averaged values of <italic>b</italic> but for large <italic>b</italic>, while the dynamics of the continuous model (9) shows coexistence of tumour cells and immune cells at equilibrium values but the dynamics of the discrete model (17) exhibits chaotic dynamics with a route to chaos through periodic doubling. <fig id="F0002" orientation="portrait" position="float"><label>Fig. 2. </label><caption><p>In this figure <italic>l</italic>=0.009, <italic>u</italic>=0.02, <italic>L</italic>=1.6 are fixed and values of rest of the parameters are chosen from <xref rid="T0001" ref-type="table">Table 1</xref>. (A) is the bifurcation diagram of the continuous model (9), while (B) is the corresponding bifurcation diagram of discrete model (17). The intrinsic growth rate of the tumour cell population <italic>b</italic> is considered as bifurcation parameter.</p></caption><graphic xlink:href="tjbd-8-117-g002"/></fig> </p></sec></sec><sec id="S006"><label>6. </label><title>Discussion and conclusions</title><p>Potassium (K <sup>+</sup>) is an essential mineral found in most foods. It is a mineral that is required along with sodium and calcium for the body to work normally. It helps regulate major body functions including normal heart rhythm, blood pressure, water balance in the body, nerve impulses, muscle contractions and pH balance. The body cannot make potassium on its own and must get it from foods. Potassium is found in foods such as apricots, potatoes, bananas, oranges, pineapples, green leafy vegetables, whole grains, beans, nuts and lean meat. Most people get all the potassium they need from what they eat and drink.</p><p>The starting point for this paper has been the protective nature of potassium against the cancer risk as has been suggested by Jansson [<xref rid="CIT0013" ref-type="bibr">13</xref>,<xref rid="CIT0014" ref-type="bibr">14</xref>] studies. An Italian case-control study [<xref rid="CIT0002" ref-type="bibr">2</xref>] recently reported significant inverse trends in pancreatic risk for increasing intake of potassium. The present paper focuses on the role of a single element, i.e. potassium on cancer etiology by developing the model using the principles of stoichiometric theory. This brings in two significant changes in the classical approach of model formulation without stoichiometric considerations. Firstly, it makes the carrying capacity of the cancer cell population dependent on the total potassium as well as the immune cell population (see Equation (9a)). Secondly, it makes the growth of immune cell population dependent on the ratio of potassium quota for cancer cell and immune cell (see Equation (9b)). Another noticeable important change that the stoichiometric theory introduces into the model (9) different from classical predator–prey relationship type formulation is the possible sign change in <italic>G</italic> <sub><italic>x</italic></sub> (<italic>x, y</italic>) from positive (+ve) to negative (−ve) and vice versa (see model (9) and its representation in Section 3). The positive sign of <italic>G</italic> <sub><italic>x</italic></sub> would denote successful immunosurveillance by the immune system while negative sign of <italic>G</italic> <sub><italic>x</italic></sub> indicates either cancer's successful immunosuppressive activities or immune systems’ favouring approach to cancer progression. Data, supporting the dual host-protecting and tumours-sculpting actions of immunity (termed as cancer immunoediting in [<xref rid="CIT0039" ref-type="bibr">39</xref>]) have been reported in the literature [<xref rid="CIT0029" ref-type="bibr">29</xref>]. Vesely <italic>et al</italic>. [<xref rid="CIT0039" ref-type="bibr">39</xref>], describe cancer immunoediting as dynamic process comprising of three distinct phases: elimination, equilibrium and escape. It is interesting to note that model (9) can produce each of immunoediting's phases for specific choices of model parameters. The elimination phase is achieved when <italic>E</italic> <sub>1</sub>=(0, <italic>u</italic>/<italic>d</italic>) is stable. Equilibrium phase is possible when <inline-formula><inline-graphic xlink:href="tjbd-8-117-m066.jpg"/></inline-formula> is stable. The escape phase can be attained either by having a stable equilibrium <italic>E</italic> <sub>1</sub> (<italic>k</italic>, 0) under the condition <italic>l</italic>=0 and <italic>u</italic>=0 (see Remark 2.2) or by having equilibrium <inline-formula><inline-graphic xlink:href="tjbd-8-117-m067.jpg"/></inline-formula> as saddle and sustenance of both cancer and immune cell populations in an oscillatory mode. The oscillatory dynamics suggested by the model (9) though is not supported by any example in solid tumours; it has been shown to occur in systemic diseases such as leukaemia [<xref rid="CIT0019" ref-type="bibr">19</xref>]. It can be noted that the discrete model (17) depicts similar dynamics as mentioned above for model (9).</p><p>As pointed out earlier, the main purpose of the present paper is to investigate the protective nature of potassium against the cancer risk. It has been shown through bifurcation diagrams in Section 5 that the results of our continuous model (9) and its discrete analogue model (17) both suggest that increasing total potassium can play a protective role against cancer. It is observed that while for a small amount of potassium the immune system fights well for averaged size tumours but it may fail for larger size tumours (<xref rid="F0001" ref-type="fig">Figure 1</xref>(A)–(D)). On the other hand, when the amount of potassium is increased, the immune system gets stronger and it shows strength even to fight larger size tumours (<xref rid="F0001" ref-type="fig">Figure 1</xref>(E)–(H)). It can be noted that although <xref rid="F0001" ref-type="fig">Figure 1</xref>(E)–(H)) do not show the eradication of the disease but they certainly suggest a longer life for the patient. It has been noticed through numerous simulations that the continuous model (9) and the discrete model (17) mostly show similar dynamics but at times they may differ in their dynamics. A situation is illustrated in bifurcation diagram (<xref rid="F0002" ref-type="fig">Figure 2</xref>) when for large cancer intrinsic growth rate <italic>b</italic>, while model (9) shows survival of patient in cancer immunoediting equilibrium phase but model (17) suggests a chaotic dynamics having a period doubling route to chaos. Of course, the chaotic dynamics suggested by the discrete model (17) is a numerical result (a typical characteristic of discrete models) that does not have support from any experimental or clinical investigations till date but such results cannot be verified clinically or experimentally in future, who knows?</p></sec>
Multiple periodic solutions of a delayed predator–prey model with non-monotonic functional response and stage structure	<p>The paper studies a periodic and delayed predator–prey system with non-monotonic functional responses and stage structure. In the system, both the predator and prey are divided into immature individuals and mature individuals by two fixed ages. It is assumed that the immature predators cannot attack preys, and the case of the mature predators attacking the immature preys is also ignored. Based on Mawhin's coincidence degree, sufficient conditions are obtained for the existence of two positive periodic solutions of the system. An example is presented to illustrate the feasibility of the main results.</p>	<contrib contrib-type="author"><name><surname>Liu</surname><given-names>Yingyuan</given-names></name><xref ref-type="aff" rid="AF1"> <sup>a</sup> </xref><xref rid="AN1" ref-type="author-notes"/><xref rid="AN2" ref-type="author-notes"/></contrib><contrib contrib-type="author"><name><surname>Zhang</surname><given-names>Xiaolan</given-names></name><xref ref-type="aff" rid="AF1"> <sup>a</sup> </xref></contrib><contrib contrib-type="author"><name><surname>Zhou</surname><given-names>Tiejun</given-names></name><xref ref-type="aff" rid="AF1"> <sup>a</sup> </xref><xref ref-type="corresp" rid="cor1"> <sup>*</sup> </xref></contrib><aff id="AF1"><label><sup>a</sup></label><institution><named-content content-type="department">College of Science</named-content>, <named-content content-type="institution-name">Hunan Agricultural University</named-content></institution>, <named-content content-type="city">Changsha</named-content>, <named-content content-type="state">Hunan</named-content><named-content content-type="postal-code">410128</named-content>, <country>People's Republic of China</country></aff>	Journal of Biological Dynamics	<sec sec-type="intro" id="S001"><label>1. </label><title>Introduction</title><p>In a classical predator–prey model, it is generally assumed that there are no differences among the individuals of each species, which implies all the predators have the same survival probability and the same fertility, and all the preys also have the same survival probability and the same fertility. It is also assumed that each individual predator has the same attacking ability and each individual prey faces the same risk of being attacked. However, this phenomenon of no differences among individuals is very rare in the natural world. For example, the fertility and the attacking ability between an infant lion and an adult lion are apparently different. It is more reasonable to divide a species into different stages based on age. A simple method is to divide a species into two stages, the immature stage and the mature stage, where the immature individuals generally have no fertilities. There are different ways to impose the stage structure in the model, but usually only one species is taken into the consideration, for example, a stage structure for predator with fertility. Some systems consider stage structures only for the predator [<xref rid="CIT0002" ref-type="bibr">2</xref>,<xref rid="CIT0011" ref-type="bibr">11</xref>,<xref rid="CIT0014" ref-type="bibr">14</xref>,<xref rid="CIT0015" ref-type="bibr">15</xref>,<xref rid="CIT0017" ref-type="bibr">17</xref>,<xref rid="CIT0025" ref-type="bibr">25</xref>], and some consider stage structures only for the prey [<xref rid="CIT0008" ref-type="bibr">8</xref>,<xref rid="CIT0009" ref-type="bibr">9</xref>,<xref rid="CIT0019" ref-type="bibr">19–22</xref>,<xref rid="CIT0028" ref-type="bibr">28</xref>,<xref rid="CIT0029" ref-type="bibr">29</xref>]. In fact, a more general and more realistic system considers a stage structure for both the predator and the prey [<xref rid="CIT0005" ref-type="bibr">5</xref>,<xref rid="CIT0006" ref-type="bibr">6</xref>,<xref rid="CIT0013" ref-type="bibr">13</xref>,<xref rid="CIT0016" ref-type="bibr">16</xref>,<xref rid="CIT0027" ref-type="bibr">27</xref>]. In these systems with stage structures for both the predator and prey, immature predator attacking prey is generally ignored [<xref rid="CIT0013" ref-type="bibr">13</xref>]. In addition, considering that the immature preys are usually protected by their parents, the probability of immature prey being attacked is very small and therefore mature predator attacking immature prey can also be ignored [<xref rid="CIT0013" ref-type="bibr">13</xref>,<xref rid="CIT0027" ref-type="bibr">27</xref>]. In [<xref rid="CIT0027" ref-type="bibr">27</xref>], the following predator–prey model with stage structure for both the predator and prey is studied. <disp-formula-group id="M0001"><disp-formula><graphic xlink:href="tjbd-8-145-e001.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <italic>x</italic> <sub>1</sub>(<italic>t</italic>) and <italic>x</italic> <sub>2</sub>(<italic>t</italic>) denote the densities of the immature and mature individual preys at time <italic>t</italic>, respectively; <italic>y</italic> <sub>1</sub>(<italic>t</italic>) and <italic>y</italic> <sub>2</sub>(<italic>t</italic>) represent the densities of the immature and mature individual predators at time <italic>t</italic>, respectively. The term <inline-formula><inline-graphic xlink:href="tjbd-8-145-m001.jpg"/></inline-formula> represents the number of immature preys that were born at time <italic>t</italic>−τ<sub>1</sub>, still survive at time <italic>t</italic>, and transfer from the immature stage to the mature stage at time <italic>t</italic>. The term <disp-formula id="UM0001"><graphic xlink:href="tjbd-8-145-u001.jpg" position="float" orientation="portrait"/></disp-formula> represents the number of immature predators that were born at time <italic>t</italic>−τ<sub>2</sub>, still survive at time <italic>t</italic>, and transfer from the immature stage to the mature stage at time <italic>t</italic>. It is assumed in Equation (1) that the immature predators do not feed on preys and the mature predators only feed on the mature preys. Sufficient conditions are given for the permanence and existence of a positive periodic solutions to model (1) in [<xref rid="CIT0027" ref-type="bibr">27</xref>]. A stage-structured predator–prey system with functional response is an important population model, and it has been extensively studied recently. In these systems, three kinds of monotone functions <italic>g</italic>(<italic>x</italic>)=<italic>mx, mx</italic>/(<italic>a</italic>+<italic>x</italic>), <inline-formula><inline-graphic xlink:href="tjbd-8-145-m002.jpg"/></inline-formula>, where <italic>x</italic> denotes the density of prey, are often used [<xref rid="CIT0002" ref-type="bibr">2</xref>,<xref rid="CIT0008" ref-type="bibr">8</xref>,<xref rid="CIT0009" ref-type="bibr">9</xref>,<xref rid="CIT0011" ref-type="bibr">11</xref>,<xref rid="CIT0013" ref-type="bibr">13–15</xref>,<xref rid="CIT0017" ref-type="bibr">17</xref>,<xref rid="CIT0019" ref-type="bibr">19</xref>,<xref rid="CIT0021" ref-type="bibr">21</xref>,<xref rid="CIT0029" ref-type="bibr">29</xref>]. These functional response functions are monotonic for prey. But some experiments and observations indicate that a non-monotonic response also occurs under certain circumstances. For this reason, Andrews [<xref rid="CIT0001" ref-type="bibr">1</xref>] suggested the following function to model the non-monotonic response: <disp-formula id="UM0002"><graphic xlink:href="tjbd-8-145-u002.jpg" position="float" orientation="portrait"/></disp-formula> where <italic>m, a</italic>, and <italic>b</italic> are positive constants, which is called the Holling type-IV function. Its simplified form is <inline-formula><inline-graphic xlink:href="tjbd-8-145-m003.jpg"/></inline-formula>. There are many researches on the predator–prey with non-monotonic response [<xref rid="CIT0018" ref-type="bibr">18</xref>,<xref rid="CIT0023" ref-type="bibr">23–26</xref>,<xref rid="CIT0028" ref-type="bibr">28</xref>]. For example, Mischaikow and Wolkowicz [<xref rid="CIT0018" ref-type="bibr">18</xref>], Wolkowicz [<xref rid="CIT0023" ref-type="bibr">23</xref>] considered a general non-monotonic response function <italic>p</italic>(<italic>x</italic>), which was assumed to satisfy the following three conditions: <disp-formula-group id="M0002"><disp-formula><graphic xlink:href="tjbd-8-145-e002.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> there exists <italic>h</italic>>0 such that <disp-formula-group id="M0003"><disp-formula><graphic xlink:href="tjbd-8-145-e003.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> and <disp-formula-group id="M0004"><disp-formula><graphic xlink:href="tjbd-8-145-e004.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Obviously, the function <italic>g</italic>(<italic>x</italic>) above satisfies these assumptions. Xia <italic>et al.</italic> [<xref rid="CIT0026" ref-type="bibr">26</xref>] also considered a general non-monotonic response function <italic>p</italic>(<italic>x</italic>), which was assumed to satisfy the following three conditions: <disp-formula-group id="M0005"><disp-formula><graphic xlink:href="tjbd-8-145-e005.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> It is easy to see that the function <italic>g</italic>(<italic>x</italic>) above also satisfies the conditions (I)–(III).</p><p>Recently, some researchers incorporated the stage structure and the non-monotonic response into the predator–prey model [<xref rid="CIT0025" ref-type="bibr">25</xref>,<xref rid="CIT0028" ref-type="bibr">28</xref>]. For example, Yang <italic>et al.</italic> [<xref rid="CIT0028" ref-type="bibr">28</xref>] considered the following predator–prey system with Holling type-IV functional response and stage structure for prey in a periodic environment: <disp-formula id="UM0003"><graphic xlink:href="tjbd-8-145-u003.jpg" position="float" orientation="portrait"/></disp-formula> where <italic>x</italic> <sub>1</sub>(<italic>t</italic>) and <italic>x</italic> <sub>2</sub>(<italic>t</italic>) denote the density of the immature and mature prey, respectively, and <italic>y</italic>(<italic>t</italic>) is the density of predator that preys on <italic>x</italic> <sub>1</sub>. Xia <italic>et al.</italic> [<xref rid="CIT0025" ref-type="bibr">25</xref>] considered the following predator–prey system with Holling type-IV functional response and stage structure for predator in a periodic environment: <disp-formula-group id="M0006"><disp-formula><graphic xlink:href="tjbd-8-145-e006.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <italic>y</italic> <sub>1</sub>(<italic>t</italic>) and <italic>y</italic> <sub>2</sub>(<italic>t</italic>) denote the density of the immature and mature predator, respectively, and <italic>x</italic>(<italic>t</italic>) is the density of prey.</p><p>Because of the periodicity of the environment, researchers not only care about the permanence and extinction of predator–prey systems, but also concern about the periodic change of these systems [<xref rid="CIT0013" ref-type="bibr">13</xref>,<xref rid="CIT0025" ref-type="bibr">25</xref>,<xref rid="CIT0027" ref-type="bibr">27</xref>]. For example, by applying the method of coincidence degree, the authors of Xu <italic>et al.</italic> [<xref rid="CIT0027" ref-type="bibr">27</xref>] studied the existence of a positive periodic solution to system (1). At the same time, in order to explain the diversity of some systems, the multistability or multiperiodicity of those system are considered [<xref rid="CIT0003" ref-type="bibr">3</xref>,<xref rid="CIT0004" ref-type="bibr">4</xref>,<xref rid="CIT0007" ref-type="bibr">7</xref>,<xref rid="CIT0025" ref-type="bibr">25</xref>]. For example, the authors of Xia <italic>et al.</italic> [<xref rid="CIT0025" ref-type="bibr">25</xref>] obtained some sufficient conditions for the existence of at least two positive periodic solutions to system (6).</p><p>However, the combined effects of Holling type-IV functional response and the stage structure for both the predator and prey on a predator–prey model has not yet been widely studied. The motivation of this paper is to study the following delayed predator–prey system by replacing the ratio-dependent response function <inline-formula><inline-graphic xlink:href="tjbd-8-145-m004.jpg"/></inline-formula> of system (1) with the Holling type-IV function <inline-formula><inline-graphic xlink:href="tjbd-8-145-m005.jpg"/></inline-formula> where <inline-formula><inline-graphic xlink:href="tjbd-8-145-m006.jpg"/></inline-formula>. <disp-formula-group id="M0007"><disp-formula><graphic xlink:href="tjbd-8-145-e007.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where α<sub>1</sub>(<italic>t</italic>), α<sub>2</sub>(<italic>t</italic>), β<sub>1</sub>(<italic>t</italic>), β<sub>2</sub>(<italic>t</italic>), γ<sub>1</sub>(<italic>t</italic>), γ<sub>2</sub>(<italic>t</italic>) and <italic>a</italic> <sub>1</sub>(<italic>t</italic>) are continuous positive periodic functions with period ω, the constants <italic>m</italic> and <italic>a</italic> are positive.</p><p>The initial conditions in Equation (7) are of the form <disp-formula-group id="M0008"><disp-formula><graphic xlink:href="tjbd-8-145-e008.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> for <italic>i</italic>=1, 2, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m007.jpg"/></inline-formula>, where <inline-formula><inline-graphic xlink:href="tjbd-8-145-m008.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m009.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m010.jpg"/></inline-formula> are continuous positive periodic functions. The symbol ℝ<sup>+</sup> in the above denotes the set of all positive real numbers, and the symbol <inline-formula><inline-graphic xlink:href="tjbd-8-145-m011.jpg"/></inline-formula> denotes the set of all the non-negative real numbers. The main purpose of this paper is to obtain some sufficient conditions for the existence of multiple positive periodic solutions to system (7).</p></sec><sec id="S002"><label>2. </label><title>Main results</title><p>In order to prove the existence of positive periodic solutions to system (7), we first summarize some relative concepts and results from [<xref rid="CIT0010" ref-type="bibr">10</xref>] in the following.</p><p>Let <italic>X</italic> and <italic>Z</italic> be normed vector spaces, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m012.jpg"/></inline-formula> be a linear mapping, and <inline-formula><inline-graphic xlink:href="tjbd-8-145-m013.jpg"/></inline-formula> be a continuous mapping. The mapping <italic>L</italic> is called a Fredholm mapping of index zero if <inline-formula><inline-graphic xlink:href="tjbd-8-145-m014.jpg"/></inline-formula> and Im <italic>L</italic> is closed in <italic>Z</italic>. If <italic>L</italic> is a Fredholm mapping of index zero then there exist continuous projectors <inline-formula><inline-graphic xlink:href="tjbd-8-145-m015.jpg"/></inline-formula>, and <inline-formula><inline-graphic xlink:href="tjbd-8-145-m016.jpg"/></inline-formula> such that <inline-formula><inline-graphic xlink:href="tjbd-8-145-m017.jpg"/></inline-formula>, and <inline-formula><inline-graphic xlink:href="tjbd-8-145-m018.jpg"/></inline-formula>. It follows that <inline-formula><inline-graphic xlink:href="tjbd-8-145-m019.jpg"/></inline-formula> is invertible. We denote the inverse of the map by <italic>K</italic> <sub><italic>p</italic></sub>. If Ω is an bounded subset of <italic>X</italic>, the mapping <italic>N</italic> is then called <italic>L</italic>-compact on <inline-formula><inline-graphic xlink:href="tjbd-8-145-m020.jpg"/></inline-formula> if <inline-formula><inline-graphic xlink:href="tjbd-8-145-m021.jpg"/></inline-formula> is bounded and <inline-formula><inline-graphic xlink:href="tjbd-8-145-m022.jpg"/></inline-formula> is compact. Since Im <italic>Q</italic> is isomorphic to ker <italic>L</italic>, there exists an isomorphism <inline-formula><inline-graphic xlink:href="tjbd-8-145-m023.jpg"/></inline-formula>.</p><statement id="E0001"><label>Lemma 2.1 [10] </label><p>Let <inline-formula><inline-graphic xlink:href="tjbd-8-145-m024.jpg"/></inline-formula> be an open bounded set, <italic>L</italic> be a Fredholm mapping of index zero, and <italic>N</italic> be <italic>L</italic>-compact on <inline-formula><inline-graphic xlink:href="tjbd-8-145-m025.jpg"/></inline-formula>. Assume <list list-type="simple"><list-item><p>(i) <inline-formula><inline-graphic xlink:href="tjbd-8-145-m026.jpg"/></inline-formula> </p></list-item><list-item><p>(ii) <inline-formula><inline-graphic xlink:href="tjbd-8-145-m027.jpg"/></inline-formula> </p></list-item><list-item><p>(iii) <inline-formula><inline-graphic xlink:href="tjbd-8-145-m028.jpg"/></inline-formula> </p></list-item></list> </p><p>Then <italic>Lx</italic>=<italic>Nx</italic> has at least one solution in <inline-formula><inline-graphic xlink:href="tjbd-8-145-m029.jpg"/></inline-formula>.</p></statement><p>Note that the second equation and the fourth equation in Equation (7) can be separated from the whole system. Consider the following subsystem of Equation (7): <disp-formula-group id="M0009"><disp-formula><graphic xlink:href="tjbd-8-145-e009.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> with the initial conditions <disp-formula id="UM0004"><graphic xlink:href="tjbd-8-145-u004.jpg" position="float" orientation="portrait"/></disp-formula> where <inline-formula><inline-graphic xlink:href="tjbd-8-145-m030.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m031.jpg"/></inline-formula> are continuous positive periodic functions. Since we require <italic>x</italic> <sub>2</sub>(0)>0 and <italic>y</italic> <sub>2</sub>(0)>0, each component of the solutions is positive as long as <italic>t</italic>>0 and the solutions are defined.</p><p>If <inline-formula><inline-graphic xlink:href="tjbd-8-145-m032.jpg"/></inline-formula> is a positive ω-periodic solution to system (9), then it is not difficult to verify that <disp-formula id="UM0005"><graphic xlink:href="tjbd-8-145-u005.jpg" position="float" orientation="portrait"/></disp-formula> are also ω-periodic by the periodicity of the coefficients of system (7). For system (7), consider the following two linear periodic differential equations: <disp-formula-group id="M0010"><disp-formula><graphic xlink:href="tjbd-8-145-e010.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0011"><disp-formula><graphic xlink:href="tjbd-8-145-e011.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Under the initial condition (8), Equation (10) has a unique solution <disp-formula-group id="M0012"><disp-formula><graphic xlink:href="tjbd-8-145-e012.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Similarly, Equation (11) also has a unique solution <disp-formula-group id="M0013"><disp-formula><graphic xlink:href="tjbd-8-145-e013.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> </p><p>If the following condition holds: <disp-formula id="UM0006"><graphic xlink:href="tjbd-8-145-u006.jpg" position="float" orientation="portrait"/></disp-formula> then from Equations (12) and (13), Equation (10) has a unique ω-periodic solution <disp-formula id="UM0007"><graphic xlink:href="tjbd-8-145-u007.jpg" position="float" orientation="portrait"/></disp-formula> and Equation (11) also has a unique ω-periodic solution <disp-formula id="UM0008"><graphic xlink:href="tjbd-8-145-u008.jpg" position="float" orientation="portrait"/></disp-formula> The positivity of <inline-formula><inline-graphic xlink:href="tjbd-8-145-m033.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m034.jpg"/></inline-formula> and the coefficients of Equation (7) implies that of <inline-formula><inline-graphic xlink:href="tjbd-8-145-m035.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-145-m036.jpg"/></inline-formula>. Therefore, in order to prove the existence of multiply positive periodic solutions for system (7), we only need to proof it for system (9).</p><p>For simplicity we adopt the following notations throughout this paper: <disp-formula id="UM0009"><graphic xlink:href="tjbd-8-145-u009.jpg" position="float" orientation="portrait"/></disp-formula> where the function <italic>g</italic>(<italic>t</italic>) is continuous on [0, ω].</p><p>Let <disp-formula id="UM0010"><graphic xlink:href="tjbd-8-145-u010.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>From now on, we assume that <disp-formula id="UM0011"><graphic xlink:href="tjbd-8-145-u011.jpg" position="float" orientation="portrait"/></disp-formula> Under assumption (H2), there exist the following six positive numbers: <disp-formula id="UM0012"><graphic xlink:href="tjbd-8-145-u012.jpg" position="float" orientation="portrait"/></disp-formula> It is not difficult to show that <disp-formula id="UM0013"><graphic xlink:href="tjbd-8-145-u013.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Furthermore, we make the third assumption, <disp-formula id="UM0014"><graphic xlink:href="tjbd-8-145-u014.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Our result on the existence of multiple periodic solutions to system (7) is stated as the following theorem.</p><statement id="E0002"><label>Theorem 2.2 </label><p>Suppose that the conditions (H1), (H2), and (H3) hold. Then system (7) with the initial condition (8) has at least two positive ω-periodic solutions.</p></statement><p> <italic>Proof</italic> From the above analysis, we only need to prove that system (9) has at least two positive ω-periodic solutions. By making the changes of variables <disp-formula id="UM0015"><graphic xlink:href="tjbd-8-145-u015.jpg" position="float" orientation="portrait"/></disp-formula> system (9) is rewritten as <disp-formula-group id="M0014"><disp-formula><graphic xlink:href="tjbd-8-145-e014.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> where <inline-formula><inline-graphic xlink:href="tjbd-8-145-m037.jpg"/></inline-formula>. We set <disp-formula id="UM0016"><graphic xlink:href="tjbd-8-145-u016.jpg" position="float" orientation="portrait"/></disp-formula> and define the norm of <italic>X</italic> and <italic>Z</italic> as <disp-formula id="UM0017"><graphic xlink:href="tjbd-8-145-u017.jpg" position="float" orientation="portrait"/></disp-formula> where \|·\| denotes the Euclidean norm. Then <italic>X</italic> and <italic>Z</italic> are Banach spaces when they are endowed with the usual operations and norm <inline-formula><inline-graphic xlink:href="tjbd-8-145-m038.jpg"/></inline-formula>. Since <inline-formula><inline-graphic xlink:href="tjbd-8-145-m039.jpg"/></inline-formula>, there exist ξ<sub><italic>i</italic></sub> and <inline-formula><inline-graphic xlink:href="tjbd-8-145-m040.jpg"/></inline-formula>, such that <disp-formula id="UM0018"><graphic xlink:href="tjbd-8-145-u018.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>For any <inline-formula><inline-graphic xlink:href="tjbd-8-145-m041.jpg"/></inline-formula>, by the periodicity of the coefficients of system (14), we can easily check that both <disp-formula id="UM0019"><graphic xlink:href="tjbd-8-145-u019.jpg" position="float" orientation="portrait"/></disp-formula> and <disp-formula id="UM0020"><graphic xlink:href="tjbd-8-145-u020.jpg" position="float" orientation="portrait"/></disp-formula> are ω-periodic with respect to <italic>t</italic>. Set <disp-formula id="UM0021"><graphic xlink:href="tjbd-8-145-u021.jpg" position="float" orientation="portrait"/></disp-formula> and <disp-formula id="UM0022"><graphic xlink:href="tjbd-8-145-u022.jpg" position="float" orientation="portrait"/></disp-formula> It is not difficult to show that <inline-formula><inline-graphic xlink:href="tjbd-8-145-m042.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-145-m043.jpg"/></inline-formula> <inline-formula><inline-graphic xlink:href="tjbd-8-145-m044.jpg"/></inline-formula> are closed in <italic>Z</italic>. Then <inline-formula><inline-graphic xlink:href="tjbd-8-145-m045.jpg"/></inline-formula> It follows that <italic>L</italic> is a Fredholm mapping of index zero. Define two mappings <italic>P</italic> and <italic>Q</italic> as <disp-formula id="UM0023"><graphic xlink:href="tjbd-8-145-u023.jpg" position="float" orientation="portrait"/></disp-formula> and <disp-formula id="UM0024"><graphic xlink:href="tjbd-8-145-u024.jpg" position="float" orientation="portrait"/></disp-formula> Then <italic>P</italic> and <italic>Q</italic> are continuous projectors such that <disp-formula id="UM0025"><graphic xlink:href="tjbd-8-145-u025.jpg" position="float" orientation="portrait"/></disp-formula> Furthermore, the generalized inverse (of <italic>L</italic>) <inline-formula><inline-graphic xlink:href="tjbd-8-145-m046.jpg"/></inline-formula> exists and is given by <disp-formula id="UM0026"><graphic xlink:href="tjbd-8-145-u026.jpg" position="float" orientation="portrait"/></disp-formula> Then <inline-formula><inline-graphic xlink:href="tjbd-8-145-m047.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-145-m048.jpg"/></inline-formula> are given, respectively, by <disp-formula-group id="M0015"><disp-formula><graphic xlink:href="tjbd-8-145-e015.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula id="UM0027"><graphic xlink:href="tjbd-8-145-u027.jpg" position="float" orientation="portrait"/></disp-formula> where <inline-formula><inline-graphic xlink:href="tjbd-8-145-m049.jpg"/></inline-formula>. Clearly, <italic>QN</italic> and <italic>K</italic> <sub><italic>p</italic></sub>(<italic>I</italic>−<italic>Q</italic>)<italic>N</italic> are continuous. By using the Arzela–Ascoli theorem, it is not difficult to prove that <inline-formula><inline-graphic xlink:href="tjbd-8-145-m050.jpg"/></inline-formula> is compact for any open bounded set <inline-formula><inline-graphic xlink:href="tjbd-8-145-m051.jpg"/></inline-formula>. Moreover, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m052.jpg"/></inline-formula> is bounded too. Hence <italic>N</italic> is <italic>L</italic>-compact on <inline-formula><inline-graphic xlink:href="tjbd-8-145-m053.jpg"/></inline-formula> for the open bounded set <inline-formula><inline-graphic xlink:href="tjbd-8-145-m054.jpg"/></inline-formula>. In order to apply Lemma 2.1 to prove the existence of two periodic solutions of system (14), we need to construct two appropriate open bounded subsets in <italic>X</italic>. Corresponding to the operator equation <inline-formula><inline-graphic xlink:href="tjbd-8-145-m055.jpg"/></inline-formula>, we have <disp-formula-group id="M0016"><disp-formula><graphic xlink:href="tjbd-8-145-e016.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0017"><disp-formula><graphic xlink:href="tjbd-8-145-e017.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Suppose <inline-formula><inline-graphic xlink:href="tjbd-8-145-m056.jpg"/></inline-formula> is a solution to Equations (16) and (17) for a certain <inline-formula><inline-graphic xlink:href="tjbd-8-145-m057.jpg"/></inline-formula>. Integrating Equations (16) and (17) over the interval [0, ω], we obtain <disp-formula-group id="M0018"><disp-formula><graphic xlink:href="tjbd-8-145-e018.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> <disp-formula-group id="M0019"><disp-formula><graphic xlink:href="tjbd-8-145-e019.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> It follows from Equations (17) and (19) that <disp-formula id="UM0028"><graphic xlink:href="tjbd-8-145-u028.jpg" position="float" orientation="portrait"/></disp-formula> Therefore, <disp-formula-group id="M0020"><disp-formula><graphic xlink:href="tjbd-8-145-e020.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Multiplying Equation (16) by <inline-formula><inline-graphic xlink:href="tjbd-8-145-m058.jpg"/></inline-formula> and then integrating it over [0, ω], we obtain <disp-formula-group id="M0021"><disp-formula><graphic xlink:href="tjbd-8-145-e021.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> With the inequality <disp-formula id="UM0029"><graphic xlink:href="tjbd-8-145-u029.jpg" position="float" orientation="portrait"/></disp-formula> and Equation (21), we get <disp-formula id="UM0030"><graphic xlink:href="tjbd-8-145-u030.jpg" position="float" orientation="portrait"/></disp-formula> Then we have <disp-formula-group id="M0022"><disp-formula><graphic xlink:href="tjbd-8-145-e022.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Multiplying Equation (17) by <inline-formula><inline-graphic xlink:href="tjbd-8-145-m059.jpg"/></inline-formula>, then integrating it over [0, ω], we obtain <disp-formula id="UM0031"><graphic xlink:href="tjbd-8-145-u031.jpg" position="float" orientation="portrait"/></disp-formula> which yields <disp-formula-group id="M0023"><disp-formula><graphic xlink:href="tjbd-8-145-e023.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> In addition, from Equation (21) we have <disp-formula id="UM0032"><graphic xlink:href="tjbd-8-145-u032.jpg" position="float" orientation="portrait"/></disp-formula> Simply, <disp-formula-group id="M0024"><disp-formula><graphic xlink:href="tjbd-8-145-e024.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> From Equations (23) and (24), we can get <disp-formula id="UM0033"><graphic xlink:href="tjbd-8-145-u033.jpg" position="float" orientation="portrait"/></disp-formula> Combining it with Equation (22), we obtain <disp-formula-group id="M0025"><disp-formula><graphic xlink:href="tjbd-8-145-e025.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> From Equations (16), (18), (22) and (25), we have <disp-formula id="UM0034"><graphic xlink:href="tjbd-8-145-u034.jpg" position="float" orientation="portrait"/></disp-formula> This means <disp-formula-group id="M0026"><disp-formula><graphic xlink:href="tjbd-8-145-e026.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Multiplying Equation (17) by <inline-formula><inline-graphic xlink:href="tjbd-8-145-m060.jpg"/></inline-formula> and integrating it over [0, ω], we also obtain <disp-formula id="UM0035"><graphic xlink:href="tjbd-8-145-u035.jpg" position="float" orientation="portrait"/></disp-formula> Then, <disp-formula id="UM0036"><graphic xlink:href="tjbd-8-145-u036.jpg" position="float" orientation="portrait"/></disp-formula> This gives <disp-formula id="UM0037"><graphic xlink:href="tjbd-8-145-u037.jpg" position="float" orientation="portrait"/></disp-formula> Therefore, we have <disp-formula-group id="M0027"><disp-formula><graphic xlink:href="tjbd-8-145-e027.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Similarly, we obtain <disp-formula-group id="M0028"><disp-formula><graphic xlink:href="tjbd-8-145-e028.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> From Equations (26) and (27), it is easy to see that <disp-formula id="UM0038"><graphic xlink:href="tjbd-8-145-u038.jpg" position="float" orientation="portrait"/></disp-formula> A special case is <disp-formula id="UM0039"><graphic xlink:href="tjbd-8-145-u039.jpg" position="float" orientation="portrait"/></disp-formula> Therefore, we have <disp-formula id="UM0040"><graphic xlink:href="tjbd-8-145-u040.jpg" position="float" orientation="portrait"/></disp-formula> From (H2), it is not difficult to obtain that <disp-formula-group id="M0029"><disp-formula><graphic xlink:href="tjbd-8-145-e029.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Similarly, by Equations (26) and (28) we get <disp-formula id="UM0041"><graphic xlink:href="tjbd-8-145-u041.jpg" position="float" orientation="portrait"/></disp-formula> Specially, we have <disp-formula-group id="M0030"><disp-formula><graphic xlink:href="tjbd-8-145-e030.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> and it follows that <disp-formula id="UM0042"><graphic xlink:href="tjbd-8-145-u042.jpg" position="float" orientation="portrait"/></disp-formula> In view of (H2), we have <disp-formula-group id="M0031"><disp-formula><graphic xlink:href="tjbd-8-145-e031.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> and <disp-formula-group id="M0032"><disp-formula><graphic xlink:href="tjbd-8-145-e032.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> From Equation (21) we can get that <disp-formula id="UM0043"><graphic xlink:href="tjbd-8-145-u043.jpg" position="float" orientation="portrait"/></disp-formula> Therefore <disp-formula id="UM0044"><graphic xlink:href="tjbd-8-145-u044.jpg" position="float" orientation="portrait"/></disp-formula> Combining it with Equation (32), we have <disp-formula-group id="M0033"><disp-formula><graphic xlink:href="tjbd-8-145-e033.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> From Equation (21), it also follows that <disp-formula id="UM0045"><graphic xlink:href="tjbd-8-145-u045.jpg" position="float" orientation="portrait"/></disp-formula> Therefore, <disp-formula id="UM0046"><graphic xlink:href="tjbd-8-145-u046.jpg" position="float" orientation="portrait"/></disp-formula> By Equations (29) and (32), we obtain <disp-formula id="UM0047"><graphic xlink:href="tjbd-8-145-u047.jpg" position="float" orientation="portrait"/></disp-formula> From (H3), we have <disp-formula id="UM0048"><graphic xlink:href="tjbd-8-145-u048.jpg" position="float" orientation="portrait"/></disp-formula> which implies <disp-formula id="UM0049"><graphic xlink:href="tjbd-8-145-u049.jpg" position="float" orientation="portrait"/></disp-formula> Together with Equation (20), it leads to <disp-formula-group id="M0034"><disp-formula><graphic xlink:href="tjbd-8-145-e034.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> From Equations (20) and (33), we also have <disp-formula-group id="M0035"><disp-formula><graphic xlink:href="tjbd-8-145-e035.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Combining Equation (34) with Equation (35), we obtain <disp-formula-group id="M0036"><disp-formula><graphic xlink:href="tjbd-8-145-e036.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Now we consider <italic>QNu</italic>, where <inline-formula><inline-graphic xlink:href="tjbd-8-145-m061.jpg"/></inline-formula>. From Equation (15) we have <disp-formula-group id="M0037"><disp-formula><graphic xlink:href="tjbd-8-145-e037.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Because of (H2) and (H3), we can show that the equation <inline-formula><inline-graphic xlink:href="tjbd-8-145-m062.jpg"/></inline-formula> has two distinct solutions <disp-formula id="UM0050"><graphic xlink:href="tjbd-8-145-u050.jpg" position="float" orientation="portrait"/></disp-formula> Choose a positive constant <italic>c</italic> such that <disp-formula-group id="M0038"><disp-formula><graphic xlink:href="tjbd-8-145-e038.jpg" position="float" orientation="portrait"/></disp-formula></disp-formula-group> Define <disp-formula id="UM0051"><graphic xlink:href="tjbd-8-145-u051.jpg" position="float" orientation="portrait"/></disp-formula> Then, both Ω<sub>1</sub> and Ω<sub>2</sub> are bounded open subsets of <italic>X</italic>. It follows from Equations (9) and (38) that <inline-formula><inline-graphic xlink:href="tjbd-8-145-m063.jpg"/></inline-formula>. With the help of Equations (9), (29)–(32), (36) and (38), it is easy to see that <inline-formula><inline-graphic xlink:href="tjbd-8-145-m064.jpg"/></inline-formula>, and Ω<sub><italic>i</italic></sub> satisfies the requirement (<italic>i</italic>) in Lemma 2.1 for <italic>i</italic>=1, 2. Moreover, <italic>QNu</italic>≠0 for <inline-formula><inline-graphic xlink:href="tjbd-8-145-m065.jpg"/></inline-formula>. Taking <inline-formula><inline-graphic xlink:href="tjbd-8-145-m066.jpg"/></inline-formula>, we obtain from Equation (37) that <disp-formula id="UM0052"><graphic xlink:href="tjbd-8-145-u052.jpg" position="float" orientation="portrait"/></disp-formula> </p><p>Since <italic>m</italic>, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m067.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m068.jpg"/></inline-formula>, exp(<italic>u</italic> <sub>1</sub>), exp(<italic>u</italic> <sub>2</sub>) and <italic>f</italic>(<italic>u</italic> <sub>1</sub>) are all positive, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m069.jpg"/></inline-formula> depends on the sign of <inline-formula><inline-graphic xlink:href="tjbd-8-145-m070.jpg"/></inline-formula>. When <inline-formula><inline-graphic xlink:href="tjbd-8-145-m071.jpg"/></inline-formula>, <italic>u</italic> <sub>1</sub> is equal to ln<italic>u</italic> <sub>−</sub>. Therefore, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m072.jpg"/></inline-formula>. When <inline-formula><inline-graphic xlink:href="tjbd-8-145-m073.jpg"/></inline-formula>, <italic>u</italic> <sub>1</sub> is equal to ln<italic>u</italic> <sub>+</sub>. Then, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m074.jpg"/></inline-formula>. Hence we obtain that <inline-formula><inline-graphic xlink:href="tjbd-8-145-m075.jpg"/></inline-formula>. Now we have proved that Ω<sub><italic>i</italic></sub> satisfies all the assumptions in Lemma 2.1. Here, system (14) has at least two ω-periodic solutions <inline-formula><inline-graphic xlink:href="tjbd-8-145-m076.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-145-m077.jpg"/></inline-formula> with <inline-formula><inline-graphic xlink:href="tjbd-8-145-m078.jpg"/></inline-formula>. Obviously, <italic>u</italic>*(<italic>t</italic>) and <italic>u</italic> <sup>+</sup>(<italic>t</italic>) are different. Let <inline-formula><inline-graphic xlink:href="tjbd-8-145-m079.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-145-m080.jpg"/></inline-formula>. Then <inline-formula><inline-graphic xlink:href="tjbd-8-145-m081.jpg"/></inline-formula> and <inline-formula><inline-graphic xlink:href="tjbd-8-145-m082.jpg"/></inline-formula> are two different positive ω-periodic solutions to system (9). This completes the proof of Theorem 2.2.</p></sec><sec id="S003"><label>3. </label><title>Example</title><p>In system (7), let <inline-formula><inline-graphic xlink:href="tjbd-8-145-m083.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m084.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m085.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m086.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m087.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m088.jpg"/></inline-formula>, <italic>m</italic>=<italic>a</italic>=1. We have <inline-formula><inline-graphic xlink:href="tjbd-8-145-m089.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m090.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m091.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m092.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m093.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m094.jpg"/></inline-formula>, <inline-formula><inline-graphic xlink:href="tjbd-8-145-m095.jpg"/></inline-formula>, and <inline-formula><inline-graphic xlink:href="tjbd-8-145-m096.jpg"/></inline-formula>.</p><p>Taking the initial condition <disp-formula id="UM0053"><graphic xlink:href="tjbd-8-145-u053.jpg" position="float" orientation="portrait"/></disp-formula> where <disp-formula id="UM0054"><graphic xlink:href="tjbd-8-145-u054.jpg" position="float" orientation="portrait"/></disp-formula> it is easy to verify the assumption (H1) holds.</p><p>By computation, we have <italic>c</italic> <sub>−</sub>=0.26, <italic>c</italic> <sub>+</sub>=2.86, <italic>u</italic> <sub>−</sub>=0.41, <italic>u</italic> <sub>+</sub>=2.45, <italic>l</italic> <sub>−</sub>=0.63,<italic>l</italic> <sub>+</sub>=1.58, <italic>B</italic>=0.15. Then we can verify the following two inequalities: <disp-formula id="UM0055"><graphic xlink:href="tjbd-8-145-u055.jpg" position="float" orientation="portrait"/></disp-formula> and <disp-formula id="UM0056"><graphic xlink:href="tjbd-8-145-u056.jpg" position="float" orientation="portrait"/></disp-formula> The above inequalities show that assumptions (H2) and (H3) hold. Thus, by Theorem 2.2, system (7) has at least two different positive periodic solutions.</p></sec><sec sec-type="conclusions" id="S004"><label>4. </label><title>Conclusion</title><p>In this paper, we study the existence of multiple positive periodic solutions to system (7), in which the coefficients are periodic, the predator functional response is non-monotonic, predator and prey species are all divided into immature individuals and mature individuals. By using Mawhin's continuation theorem of coincidence degree theory, we have proved that there exist at least two positive periodic solutions to system (7) under the assumptions (H1),(H2) and (H3). From (H2) and (H3), we know that all parameters of system (7) have effects on the existence of positive periodic solutions and the period ω of the coefficients is an important influence factor on the existence of positive periodic solutions. We found that, when the period ω enlarges, for the existence of periodic solutions to system (7), the infimums of birth rates of prey and the conversion of nutrients into the reproduction rate of mature predator must be increased. In other words, to shorten the period of the environmental change can increase the possibility of the existence of multiple periodic solutions.</p></sec>
Clinical impact of ultra deep versus Sanger sequencing detection of minority mutations on HIV-1 drug resistance genotype interpretation after virological failure	Could not extract abstract	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Mohamed</surname><given-names>Sofiane</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I4">4</xref></contrib><contrib contrib-type="author" id="A2"><name><surname>Penaranda</surname><given-names>Guillaume</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A3"><name><surname>Gonzalez</surname><given-names>Dimitri</given-names></name><xref ref-type="aff" rid="I2">2</xref></contrib><contrib contrib-type="author" id="A4"><name><surname>Camus</surname><given-names>Claire</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A5"><name><surname>Khiri</surname><given-names>Hacène</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A6"><name><surname>Boulmé</surname><given-names>Ronan</given-names></name><xref ref-type="aff" rid="I2">2</xref></contrib><contrib contrib-type="author" id="A7"><name><surname>Sayada</surname><given-names>Chalom</given-names></name><xref ref-type="aff" rid="I2">2</xref></contrib><contrib contrib-type="author" id="A8"><name><surname>Philibert</surname><given-names>Patrick</given-names></name><xref ref-type="aff" rid="I3">3</xref></contrib><contrib contrib-type="author" id="A9"><name><surname>Olive</surname><given-names>Daniel</given-names></name><xref ref-type="aff" rid="I4">4</xref></contrib><contrib contrib-type="author" id="A10"><name><surname>Halfon</surname><given-names>Philippe</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib>	BMC Infectious Diseases	<sec><title>Aim</title><p>Drug resistance mutations are routinely detected using standard Sanger sequencing, which does not detect minor variants with a frequency below 20%. The impact of detecting minor variants generated by ultra-deep sequencing (UDS) on HIV drug-resistance (DR) interpretations has not yet been studied.</p></sec><sec sec-type="methods"><title>Methods</title><p>Fifty HIV-1 patients who experienced virological failure were included in this retrospective study. The HIV-1 UDS protocol was performed using the GS Junior (Roche 454 Life Sciences Branford, CT). This UDS protocol allowed the detection and quantification of minor and major HIV-1 protease and reverse transcriptase variants related to genotypes A, B, C, E, F and G. DeepChek®-HIV (ABL, SA and TherapyEdgeTM, USA) simplified drug resistance (DR) interpretation software was used to compare Sanger sequencing and UDS at two different thresholds (≥1% and ≥20%). DeepChek®-HIV utilizes the ANRS, HIVdb and Rega algorithms.</p></sec><sec sec-type="results"><title>Results</title><p>The total time required for the UDS protocol was found to be approximately three times longer than Sanger sequencing with equivalent reagent costs. UDS detected all of the mutations found by population sequencing and identified additional resistance variants in all patients, primarily by using 1% sensitivity. An analysis of DR revealed a total of 643 and 224 clinically relevant mutations by UDS and Sanger sequencing, respectively. Three resistance mutations with >20% prevalence were detected solely by UDS: A98S (23%), E138A (21%) and V179I (25%). A significant difference in the DR interpretations for 19 antiretroviral drugs was observed between the UDS and Sanger sequencing methods. Y181C and T215Y were the most frequent mutations associated with interpretation differences. The major discrepancies between Sanger and UDS were primarily found at the 1% threshold in the three algorithms.</p></sec><sec sec-type="conclusions"><title>Conclusion</title><p>UDS was more sensitive than the standard Sanger sequencing. A combination of UDS and DeepChek® software for the interpretation of DR results saved a considerable amount of time and would help clinicians provide suitable treatments. A cut-off of 1% allowed a better characterization of the viral population by identifying additional resistance mutations and improving the DR interpretation.</p></sec>
Neonatal macaque model to study <italic>Mycobacterium tuberculosis</italic> infection in pediatric AIDS	Could not extract abstract	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Gauduin</surname><given-names>Marie-Claire</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A2"><name><surname>Cepeda</surname><given-names>Magdalena</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A3"><name><surname>Salas</surname><given-names>Mary</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A4"><name><surname>White</surname><given-names>Robert</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A5"><name><surname>de la Garza</surname><given-names>Melissa</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A6"><name><surname>Dick</surname><given-names>Edward J</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A7"><name><surname>Owston</surname><given-names>Michael</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A8"><name><surname>Armitige</surname><given-names>Lisa Y</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib>	BMC Infectious Diseases	<sec sec-type="intro"><title>Introduction</title><p>Tuberculosis (TB) is the leading cause of death in AIDS patients worldwide; the earliest opportunistic infection occurring in conjunction with HIV; an accelerant of HIV replication and immune system deterioration. Co-infection with <italic>Mycobacterium tuberculosis</italic> (Mtb) and HIV is an increasing global emergency. Little is known regarding the early events of Mtb infection in humans especially infants. There is an urgent need for a clinically relevant animal model that mimics TB disease in human children to better understand early immune responses, pathogenicity and disease progression in newborn/infants.</p></sec><sec sec-type="methods"><title>Methods</title><p>Aerosolized Mtb transmission was performed in neonatal macaques to mimic mother-to-child transmission and to: i. Characterize early TB-specific immune responses during acute Mtb-infection, ii. Investigate the dynamics of Mtb-specific T cell responses following early infection in infants; and, iii. Characterize the distribution and frequency of Mtb-specific responses in various host tissues. Six newborn macaques (6 weeks old) were infected via broncho or aerosol routes of infection with various Mtb doses (Erdman or Rh37rv strains). All clinical, immunologic, microbiologic, and pathologic events were assessed for 84 days post-infection.</p></sec><sec sec-type="results"><title>Results</title><p>Gross pathological abnormalities were observed as early as 35 et 42 days post infection including Ghon complex formation and granulomas in the lung. Caseous granulomas were also observed in the lungs at these early time-points, reflecting strong initial responses. Using ELISPOT assays, we found that IL-12 production correlated with early Mtb infection lesions seen by routine thoracic radiographs. Flow cytometry revealed robust granzyme B productions by NK8/NK cells that increased over time and transient Perforin productions that peak at day 56. Mtb-specific T cells responses were also detected and increased IL-12 and MIP-1β production in PBMC, Spleen and LN collected day 84.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>Overall, we successfully developed/optimized an aerosol neonatal macaque model that mimics clinical and bacteriological characteristics of early Mtb-infection as seen in human newborns/infants. This model represents a unique opportunity to characterize neonatal Mtb infection and, further understand all interactions between TB and HIV co-infection in pediatric AIDS and develop appropriate therapeutic interventions.</p></sec>
Persistent production of an integrase-deleted HIV-1 variant with no resistance mutation and wild-type proviral DNA in a treated patient	Could not extract abstract	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Trabaud</surname><given-names>MA</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A2"><name><surname>Cotte</surname><given-names>L</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A3"><name><surname>Saison</surname><given-names>J</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A4"><name><surname>Ramiere</surname><given-names>C</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A5"><name><surname>Ronfort</surname><given-names>C</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A6"><name><surname>Tardy</surname><given-names>JC</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A7"><name><surname>Andre</surname><given-names>P</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib>	BMC Infectious Diseases	<sec sec-type="intro"><title>Introduction</title><p>An HIV-1-infected patient with suppressed viremia for several years, in whom a variant carrying a deleted integrase (IN) gene, without reverse transcriptase (RT) or protease (PR) resistance mutations, emerged in the plasma and persisted is described.</p></sec><sec sec-type="materials\|methods"><title>Materials and methods</title><p>Viral load (VL) was tested by routine assays following manufacturer's instructions. RT, PR and IN genes were sequenced with the ANRS consensus techniques.</p><p>Nested PCRs with patient virus IN-specific primers and probes were developed to detect the deleted variant from plasma, blood lymphocytes, rectal biopsies, and sperm.</p></sec><sec sec-type="results"><title>Results</title><p>VL remained undetectable for more than two years under therapy, excepted for 1 observed blip. Thereafter HIV RNA increased slightly but persistently, fluctuating from 56 to 466 copies/ml during more than five years. By population sequencing a 38 nucleotides deletion was observed in the IN C-terminal domain (CTD) encoding sequence (residues 215 to 227 of the WT IN).</p><p>In plasma, the variant progressively emerged during therapy-induced virosuppression, HIV RNA being undetectable by routine viral load assay, and then persisted during detectable viremia. The WT IN, not detected by bulk sequencing, was present but at stable low level.</p><p>HIV DNA and RNA with WT IN were amplified from each cell extract. Detection of the deleted IN was a very rare event in blood and rectal cells.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>Although the reservoir of this virus is yet unknown it is likely in a tissue compartment and not predominantly in cells migrating through blood.</p><p>Questions arising from this case are how and why this certainly defective variant emerged during efficient virosuppression, and how it could be preferentially and stably produced.</p><p>As for interpretation of residual viremia found in long-term treated patients with undetectable viral load, virus production can originate from on-going replication, but the cells had to be continuously co-infected with the two viral forms, or synthesis by chronically infected cells. However, the defective virus production at detectable level suggests unusual circumstances.</p><p>Anyway, this case raises concern about the possible long term synthesis of defective viruses.</p><fig id="F1" position="float"><label>Figure 1</label><graphic xlink:href="1471-2334-14-S2-O12-1"/></fig></sec>
Competition between HIV-1-encoded RRE RNA and miRNA-TRBP interactions alters RNA interference activity and gene expression	Could not extract abstract	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Gatignol</surname><given-names>A</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A2"><name><surname>Daniels</surname><given-names>SM</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A3"><name><surname>Sinck</surname><given-names>L</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A4"><name><surname>Ward</surname><given-names>NJ</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A5"><name><surname>Melendez-Peña</surname><given-names>CE</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A6"><name><surname>Scarborough</surname><given-names>RJ</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A7"><name><surname>Azar</surname><given-names>I</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A8"><name><surname>Daher</surname><given-names>A</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A9"><name><surname>Pang</surname><given-names>KM</given-names></name><xref ref-type="aff" rid="I2">2</xref></contrib><contrib contrib-type="author" id="A10"><name><surname>Rossi</surname><given-names>JJ</given-names></name><xref ref-type="aff" rid="I2">2</xref></contrib>	BMC Infectious Diseases	<sec><title>Aim</title><p>RNA interference (RNAi) is a mechanism by which small double-stranded RNAs called micro(mi) or small interfering(si) RNAs bind messenger RNAs (mRNAs) to inhibit their expression. The mechanism involves the RNA-induced silencing complex (RISC) composed of Dicer, TRBP and Ago2 proteins in which TRBP loads miRNAs into the active complex [<xref ref-type="bibr" rid="B1">1</xref>]. Several mammalian viruses interfere with RNAi activity. Changes in miRNA and mRNA expression have been observed in patients infected by HIV-1, but the mechanisms are not understood. To explain part of the relationship between RNAi and HIV-1, we investigated the ability and the mechanism of the HIV-1-encoded RNA Rev-Response Element (RRE) to suppress RNAi.</p></sec><sec sec-type="methods"><title>Methods</title><p>We used a model based on miRNA Let7 activity on a reporter gene (RL or EGFP) linked to a complementary sequence (cLet7) to measure RNAi activity or its suppression. We used RNA-immunoprecipitation (IP) and gel mobility shift assays to compare TRBP binding to RRE or siRNAs. We studied RRE activity on RNAi in the context of the entire HIV-1, a lentiviral or an adenoviral vector.</p></sec><sec sec-type="results"><title>Results</title><p>We observed that RRE, acts as an RNAi suppressor with no modification of the endogenous RISC (Daniels et al., submitted). In contrast, RRE RNA displaces siARNs from TRBP, which suggests a change in miRNA incorporation into the RISC. RNAi remains functional in HIV-1 infected cells, whereas a lentiviral vector expressing RRE has a suppressive activity. The suppression is alleviated when Rev or GagPol is expressed. Adenovirus is known to be suppressed by RNAi and RRE reverses this inhibition as seen by increased viral replication.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>RRE is a new RNAi suppressor, which acts by competition with siRNA and miRNA for binding to TRBP and therefore incorporation into the RISC. This could explain in part the alteration of certain gene expression and modifications of the cell metabolism in patients with long-term HIV-1 infection.</p></sec>
IDO-induced immunosuppressive tryptophan catabolism following primary HIV infection	Could not extract abstract	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Jenabian</surname><given-names>Mohammad-Ali</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A2"><name><surname>Vyboh</surname><given-names>Kishanda</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A3"><name><surname>Kema</surname><given-names>Ido</given-names></name><xref ref-type="aff" rid="I3">3</xref></contrib><contrib contrib-type="author" id="A4"><name><surname>Kanagaratham</surname><given-names>Cynthia</given-names></name><xref ref-type="aff" rid="I2">2</xref></contrib><contrib contrib-type="author" id="A5"><name><surname>Radzioch</surname><given-names>Danuta</given-names></name><xref ref-type="aff" rid="I2">2</xref></contrib><contrib contrib-type="author" id="A6"><name><surname>Gilmore</surname><given-names>Norbert</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref></contrib><contrib contrib-type="author" id="A7"><name><surname>Ancuta</surname><given-names>Petronela</given-names></name><xref ref-type="aff" rid="I4">4</xref><xref ref-type="aff" rid="I5">5</xref></contrib><contrib contrib-type="author" id="A8"><name><surname>Tremblay</surname><given-names>Cécile</given-names></name><xref ref-type="aff" rid="I4">4</xref><xref ref-type="aff" rid="I5">5</xref></contrib><contrib contrib-type="author" id="A9"><name><surname>Routy</surname><given-names>Jean-Pierre</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref></contrib>	BMC Infectious Diseases	<sec><title>Background</title><p>We showed in cross-sectional studies that tryptophan (Trp) catabolism into kynurenine (Kyn) by IDO enzyme expressed by dendritic cells (DC) contributes to regulatory T-cells (Tregs) expansion and immune suppression in chronic HIV infection. We prospectively assessed Trp catabolism and anti-inflammatory response following primary HIV infection (PHI).</p></sec><sec sec-type="methods"><title>Methods</title><p>Plasma and Peripheral blood mononuclear cells (PBMCs) were longitudinally collected in 41 PHI patients (infection <90 days), 24 remained untreated (ART-naive) and 17 were ART-treated one year later. In addition, samples from elite controllers (EC, n=12) and healthy subjects (HS, n=12) were also assessed. IDO enzymatic activity marker (Kyn/Trp ratio) was measured by isotope dilution tandem mass spectrometry. IL-6, IL-18, TNF-α and IP-10 plasma levels were assessed by Luminex. Frequency of Tregs (CD4+CD25highCD127lowFOXP3high), CD11c+ myeloid DC (mDC) and CD123+ plasmacytoid DC (pDC) as well as HLA-DR/CD38 co-expression of on T-cells were assessed.</p></sec><sec sec-type="results"><title>Results</title><p>PHI patients had elevated Kyn/Trp ratio compared to HS and EC and further increased during the chronic phase, while normalized following ART. Accordingly, an increase of Treg frequency was observed at the baseline and continues to increase in the chronic phase only for those remaining untreated, when compared to HS and EC. Conversely, the frequency of mDC and pDC decreases over time only for those who remained untreated. Higher Kyn/Trp ratios were inversely correlated with the frequency of mDC and pDC at PHI and for those untreated. Importantly, the highest level of immune activation (HLA-DR+CD38+ CD8 T-cells) was observed during PHI followed by a decrease in chronic phase in ART-naïve and became comparable to EC and HS when receiving ART. Importantly, Kyn/Trp ratio was correlated with level of CD8 T-cell activation during PHI and for those who remained untreated. In line with this, positive correlations were observed between Kyn/Trp ratio and levels of IL-18 and TNF-α as well as markers of HIV disease progression IL-6 and IP-10.</p></sec><sec sec-type="conclusions"><title>Conclusion</title><p>The progressive increase of Kyn/Trp ratio observed in the chronic phase of HIV infection in contrast to decreased viral load and T-cell activation, support the contribution of tissue damage and/or myeloid inflammatory syndrome in addition to viral replication for the development of immunosuppression.</p></sec>
Carcinovic cohort: prognostic factors of death in HIV/HCV coinfected patients with hepatocellular carcinoma (HCC)	Could not extract abstract	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Gelu-Simeon</surname><given-names>M</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A2"><name><surname>Lewin</surname><given-names>M</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A3"><name><surname>Sobesky</surname><given-names>R</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A4"><name><surname>Ostos</surname><given-names>M</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A5"><name><surname>Bayan</surname><given-names>T</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A6"><name><surname>Boufassa</surname><given-names>F</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A7"><name><surname>Meyer</surname><given-names>L</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A8"><name><surname>Persoz</surname><given-names>A</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A9"><name><surname>Teicher</surname><given-names>E</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A10"><name><surname>Fontaine</surname><given-names>H</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A11"><name><surname>Salmon-Céron</surname><given-names>D</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A12"><name><surname>Seror</surname><given-names>O</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A13"><name><surname>Trinchet</surname><given-names>J-C</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A14"><name><surname>Duclos-Vallée</surname><given-names>J-C</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib>	BMC Infectious Diseases	<sec><title>Background and aim</title><p>We have previously reported a more advanced radiological presentation in HIV+/HCV+ than HIV-/HCV+ patients (pts). The aim of our study was to define prognostic factors of death in HIV+/HCV+ pts with HCC.</p></sec><sec sec-type="methods"><title>Methods</title><p>Cases of HCC in HIV+/HCV+ pts were obtained from the 3 ANRS Prethevic, HepaVih and CirVir cohorts. Imaging was reviewed according to EASL criteria.</p></sec><sec sec-type="results"><title>Results</title><p>Fifty HIV+/HCV+ coinfected pts (n=44 men (88%), median age 50 years [40-74], median CD4 cell count 334/mm3 [58-1621], n=28 Child A cirrhosis (60%)) developed HCC. Thirty-one (63%) pts presented cirrhosis decompensation before HCC diagnosis. At HCC diagnosis, median serum aFP was 20.4 [1.9-198,900] ng/ml, 38 (76%) pts had a nodular tumor (median main diameter 23.5 [11-70] cm) and 12 (24%) pts an infiltrating form (62.5 [10-130] cm), p=0.007. Tumor portal thrombosis was diagnosed in 14 (28%) pts. A curative or a palliative procedure was further performed in 22 (44%) pts and 20 (40%) pts, respectively. The 2-years and 4-years overall survival rates were 51% and 28%, respectively. Age (p=0.0005), infiltrating or nodular tumor (p= 0.0009) and tumor portal thrombosis (p=0.004) were associated to survival. In a Cox model, two prognostic factors of deaths were found: prior episode of cirrhosis decompensation (aRR 11.43 [3.01-43.34], p=0.0003) and tumor portal thrombosis (aRR 4.66 [1.19-18.27], p=0.03), adjusted on age, CD4 cell count and the therapeutic strategy for HCC.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>Cirrhosis decompensation and tumor portal thrombosis significantly impact the survival of HIV+/HCV+ pts with HCC. Our results suggest new rules of screening HCC in HIV+/HCV+ pts with advanced liver disorders.</p></sec>
Rational Design of Superoxide Dismutase (SOD) Mimics: The Evaluation of the Therapeutic Potential of New Cationic Mn Porphyrins with Linear and Cyclic Substituents	<p content-type="toc-graphic"><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-01329p_0011" id="ab-tgr1"/></p><p>Our goal herein has been to gain further insight into the parameters which control porphyrin therapeutic potential. Mn porphyrins (MnTnOct-2-PyP<sup>5+</sup>, MnTnHexOE-2-PyP<sup>5+</sup>, MnTE-2-PyPhP<sup>5+</sup>, and MnTPhE-2-PyP<sup>5+</sup>) that bear the same positive charge and same number of carbon atoms at <italic>meso</italic> positions of porphyrin core were explored. The carbon atoms of their <italic>meso</italic> substituents are organized to form either linear or cyclic structures of vastly different redox properties, bulkiness, and lipophilicities. These Mn porphyrins were compared to frequently studied compounds, MnTE-2-PyP<sup>5+</sup>, MnTE-3-PyP<sup>5+</sup>, and MnTBAP<sup>3–</sup>. All Mn(III) porphyrins (MnPs) have metal-centered reduction potential, <italic>E</italic><sub>1/2</sub> for Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple, ranging from −194 to +340 mV versus NHE, log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) from 3.16 to 7.92, and log <italic>k</italic><sub>red</sub>(ONOO<sup>–</sup>) from 5.02 to 7.53. The lipophilicity, expressed as partition between n-octanol and water, log <italic>P</italic><sub>OW</sub>, was in the range −1.67 to −7.67. The therapeutic potential of MnPs was assessed via: (i) <italic>in vitro</italic> ability to prevent spontaneous lipid peroxidation in rat brain homogenate as assessed by malondialdehyde levels; (ii) <italic>in vivo</italic> O<sub>2</sub><sup>•–</sup> specific assay to measure the efficacy in protecting the aerobic growth of SOD-deficient <italic>Saccharomyces cerevisiae</italic>; and (iii) aqueous solution chemistry to measure the reactivity toward major <italic>in vivo</italic> endogenous antioxidant, ascorbate. Under the conditions of lipid peroxidation assay, the transport across the cellular membranes, and in turn shape and size of molecule, played no significant role. Those MnPs of <italic>E</italic><sub>1/2</sub> ∼ +300 mV were the most efficacious, significantly inhibiting lipid peroxidation in 0.5–10 μM range. At up to 200 μM, MnTBAP<sup>3–</sup> (<italic>E</italic><sub>1/2</sub> = −194 mV vs NHE) failed to inhibit lipid peroxidation, while MnTE-2-PyPhP<sup>5+</sup> with 129 mV more positive <italic>E</italic><sub>1/2</sub> (−65 mV vs NHE) was fully efficacious at 50 μM. The <italic>E</italic><sub>1/2</sub> of Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple is proportional to the log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>), <italic>i.e</italic>., the SOD-like activity of MnPs. It is further proportional to <italic>k</italic><sub><italic>r</italic>ed</sub>(ONOO<sup>–</sup>) and the ability of MnPs to prevent lipid peroxidation. In turn, the inhibition of lipid peroxidation by MnPs is also proportional to their SOD-like activity. In an <italic>in vivo S. cerevisiae</italic> assay, however, while <italic>E</italic><sub>1/2</sub> predominates, lipophilicity significantly affects the efficacy of MnPs. MnPs of similar log <italic>P</italic><sub>OW</sub> and <italic>E</italic><sub>1/2</sub>, that have linear alkyl or alkoxyalkyl pyridyl substituents, distribute more easily within a cell and in turn provide higher protection to <italic>S. cerevisiae</italic> in comparison to MnP with bulky cyclic substituents. The bell-shape curve, with MnTE-2-PyP<sup>5+</sup> exhibiting the highest ability to catalyze ascorbate oxidation, has been established and discussed. Our data support the notion that the SOD-like activity of MnPs parallels their therapeutic potential, though species other than O<sub>2</sub><sup>•–</sup>, such as peroxynitrite, H<sub>2</sub>O<sub>2</sub>, lipid reactive species, and cellular reductants, may be involved in their mode(s) of action(s).</p>	<contrib contrib-type="author" id="ath1"><name><surname>Tovmasyan</surname><given-names>Artak</given-names></name><xref rid="aff1" ref-type="aff">†</xref></contrib><contrib contrib-type="author" id="ath2"><name><surname>Carballal</surname><given-names>Sebastian</given-names></name><xref rid="aff2" ref-type="aff">§</xref></contrib><contrib contrib-type="author" id="ath3"><name><surname>Ghazaryan</surname><given-names>Robert</given-names></name><xref rid="aff3" ref-type="aff">∥</xref></contrib><contrib contrib-type="author" id="ath4"><name><surname>Melikyan</surname><given-names>Lida</given-names></name><xref rid="aff3" ref-type="aff">∥</xref></contrib><contrib contrib-type="author" id="ath5"><name><surname>Weitner</surname><given-names>Tin</given-names></name><xref rid="aff1" ref-type="aff">†</xref></contrib><contrib contrib-type="author" id="ath6"><name><surname>Maia</surname><given-names>Clarissa G. C.</given-names></name><xref rid="aff4" ref-type="aff">⊥</xref></contrib><contrib contrib-type="author" id="ath7"><name><surname>Reboucas</surname><given-names>Julio S.</given-names></name><xref rid="aff4" ref-type="aff">⊥</xref></contrib><contrib contrib-type="author" id="ath8"><name><surname>Radi</surname><given-names>Rafael</given-names></name><xref rid="aff2" ref-type="aff">§</xref></contrib><contrib contrib-type="author" id="ath9"><name><surname>Spasojevic</surname><given-names>Ivan</given-names></name><xref rid="aff1" ref-type="aff">‡</xref></contrib><contrib contrib-type="author" id="ath10"><name><surname>Benov</surname><given-names>Ludmil</given-names></name><xref rid="aff5" ref-type="aff">#</xref></contrib><contrib contrib-type="author" corresp="yes" id="ath11"><name><surname>Batinic-Haberle</surname><given-names>Ines</given-names></name><xref rid="cor1" ref-type="other">*</xref><xref rid="aff1" ref-type="aff">†</xref></contrib><aff id="aff1"><sup>†</sup>Departments of Radiation Oncology and <sup>‡</sup>Medicine, <institution>Duke University Medical Center</institution>, Research Drive, 281b MSRB I, Durham, North Carolina 27710, <country>United States</country></aff><aff id="aff2"><label>§</label>Departamento de Bioquímica and Center for Free Radical and Biomedical Research, Facultad de Medicina, <institution>Universidad de la República</institution>, Montevideo, <country>Uruguay</country></aff><aff id="aff3"><label>∥</label>Department of Organic Chemistry, Faculty of Pharmacy, <institution>Yerevan State Medical University</institution>, Yerevan, <country>Armenia</country></aff><aff id="aff4"><label>⊥</label>Departamento de Quimica, CCEN, <institution>Universidade Federal de Paraiba</institution>, Joao Pessoa, PB 58051-900, <country>Brazil</country></aff><aff id="aff5"><label>#</label>Department of Biochemistry, Faculty of Medicine, <institution>Kuwait University</institution>, Kuwait City, <country>Kuwait</country></aff>	Inorganic Chemistry	<sec sec-type="intro" id="sec1"><title>Introduction</title><p>Our continuous goal has been to learn how to improve the therapeutic potential of porphyrin-based SOD mimics for the treatment of disorders with perturbed cellular redox environment, commonly described as oxidative stress. While maintaining the most appropriate thermodynamics and kinetics for SOD-like activity, the efforts have recently been directed toward the increase in the biodistribution of SOD mimics and decrease in their toxicity. The structure–activity relationship (SAR), which has guided us in our efforts to improve the drug quality,<sup><xref ref-type="bibr" rid="ref1">1</xref>,<xref ref-type="bibr" rid="ref2">2</xref></sup> correlates the thermodynamic (metal-centered reduction potential, <italic>E</italic><sub>1/2</sub>, for Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple) and kinetic properties of Mn(III) porphyrins (MnPs), log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) (O<sub>2</sub><sup>•–</sup>, superoxide). The <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) describes the ability of MnP to catalyze O<sub>2</sub><sup>•–</sup> dismutation to O<sub>2</sub> and H<sub>2</sub>O<sub>2</sub>. SAR is universally valid, not only for metalloporphyrins but for other redox-active drugs also.<sup><xref ref-type="bibr" rid="ref2">2</xref>,<xref ref-type="bibr" rid="ref3">3</xref></sup> Further, <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) parallels the ability of MnPs to reduce peroxynitrite, described by the rate constant for ONOO<sup>–</sup> (peroxynitrite) reduction, <italic>k</italic><sub>red</sub>(ONOO<sup>–</sup>).<sup><xref ref-type="bibr" rid="ref4">4</xref></sup> Both properties are controlled by the electron-deficiency of a metal site which favors exchanging electrons with O<sub>2</sub><sup>•–</sup> (reducing and oxidizing it during dismutation process) and binding of electron-rich ONOO<sup>–</sup> with its subsequent reduction to either <sup>•</sup>NO<sub>2</sub> (one-electronically) or NO<sub>2</sub><sup>–</sup> (two-electronically).<sup><xref ref-type="bibr" rid="ref4">4</xref>−<xref ref-type="bibr" rid="ref6">6</xref></sup> We have further shown that such property of the metal site also favors reactions with other electron-rich nucleophiles such as ClO<sup>–</sup> (deprotonated hypochlorite),<sup><xref ref-type="bibr" rid="ref7">7</xref></sup> HO<sub>2</sub><sup>–</sup> (a deprotonated reactive species of H<sub>2</sub>O<sub>2</sub>), lipid radicals,<sup><xref ref-type="bibr" rid="ref8">8</xref>,<xref ref-type="bibr" rid="ref9">9</xref></sup> CO<sub>3</sub><sup>•–</sup>,<sup><xref ref-type="bibr" rid="ref4">4</xref></sup> ascorbate, HA<sup>–</sup> (monodeprotonated ascorbic acid), and deprotonated thiols, RS<sup>–</sup>.<sup><xref ref-type="bibr" rid="ref10">10</xref>−<xref ref-type="bibr" rid="ref12">12</xref></sup> The reaction of MnPs with simple and protein thiols as well as with ascorbate coupled to peroxide production seems to be heavily involved in their mechanism(s) of action(s).<sup><xref ref-type="bibr" rid="ref2">2</xref>,<xref ref-type="bibr" rid="ref10">10</xref></sup></p><p>With the goal to enhance the biodistribution of MnPs, we modified the original structure of MnTE-2-PyP<sup>5+</sup> (AEOL10113, Mn(III) <italic>meso</italic>-tetrakis(<italic>N</italic>-ethylpyridinium-2-yl)porphyrin) and synthesized a first generation of lipophilic analogs, via lengthening the alkyl chains of MnTE-2-PyP<sup>5+</sup> to MnTnOct-2-PyP<sup>5+</sup> (Mn(III) <italic>meso</italic>-tetrakis(<italic>N</italic>-n-octylpyridinium-2-yl)porphyrin).<sup><xref ref-type="bibr" rid="ref13">13</xref></sup> MnTnHex-2-PyP<sup>5+</sup> (Mn(III) <italic>meso</italic>-tetrakis(<italic>N</italic>-<italic>n</italic>-hexylpyridinium-2-yl)porphyrin) has been a well-explored lipophilic analog with much higher brain and mitochondrial distribution than MnTE-2-PyP<sup>5+</sup>.<sup><xref ref-type="bibr" rid="ref14">14</xref>,<xref ref-type="bibr" rid="ref15">15</xref></sup> Yet, its toxicity at higher concentration and prolonged administration may limit its use. We thus designed and characterized MnTnBuOE-2-PyP<sup>5+</sup> (BMX-001; Mn(III) <italic>meso</italic>-tetrakis(<italic>N</italic>-(2′-<italic>n</italic>-butoxyethyl)pyridinium-2-yl)porphyrin), which has 4–5-fold reduced toxicity relative to MnTnHex-2-PyP<sup>5+</sup>, while lipophilicity and redox-activity have not been compromised.<sup><xref ref-type="bibr" rid="ref16">16</xref></sup></p><p>Herein we continued with the rational design of MnPs. New Mn porphyrins were synthesized and compared to MnTnOct-2-PyP<sup>5+</sup> and several other compounds (Figures <xref rid="fig1" ref-type="fig">1</xref> and <xref rid="fig2" ref-type="fig">2</xref>) mostly studied by us and others, MnTE-2-PyP<sup>5+</sup>, MnTE-3-PyP<sup>5+</sup> (Mn(III) <italic>meso</italic>-tetrakis(<italic>N</italic>-ethylpyridinium-3-yl)porphyrin), and MnTBAP<sup>3–</sup> (Mn(III) <italic>meso</italic>-tetrakis(4-carboxylatophenyl)porphyrin).<sup><xref ref-type="bibr" rid="ref2">2</xref>,<xref ref-type="bibr" rid="ref17">17</xref></sup> While of entirely different redox properties, both an SOD mimic, MnTE-2-PyP<sup>5+</sup>, and a non-SOD mimic, MnTBAP<sup>3–</sup>, reportedly exhibit beneficial effects in <italic>in vitro</italic> and <italic>in vivo</italic> models of numerous oxidative stress-related disorders, such as stroke, cancer, lung diseases, radiation injuries, spinal cord injury, Alzheimer disease, cardiac injuries, pain, and morphine tolerance and autoimmune diseases, some of which are shown in Figure <xref rid="fig1" ref-type="fig">1</xref>.<sup><xref ref-type="bibr" rid="ref18">18</xref>−<xref ref-type="bibr" rid="ref68">68</xref></sup></p><fig id="fig1" position="float"><label>Figure 1</label><caption><p>Structures of MnTBAP<sup>3-</sup>,<sup><xref ref-type="bibr" rid="ref18">18</xref>−<xref ref-type="bibr" rid="ref42">42</xref></sup> and <italic>ortho</italic> (2) and <italic>meta</italic> (3) isomers, MnTE-2(and 3)-PyP<sup>5+</sup>.<sup><xref ref-type="bibr" rid="ref43">43</xref>−<xref ref-type="bibr" rid="ref68">68</xref></sup> Also listed are their <italic>in vivo</italic> efficacy studies.</p></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-01329p_0001" id="gr1" position="float"/></fig><p>We have been puzzled with the therapeutic efficacy of the MnTBAP<sup>3–</sup> [see also <xref rid="sec3" ref-type="other">Results and Discussion</xref>]. On the basis of our present knowledge, this compound has inferior redox properties (<italic>E</italic><sub>1/2</sub> = −194 mV vs NHE), relative to MnTE-2-PyP<sup>5+</sup> (<italic>E</italic><sub>1/2</sub> = +228 mV vs NHE) and thus does not favor interactions with biological targets. Most of the reactive species are anionic and would disfavor interacting with anionic MnTBAP<sup>3–</sup> on electrostatic grounds.<sup><xref ref-type="bibr" rid="ref69">69</xref></sup> Still strongly oxidizing species such as ONOO<sup>–</sup> and CO<sub>3</sub><sup>•–</sup> are able to oxidize it.<sup><xref ref-type="bibr" rid="ref4">4</xref>,<xref ref-type="bibr" rid="ref5">5</xref>,<xref ref-type="bibr" rid="ref9">9</xref></sup> Despite claims,<sup><xref ref-type="bibr" rid="ref70">70</xref></sup> MnTBAP<sup>3–</sup> is neither reactive toward O<sub>2</sub><sup>•–</sup> nor to H<sub>2</sub>O<sub>2</sub>. Recent data indicate that its RNS-related chemistry may account for its biological effects;<sup><xref ref-type="bibr" rid="ref71">71</xref></sup> the neutrality of <sup>•</sup>NO or HNO would work in favor of such reactions.<sup><xref ref-type="bibr" rid="ref72">72</xref></sup> Finally, its negative charge would not facilitate its transport across anionic phospholipid membranes. The fact that impure preparations of MnTBAP<sup>3–</sup>, provided by several commercial sources, were often used without prior characterization and purification has complicated things further.<sup><xref ref-type="bibr" rid="ref69">69</xref>,<xref ref-type="bibr" rid="ref73">73</xref></sup> Still the abundance of published data, including a few of our studies, indicates that under certain conditions MnTBAP<sup>3–</sup> is efficacious.<sup><xref ref-type="bibr" rid="ref5">5</xref>,<xref ref-type="bibr" rid="ref9">9</xref>,<xref ref-type="bibr" rid="ref18">18</xref>−<xref ref-type="bibr" rid="ref42">42</xref>,<xref ref-type="bibr" rid="ref72">72</xref></sup> Recently, a manuscript was published where PEG-ylated amid of MnTBAP<sup>3–</sup> was synthesized and characterized.<sup><xref ref-type="bibr" rid="ref74">74</xref></sup> Such derivatization removed the unfavorable electron-donating effect of COO<sup>–</sup> groups upon Mn site. In turn, the <italic>E</italic><sub>1/2</sub> is nearly 200 mV more positive relative to MnTBAP<sup>3–</sup>. Along with the improved electrostatics, this modification increased log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) from 3.16 to 5.6. The impact of such derivatization agrees well with our data on the contribution of electrostatics in O<sub>2</sub><sup>•–</sup> dismutation.<sup><xref ref-type="bibr" rid="ref75">75</xref></sup> In order to gain further insight into the possible therapeutic effects of MnTBAP<sup>3–</sup>, we have used it in all studies performed herein.</p><fig id="fig2" position="float"><label>Figure 2</label><caption><p>Impact of structural features of MnTnOct-2-PyP<sup>5+</sup> and three new Mn porphyrins (MnPs) on their <italic>in vitro</italic> and <italic>in vivo</italic> therapeutic potential. The figure illustrates which properties of MnPs were studied herein with a goal to (i) further our knowledge on their impact on the therapeutic potential of redox-active drugs, and in turn (ii) facilitate drug development. Metal-centered reduction potential, <italic>E</italic><sub>1/2</sub> of Mn<sup>III</sup>P/Mn<sup>II</sup>P, controls the rate constant for the catalysis of O<sub>2</sub><sup>•–</sup>, <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>), rate constant for the peroxynitrite reduction, <italic>k</italic><sub>red</sub>(ONOO<sup>–</sup>), as well as the ability of MnP to catalyze ascorbate oxidation to ascorbyl radical A<sup>•</sup>. The <italic>in vitro</italic> consequences of appropriate thermodynamics were also witnessed in the lipid peroxidation of rat brain homogenate. This is so because the reduction of highly reactive species, such as ONOO<sup>–</sup> and lipid reactive species, involves their binding to Mn site in the first step. Binding is controlled by electron-deficiency of porphyrin and its Mn site and could be best described by the protonation equilibria of porphyrin inner pyrrolic nitrogens<sup><xref ref-type="bibr" rid="ref76">76</xref></sup> and axial waters,<sup><xref ref-type="bibr" rid="ref4">4</xref></sup> which in turn control the <italic>E</italic><sub>1/2</sub> of Mn<sup>III</sup>P/Mn<sup>II</sup>P. With <italic>E</italic><sub>1/2</sub> value beyond 0 mV vs NHE, the fair deficiency in electron density of the metal site is indicated which in turn suggests the high affinity of Mn toward binding of an electron-rich ligand, such as ONOO<sup>–</sup> or lipid reactive species. Ligand binding is followed by Mn<sup>III</sup>P oxidation to O=Mn<sup>IV</sup>P. Therefore, the <italic>E</italic><sub>1/2</sub> of Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple correlates well with rates of reactions involving the O=Mn<sup>IV</sup>P/Mn<sup>III</sup>P redox couple. The <italic>E</italic><sub>1/2</sub> of O=Mn<sup>IV</sup>P/Mn<sup>III</sup>P redox couple is similar for a variety of different Mn and Fe porphyrins, implying that the ligand (such as ONOO<sup>−</sup>) binding is a rate-limiting step in metal oxidation and ligand reduction (see also <xref rid="sec3" ref-type="other">Results and Discussion</xref>). The other major property that controls the therapeutic potential of MnP is its lipophilicity, and it was herein explored in aerobic growth of SOD-deficient yeast <italic>S. cerevisiae</italic>.</p></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-01329p_0002" id="gr2" position="float"/></fig><p>From a therapeutic point of view, if the drug is efficacious it may not quite matter what exactly it is doing <italic>in vivo</italic>. This fact may not preclude its clinical development. Yet, understanding the drug biology matters largely if it is used to clarify the metabolic pathways involved in certain models of diseases. Importantly, it provides us with valuable feedback for improving the design of not only MnPs, but also other redox-active drugs. With this in mind, we are continuing here with rational design of redox-active SOD mimics. The compounds of the same charge and same number of carbon atoms, organized in different ways, are synthesized and characterized as indicated in Figure <xref rid="fig2" ref-type="fig">2</xref>. These MnPs have different redox properties, bulkiness (size and shape), and lipophilicities. They were tested in an <italic>in vitro</italic> model of lipid peroxidation and in an <italic>in vivo</italic> O<sub>2</sub><sup>•–</sup>-specific model of aerobic growth of <italic>S. cerevisiae.</italic> This model has over the years unambiguously identified the clinical drug candidates.<sup><xref ref-type="bibr" rid="ref2">2</xref>,<xref ref-type="bibr" rid="ref67">67</xref></sup> Upon entering the cell, MnPs encounter ascorbate due to its high <italic>in vivo</italic> abundance. Thus, cycling with ascorbate seems to be heavily involved in their actions. Moreover, the combination of ascorbate and MnP holds a promising therapeutic modality for cancer treatment.<sup><xref ref-type="bibr" rid="ref2">2</xref>,<xref ref-type="bibr" rid="ref77">77</xref>,<xref ref-type="bibr" rid="ref78">78</xref></sup> Therefore, the reactivity of MnPs toward ascorbate has been explored also.</p></sec><sec id="sec2"><title>Experimental Section</title><sec id="sec2.1"><title>General</title><p><italic>meso</italic>-Tetrakis(2-<italic>N</italic>-pyridyl)porphyrin (H<sub>2</sub>T-2-PyP) and <italic>meso</italic>-tetrakis(3-<italic>N</italic>-pyridyl)porphyrin (H<sub>2</sub>T-3-PyP) were purchased from Frontier Scientific. Ethyl <italic>p</italic>-toluenesulfonate (98%) was from Sigma-Aldrich. The <italic>n</italic>-octyl <italic>p</italic>-toluenesulfonate and methyl-tri-n-octylammonium chloride (>95%) were from TCI America. MnCl<sub>2</sub>·4H<sub>2</sub>O (99.7%) was supplied by J. T. Baker, FeCl<sub>2</sub> (98%) was from Sigma-Aldrich, and NH<sub>4</sub>PF<sub>6</sub> (99.99%) was from GFS chemicals. Anhydrous diethyl ether and acetone were from EMD chemicals, while dichloromethane, chloroform, acetonitrile, EDTA, and KNO<sub>3</sub> were purchased from Mallinckrodt. Anhydrous <italic>N,N</italic>-dimethylformamide (DMF) of 99.8% purity (kept over 4-Å molecular sieves) and plastic-backed silica gel TLC plates (Z122777-25EA) were from Sigma-Aldrich. Xanthine, equine ferricytochrome <italic>c</italic> (lot 7752), and (+)-sodium <sc>l</sc>-ascorbate (>98%) were from Sigma, whereas xanthine oxidase was prepared by R. Wiley.<sup><xref ref-type="bibr" rid="ref1">1</xref></sup> Triethylamine (Et<sub>3</sub>N) of >99.5% purity was obtained from Thermo Scientific Pierce. All chemicals were used as received without further purification. The <sup>1</sup>H NMR spectra were recorded on a spectrometer “Mercury Varian 300” with deuterated chloroform as solvent.</p></sec><sec id="sec2.2"><title>Synthesis of <italic>meso</italic>-Tetrakis(<italic>N</italic>-substituted pyridinium-2-yl)porphyrins</title><p>The general synthetic procedure for <italic>meso</italic>-tetrakis(<italic>N</italic>-substituted pyridinium-2-yl)porphyrins and their Mn complexes is shown in Figure <xref rid="fig3" ref-type="fig">3</xref>. The synthesis, isolation, purification, and characterization of Mn porphyrins, MnTE-2-PyPCl<sub>5</sub>, MnTE-3-PyPCl<sub>5</sub>, MnTnOct-2-PyPCl<sub>5</sub>, and MnTBAP<sup>3–</sup>, were performed as described earlier.<sup><xref ref-type="bibr" rid="ref76">76</xref>,<xref ref-type="bibr" rid="ref79">79</xref></sup> The appropriate tosylates, phenylethyl <italic>p</italic>-toluenesulfonate and 2-n-hexoxyethyl <italic>p</italic>-toluenesulfonate, were obtained, purified, and characterized according to the methods earlier reported for analogous compounds.<sup><xref ref-type="bibr" rid="ref80">80</xref></sup> The synthesis of new porphyrinic ligands H<sub>2</sub>TPhE-2-PyPCl<sub>4</sub> (<italic>meso</italic>-tetrakis(<italic>N</italic>-(2′-phenylethyl)pyridinium-2-yl)porphyrin tetrachloride) and H<sub>2</sub>TnHexOE-2-PyPCl<sub>4</sub> (<italic>meso</italic>-tetrakis(<italic>N</italic>-(2′-n-hexoxyethyl)pyridinium-2-yl)porphyrin tetrachloride)) and their Mn complexes is illustrated in Figure <xref rid="fig3" ref-type="fig">3</xref>.</p><fig id="fig3" position="float"><label>Figure 3</label><caption><p>Synthesis of new porphyrinic ligands, H<sub>2</sub>TPhE-2-PyPCl<sub>4</sub> and H<sub>2</sub>TnHexOE-2-PyPCl<sub>4</sub>, and their Mn complexes, MnTPhE-2-PyPCl<sub>5</sub> and MnTnHexOE-2-PyPCl<sub>5</sub>.</p></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-01329p_0003" id="gr3" position="float"/></fig><sec id="sec2.2.1"><title>H<sub>2</sub>TPhE-2-PyPCl<sub>4</sub>, <italic>meso</italic>-Tetrakis(<italic>N</italic>-(2′-phenylethyl)pyridinium-2-yl)porphyrin Tetrachloride</title><p>H<sub>2</sub>T-2-PyP (100 mg; 0.162 mmol) was dissolved in 4.6 mL of DMF preheated for 10 min at 115 °C. To the resulting solution was added the 9 g (0.032 mol) of phenylethyl <italic>p</italic>-toluenesulfonate. The course of <italic>N</italic>-quaternization was followed by thin-layer chromatography (TLC) on silica gel plates using acetonitrile/KNO<sub>3</sub>(sat)/water = 8/1/1 as a mobile phase. Also, methanol/chloroform (1/4) solvent system has been used to monitor the reaction progress. The reaction was completed within 25 h. Porphyrin was precipitated from the reaction mixture by diethyl ether, filtered, and washed with diethyl ether (5 × 30 mL). The porphyrin tosylate was then dissolved in 100 mL of hot water and precipitated as the PF<sub>6</sub><sup>–</sup> salt with saturated aqueous solution of NH<sub>4</sub>PF<sub>6</sub>. The precipitate was thoroughly washed with diethyl ether. The dried precipitate was then dissolved in acetone, solution filtered and porphyrin precipitated from it as a chloride salt with saturated acetone solution of methyl-tri-n-octylammonium chloride. The precipitate was washed with acetone and dissolved in water. The double precipitation was repeated once again to ensure the highest purity of preparation. The porphyrin was dried in vacuum oven in the form of Cl<sup>–</sup> salt. Yield (calculated based on elemental analysis): 180 mg (94.3%).</p></sec><sec id="sec2.2.2"><title>MnTPhE-2-PyPCl<sub>5</sub>, Mn(III) <italic>meso</italic>-Tetrakis(<italic>N</italic>-(2′-phenylethyl)pyridinium-2-yl)porphyrin Pentachloride</title><p>The pH of 40 mL aqueous solution of H<sub>2</sub>TPhE-2-PyPCl<sub>4</sub> (50 mg) was adjusted to 10.9 (with 1 M NaOH), and a 20-fold molar excess of MnCl<sub>2</sub>·4H<sub>2</sub>O (0.847 mmol; 167.4 mg) was added into the solution at 25 °C while stirring. The pH of the solution dropped to 7.6. The stirring was continued at 100 °C for 3.5 h until metalation was completed. The course of metalation was followed on silica gel TLC plates using acetonitrile/KNO<sub>3</sub>(sat)/water = 8/1/1 as a mobile phase. The pH of the solution was periodically adjusted to 7.2. Additionally, the course of metalation was monitored as a disappearance of porphyrin ligand fluorescence under UV light at ∼350 nm. The porphyrin solution was filtered first through coarse and then through fine filter paper. The MnP was precipitated as a PF<sub>6</sub><sup>–</sup> salt with saturated aqueous solution of NH<sub>4</sub>PF<sub>6</sub>. The precipitate was thoroughly washed with diethyl ether. The dried precipitate was then dissolved in acetone, filtered, and precipitated as the chloride salt with saturated acetone solution of methyl-tri-n-octylammonium chloride. The precipitate was washed with acetone and dissolved in water. The double precipitation was repeated once again to ensure the highest purity of porphyrin and complete removal of free manganese species.</p></sec><sec id="sec2.2.3"><title>H<sub>2</sub>TnHexOE-2-PyPCl<sub>4</sub>, <italic>meso</italic>-Tetrakis(<italic>N</italic>-(2′-n-hexoxyethyl)pyridinium-2-yl)porphyrin Tetrachloride</title><p>The synthesis was similar to the one described for H<sub>2</sub>TPhE-2-PyPCl<sub>4</sub>. Briefly, to a 70 mg portion of H<sub>2</sub>T-2-PyP in 4 mL of DMF, preheated for ∼5 min at 115 °C, was added the 8.5 g of 2-n-hexoxyethyl <italic>p</italic>-toluenesulfonate (0.028 mol). The reaction was completed within 48 h. The porphyrin was isolated as described for H<sub>2</sub>TPhE-2-PyPCl<sub>4</sub>.</p></sec><sec id="sec2.2.4"><title>MnTnHexOE-2-PyPCl<sub>5</sub>, Mn(III) <italic>meso</italic>-Tetrakis(<italic>N</italic>-(2′-n-hexoxyethyl)pyridinium-2-yl)porphyrin Pentachloride</title><p>The metalation was similar to the procedure described for the MnTPhE-2-PyPCl<sub>5</sub>. Briefly, the pH of 80 mL of H<sub>2</sub>TnHexOE-2-PyPCl<sub>4</sub> aqueous solution (100 mg, 0.078 mmol) was adjusted to 10.9, and a 40-fold excess of MnCl<sub>2</sub> (620 mg, 3.1 mmol) was added into the solution while stirring at 25 °C for 2.5 h. The porphyrin was isolated and purified in quantitative yield as described for MnTPhE-2-PyPCl<sub>5</sub>.</p></sec></sec><sec id="sec2.3"><title>Synthesis of <italic>meso</italic>-Tetrakis(phenyl-4-(2′-<italic>N</italic>-pyridyl))porphyrins</title><p>Synthesis of new porphyrinic ligands H<sub>2</sub>T-2-PyPhP (<italic>meso</italic>-tetrakis(phenyl-4-(2′-<italic>N</italic>-pyridyl))porphyrin), and H<sub>2</sub>TE-2-PyPhPCl<sub>4</sub> (<italic>meso</italic>-tetrakis(phenyl-4-(<italic>N</italic>-ethylpyridinium-2′-yl))porphyrin tetrachloride), and MnTE-2-PyPhPCl<sub>5</sub> (Mn(III) <italic>meso</italic>-tetrakis(phenyl-4-(<italic>N</italic>-ethylpyridinium-2′-yl))porphyrin pentachloride) was performed as illustrated in Figure <xref rid="fig4" ref-type="fig">4</xref>.</p><fig id="fig4" position="float"><label>Figure 4</label><caption><p>Synthesis of new porphyrinic ligands, H<sub>2</sub>T-2-PyPhP and H<sub>2</sub>TE-2-PyPhPCl<sub>4</sub>, and a Mn complex, MnTE-2-PyPhPCl<sub>5</sub>.</p></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-01329p_0004" id="gr4" position="float"/></fig><sec id="sec2.3.1"><title>H<sub>2</sub>T-2-PyPhP, <italic>meso</italic>-Tetrakis(phenyl-4-(2′-<italic>N</italic>-pyridyl))porphyrin</title><p>4-(2-Pyridyl)benzaldehyde (5 g, 0.027 mol) was added to a boiling propionic acid (100 mL). Pyrrole (1.85 g, 0.027 mol) was added to the reaction mixture and was stirred for 45 min. The solution was stirred for another 2 h at room temperature and was left overnight in dark. The precipitate formed was filtered, washed with diluted aqueous solution of NaHCO<sub>3</sub>, cold water, hot water, cold water and finally with small portions of methanol, and was left overnight to dry. The obtained violet crystals were dissolved in chloroform and were purified by column chromatography (absorbent, alumina; eluent, chloroform). The solvent was evaporated under reduced pressure, and the porphyrin was air-dried. Yield: 1.1 g (17.47%). H<sub>2</sub>T-2-PyPhP porphyrin (C<sub>64</sub>H<sub>42</sub>N<sub>8</sub>) <italic>M</italic><sub>r</sub> = 923.1. <sup>1</sup>H NMR (300 MHz; CDCl<sub>3</sub>; Me<sub>4</sub>Si) δ<sub>H</sub>, ppm: −2.72 (2H, s, pyrrole-N<italic>H</italic>); 7.38 (4H, dd, <italic>J</italic> = 7.5, <italic>J</italic> = 4.8, pyridine-4-<italic>H</italic>); 7.94 (4H, dd, <italic>J</italic> = 8.0, <italic>J</italic> = 7.5, <italic>J</italic> = 1.8, pyridine-5-<italic>H</italic>); 8.16 (4H, d, <italic>J</italic> = 8.0, pyridine-6-<italic>H</italic>); 8.31–8.36 (8H, m, phenyl-<italic>H</italic>); 8.49–8.54 (8H, m, phenyl-<italic>H</italic>); 8.79 (4H, ddd, <italic>J</italic> = 4.8, <italic>J</italic> = 1.8, <italic>J</italic> = 0.9, pyridine-3-<italic>H</italic>); 8.91 (8H, s, β-pyrrole-<italic>H</italic>). Elemental Analysis H<sub>2</sub>T-2-PyPhP·2.5H<sub>2</sub>O, Anal. Calcd for C<sub>72</sub>H<sub>47</sub>N<sub>8</sub>O<sub>2.5</sub>: H, 4.89; C, 79.40; N, 11.57%. Found: H, 5.00; C, 79.36; N, 11.37%. UV–vis (CHCl<sub>3</sub>): λ<sub>max</sub>, nm (log ε) 252.5 (4.75), 275.3 (4.66), 373.9 (4.43), 422.5 (5.71), 517.5 (4.30), 553.8 (4.10), 592.1 (3.81), 648.6 (3.79).</p></sec><sec id="sec2.3.2"><title>H<sub>2</sub>TE-2-PyPhPCl<sub>4</sub>, <italic>meso</italic>-Tetrakis(phenyl-4-(<italic>N</italic>-ethylpyridinium-2′-yl))porphyrin Tetrachloride</title><p>H<sub>2</sub>T-2-PyPhP (50 mg; 0.542 mmol) was dissolved in 4.5 mL of DMF at 115 °C, preheated for 20 min while purged with nitrogen. To the resulting solution was added 3.8 g (0.019 mol) of ethyl <italic>p</italic>-toluenesulfonate. The course of <italic>N</italic>-ethylation was followed on TLC silica gel plates using acetonitrile/KNO<sub>3</sub>(sat)/water = 8/1/1 as a mobile phase. After 20 h of stirring, a new portion of 1.64 g of ethyl <italic>p</italic>-toluenesulfonate was added and stirred for another 26 h at 115 °C until reaction was completed. The <italic>N</italic>-quaternization with ethyl <italic>p</italic>-toluenesulfonate led to the formation of tetracationic ligand, H<sub>2</sub>TE-2-PyPhP<sup>4+</sup>. Interestingly, the reaction progressed very slowly when compared to the 3 h reaction time of <italic>N</italic>-ethylation of H<sub>2</sub>T-2-PyP under similar conditions. Such a difference in reaction rates is due to the steric and electronic effects imposed by the coplanarity of phenyl and pyridyl moieties of H<sub>2</sub>T-2-PyPhP. Due to the prolonged reaction time, additional spots appeared on TLC plate, and were separated by column chromatography. Though not fully characterized, these impurities are most likely the products of alkylation of inner pyrrolic nitrogens.</p><p>H<sub>2</sub>TE-2-PyPhP<sup>4+</sup> was precipated from the reaction mixture by diethyl ether, filtrated, and washed with diethyl ether (5 × 50 mL). It was isolated as a Cl<sup>–</sup> salt as described above via two PF<sub>6</sub><sup>–</sup>/Cl<sup>–</sup> sequential precipitations. The solid was then chromatographed on column chromatography with 1/500 = Et<sub>3</sub>N/(1/1/8 = KNO<sub>3</sub>/saturated H<sub>2</sub>O/acetonitrile) as a solvent system. Solvent was evaporated under reduced pressure and the porphyrin isolated after two PF<sub>6</sub><sup>–</sup>/Cl<sup>–</sup> sequential precipitations. The Cl<sup>–</sup> salt was dried in a vacuum oven. Yield (calculated based on elemental analysis): 51 mg (80%).</p></sec><sec id="sec2.3.3"><title>MnTE-2-PyPhPCl<sub>5,</sub> Mn(III) <italic>meso</italic>-Tetrakis(phenyl-4-(<italic>N</italic>-ethylpyridinium-2′-yl))porphyrin Pentachloride</title><p>The 40 mg portion of H<sub>2</sub>TE-2-PyPhPCl<sub>4</sub> (0.34 mmol) was dissolved in 40 mL of water, and the pH of the resulting solution was adjusted to 11.7. A 40-fold excess of MnCl<sub>2</sub>·4H<sub>2</sub>O (1.35 mmol, 0.27 g) was added into the solution at 25 °C while stirring, and was accompanied by the pH drop to ∼8.5. The solution was then heated for another hour at 100 °C to allow for the completion of metalation. The course of reaction was followed on silica gel TLC plates using acetonitrile/KNO<sub>3</sub>(sat)/water = 8/1/1 as a mobile phase. The isolation and purification of the MnTE-2-PyPhPCl<sub>5</sub> was done as described above for the <italic>ortho</italic> Mn pyridylporphyrins. The isolated yield was quantitative, 41 mg (95.3%).</p></sec></sec><sec id="sec2.4"><title>Elemental Analysis</title><p>Elemental analyses of porphyrins and their Mn complexes were performed in duplicates with Atlantic MicroLab (Norcross, GA) and average values presented.</p><sec id="sec2.4.1"><title>H<sub>2</sub>TPhE-2-PyPCl<sub>4</sub>·10H<sub>2</sub>O</title><p>Anal. Calcd for C<sub>72</sub>H<sub>82</sub>Cl<sub>4</sub>N<sub>8</sub>O<sub>10</sub>: H, 6.07; C, 63.53; N, 8.23; Cl, 10.42%. Found: H, 6.14; C, 63.29; N, 8.20; Cl, 10.17%.</p></sec><sec id="sec2.4.2"><title>MnTPhE-2-PyPCl<sub>5</sub>·9H<sub>2</sub>O</title><p>Anal. Calcd for C<sub>72</sub>H<sub>18</sub>Cl<sub>5</sub>MnN<sub>8</sub>O<sub>9</sub>: H, 5.49; C, 60.41; N, 7.83; Cl, 12.38%. Found: H, 5.70; C, 60.37; N, 7.85; Cl, 12.11%.</p></sec><sec id="sec2.4.3"><title>H<sub>2</sub>TnHexOE-2-PyPCl<sub>4</sub>·8H<sub>2</sub>O</title><p>Anal. Calcd for C<sub>72</sub>H<sub>110</sub>Cl<sub>4</sub>N<sub>8</sub>O<sub>12</sub>: H, 7.8; C, 60.84; N, 7.88%. Found: H, 7.72; C, 60.56; N, 7.92%.</p></sec><sec id="sec2.4.4"><title>MnTnHexOE-2-PyPCl<sub>5</sub>·8.5H<sub>2</sub>O</title><p>Anal. Calcd for C<sub>64</sub>H<sub>94</sub>Cl<sub>5</sub>MnN<sub>8</sub>O<sub>9</sub>: H, 7.23; C, 56.93; N, 7.38; Cl, 11.67%. Found: H, 7.05; C, 56.58; N, 7.68; Cl, 11.28%.</p></sec><sec id="sec2.4.5"><title>H<sub>2</sub>TE-2-PyPhPCl<sub>4</sub>·10.5H<sub>2</sub>O·0.5KNO<sub>3</sub>·2KCl</title><p>Anal. Calcd for C<sub>72</sub>H<sub>83</sub>Cl<sub>6</sub>MnN<sub>8</sub>O<sub>12</sub>K<sub>2.5</sub>: H, 5.30; C, 55.40; N, 7.63; Cl, 13.63%. Found: H, 5.40; C, 55.74; N, 7.66; Cl, 14.60%.</p></sec><sec id="sec2.4.6"><title>MnTE-2-PyPhPCl<sub>5</sub>·10.5H<sub>2</sub>O</title><p>Anal. Calcd for C<sub>72</sub>H<sub>81</sub>Cl<sub>5</sub>MnN<sub>8</sub>O<sub>10.5</sub>: H, 5.60; C, 59.29; N, 7.68; Cl, 12.15%. Found: H, 5.48; C, 59.42; N, 7.50; Cl, 11.93%.</p></sec></sec><sec id="sec2.5"><title>UV–Vis Spectroscopy</title><p>UV–vis spectra were recorded in water at room temperature on a UV-2501PC Shimadzu spectrophotometer with 0.5 nm resolution in 1 cm quartz cuvette (Table <xref rid="tbl1" ref-type="other">1</xref>). The UV–vis spectra for new compounds are provided in <xref rid="notes-1" ref-type="notes">Supporting Information</xref> (Figures S1–S3).</p><table-wrap id="tbl1" position="float"><label>Table 1</label><caption><title>Spectral Properties of Porphyrins and Their Mn Complexes</title></caption><table frame="hsides" rules="groups" border="0"><colgroup><col align="left"/><col align="left"/></colgroup><thead><tr><th style="border:none;" align="center">(metallo)porphyrin</th><th style="border:none;" align="center">λ<sub>max</sub> nm (log ε)<xref rid="t1fn1" ref-type="table-fn">a</xref></th></tr></thead><tbody><tr><td style="border:none;" align="left">MnTBAP<xref rid="t1fn2" ref-type="table-fn">b</xref></td><td style="border:none;" align="left">230.0 (4.93), 290.0 (4.49), 381.0 (4.84), 401.0 (4.84), 420.0 (sh, 4.70), 468.0 (5.04), 515.0 (3.92), 566.0 (4.16), 599.0 (4.07), 684.0 (sh, 3.23), 712.0 (sh, 3.20), 780.0 (3.24), 811.0 (sh, 3.17)</td></tr><tr><td style="border:none;" align="left">MnTE-2-PyPCl<sub>5</sub><xref rid="t1fn2" ref-type="table-fn">b</xref></td><td style="border:none;" align="left">363.5 (4.68), 409.0 (4.32), 454.0 (5.14), 499.0 (3.75), 558.0 (4.08), 782.0 (3.26)</td></tr><tr><td style="border:none;" align="left">MnTE-3-PyPCl<sub>5</sub><xref rid="t1fn2" ref-type="table-fn">b</xref></td><td style="border:none;" align="left">214.0 (4.77), 260.0 (4.60), 373.0 (4.74), 395.0 (4.78), 460.0 (5.19), 502.0 (3.85), 557.0 (4.16), 674.0 (3.25), 766.0 (3.37), 837.0 (2.40)</td></tr><tr><td style="border:none;" align="left">H<sub>2</sub>TnHexOE-2-PyPCl<sub>4</sub></td><td style="border:none;" align="left">264.4 (4.38), 419.4 (5.35), 513.5 (4.27), 545.5 (3.64), 586.4 (3.86), 640 (3.43)</td></tr><tr><td style="border:none;" align="left">MnTnHexOE-2-PyPCl<sub>5</sub></td><td style="border:none;" align="left">212.5 (4.72), 261.7 (4.56), 365.4 (4.74), 411.4 (4.39), 455.5 (5.26), 561.1 (4.16), 786.5 (3.38)</td></tr><tr><td style="border:none;" align="left">H<sub>2</sub>TPhE-2-PyPCl<sub>4</sub></td><td style="border:none;" align="left">263.2 (4.41), 419.4 (5.34), 514.6 (4.24), 585.6 (3.85), 638.6 (3.26)</td></tr><tr><td style="border:none;" align="left">MnTPhE-2-PyPCl<sub>5</sub></td><td style="border:none;" align="left">260.7 (4.60), 364.9 (4.72), 455.5 (5.27), 560.4 (4.17), 783.1 (3.41)</td></tr><tr><td style="border:none;" align="left">H<sub>2</sub>TE-2-PyPhPCl<sub>4</sub></td><td style="border:none;" align="left">274.2 (4.62), 414.2 (5.77), 515.5 (4.29), 552.1 (3.95), 579.8 (3.88), 634.1 (3.64)</td></tr><tr><td style="border:none;" align="left">MnTE-2-PyPhPCl<sub>5</sub></td><td style="border:none;" align="left">273.1 (4.71), 378.7 (4.82), 400 (4.83), 466.3 (5.06), 514 (3.90), 562.5 (4.15), 597.1 (4.01), 773.8 (3.27)</td></tr><tr><td style="border:none;" align="left">MnTnOct-2-PyPCl<sub>5</sub><xref rid="t1fn2" ref-type="table-fn">b</xref></td><td style="border:none;" align="left">364.0 (4.72), 414.0 (4.44), 454.5 (5.24), 500.5 (3.84), 559.5 (4.14), 781.0 (3.25)</td></tr></tbody></table><table-wrap-foot><fn id="t1fn1"><label>a</label><p>Spectra were recorded in water at room temperature unless otherwise noted. Molar absorption coefficients (M<sup>–1</sup> cm<sup>–1</sup>) were determined within 5% errors. λ<sub>max</sub> (nm) were determined with errors inside ±0.5 nm.</p></fn><fn id="t1fn2"><label>b</label><p>Data are taken from ref (<xref ref-type="bibr" rid="ref79">79</xref>).</p></fn></table-wrap-foot></table-wrap></sec><sec id="sec2.6"><title>Electrospray-Ionization Mass Spectrometry</title><p>Electrospray ionization mass spectrometric (ESI-MS) analyses were performed on Applied Biosystems MDS Sciex 3200 Q Trap LC/MS/MS spectrometer at Duke Comprehensive Cancer Center, Shared Resource PK Laboratories, as described elsewhere.<sup><xref ref-type="bibr" rid="ref73">73</xref>,<xref ref-type="bibr" rid="ref80">80</xref>,<xref ref-type="bibr" rid="ref81">81</xref></sup> Samples of ∼1 μM concentrations were prepared in acetonitrile/H<sub>2</sub>O mixture (1/1, v/v) containing 0.01% v/v heptafluorobutyric acid, and infused for 1 min at 10 μL/min into the spectrometer (curtain gas 20 V, ion spray voltage 3500 V, ion source 30 V, <italic>T</italic> = 300 °C, declustering potential 20 V, entrance potential 1 V, collision energy 5 V, gas N<sub>2</sub>). Under given conditions, in the presence of ion-pairing heptafluorobutyrate anion (HFBA<sup>–</sup>), no fragmentation was observed; the data relate to species originally present in solutions. The absence of peaks associated with partially alkylated and nonmetalated species unambiguously indicates the purity of the sample. Data are summarized in Table <xref rid="tbl2" ref-type="other">2</xref>. All MS spectra are provided in <xref rid="notes-1" ref-type="notes">Supporting Information</xref> (Figure S4).</p><table-wrap id="tbl2" position="float"><label>Table 2</label><caption><title>Electrospray Ionization Mass Spectrometry (ESI-MS) Data for New Porphyrins, H<sub>2</sub>P, and their Mn(III) Complexes<xref rid="t2fn1" ref-type="table-fn">a</xref></title></caption><table frame="hsides" rules="groups" border="0"><colgroup><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/><col align="center"/></colgroup><thead><tr><th style="border:none;" align="center"> </th><th colspan="6" align="center"><italic>m</italic>/<italic>z</italic> [found (calcd)]<hr/></th></tr><tr><th style="border:none;" align="center">species<xref rid="t2fn1" ref-type="table-fn">a</xref></th><th style="border:none;" align="center">H<sub>2</sub>TnHexOE-2-PyP<sup><bold>4+</bold></sup></th><th style="border:none;" align="center">MnTnHexOE-2-PyP<sup>5+</sup></th><th style="border:none;" align="center">H<sub>2</sub>TPhE-2-PyP<sup>4+</sup></th><th style="border:none;" align="center">MnTPhE-2-PyP<sup>5+</sup></th><th style="border:none;" align="center">H<sub>2</sub>TE-2-PyPhP<sup>4+</sup></th><th style="border:none;" align="center">MnTE-2-PyPhP<sup>5+</sup></th></tr></thead><tbody><tr><td style="border:none;" align="left">[P<sup><italic>n</italic>+</sup> + HFBA<sup><bold>-</bold></sup>]<sup>(<italic>n</italic>−1)+</sup>/(<italic>n</italic> – 1)</td><td style="border:none;" align="left">449.4 (449.2)</td><td style="border:none;" align="left">350.2 (350.2)</td><td style="border:none;" align="left">417.8 (417.2)</td><td style="border:none;" align="left">326.4 (326.1)</td><td style="border:none;" align="left">417.4 (417.2)</td><td style="border:none;" align="left">326.6 (326.1)</td></tr><tr><td style="border:none;" align="left">[P<sup><italic>n</italic>+</sup> + 2HFBA<sup><bold>–</bold></sup>]<sup>(<italic>n</italic>−2)+</sup>/(<italic>n</italic> – 2)</td><td style="border:none;" align="left">780.2 (780.4)</td><td style="border:none;" align="left">537.7 (537.9)</td><td style="border:none;" align="left">732.9 (732.2)</td><td style="border:none;" align="left">505.6 (505.8)</td><td style="border:none;" align="left">732.1 (732.2)</td><td style="border:none;" align="left">506.2 (505.8)</td></tr><tr><td style="border:none;" align="left">[P<sup><italic>n</italic>+</sup> + 3HFBA<sup>–</sup>]<sup>(<italic>n</italic>−3)+</sup>/(<italic>n</italic> – 3)</td><td style="border:none;" align="left"> </td><td style="border:none;" align="left">913.0 (913.3)</td><td style="border:none;" align="left"> </td><td style="border:none;" align="left">864.8 (865.2)</td><td style="border:none;" align="left"> </td><td style="border:none;" align="left">865.8 (865.2)</td></tr><tr><td style="border:none;" align="left">[H<sub>2</sub>P]<sup>4+</sup>/4</td><td style="border:none;" align="left">283.6 (283.7)</td><td style="border:none;" align="left"> </td><td style="border:none;" align="left">260.2 (259.6)</td><td style="border:none;" align="left"> </td><td style="border:none;" align="left">259.9 (259.6)</td><td style="border:none;" align="left"> </td></tr><tr><td style="border:none;" align="left">[H<sub><bold>2</bold></sub>P<sup>4+</sup> – H<sup>+</sup>]<sup>3+</sup>/3</td><td style="border:none;" align="left">378.1 (377.9)</td><td style="border:none;" align="left"> </td><td style="border:none;" align="left">346.5 (345.8)</td><td style="border:none;" align="left"> </td><td style="border:none;" align="left"> </td><td style="border:none;" align="left"> </td></tr><tr><td style="border:none;" align="left">[P<sup><italic>n</italic>+</sup> + H<sup>+</sup>+ 2HFBA<sup>–</sup>]<sup>(<italic>n</italic>−1)+</sup>/(<italic>n</italic> – 1)</td><td style="border:none;" align="left"> </td><td style="border:none;" align="left"> </td><td style="border:none;" align="left"> </td><td style="border:none;" align="left"> </td><td style="border:none;" align="left">488.9 (488.5)</td><td style="border:none;" align="left"> </td></tr><tr><td style="border:none;" align="left">[P<sup><italic>n</italic>+</sup> – H<sup>+</sup>+ HFBA<sup>–</sup>]<sup>(<italic>n</italic>−2)+</sup>/(<italic>n</italic> – 2)</td><td style="border:none;" align="left">673.3 (673.4)</td><td style="border:none;" align="left"> </td><td style="border:none;" align="left">625.9 (625.2)</td><td style="border:none;" align="left"> </td><td style="border:none;" align="left"> </td><td style="border:none;" align="left"> </td></tr><tr><td style="border:none;" align="left">[P<sup><italic>n</italic>+</sup> + H<sup>+</sup> + 3HFBA<sup>–</sup>]<sup>(<italic>n</italic>−2)+</sup>/(<italic>n</italic> – 2)</td><td style="border:none;" align="left">887.0 (887.3)</td><td style="border:none;" align="left"> </td><td style="border:none;" align="left"> </td><td style="border:none;" align="left"> </td><td style="border:none;" align="left">840.3 (839.2)</td><td style="border:none;" align="left"> </td></tr></tbody></table><table-wrap-foot><fn id="t2fn1"><label>a</label><p>∼1 μM solution of porphyrins and metalloporphyrins in 1/1 v/v acetonitrile/H<sub>2</sub>O [containing 0.01% v/v heptafluorobutyric acid (HFBA)] mixture, 20 V cone voltage; <italic>n</italic> = 4 or 5 corresponding to H<sub>2</sub>P or MnP accordingly.</p></fn></table-wrap-foot></table-wrap></sec><sec id="sec2.7"><title>Lipophilicity</title><p>Both TLC retention factor, <italic>R<sub>f</sub></italic> (compound path/solvent path), and the partition coefficient between n-octanol and water, log <italic>P</italic><sub>OW</sub>, are equally valid parameters in assessing lipophilicity of the free ligands and their Mn complexes.<sup><xref ref-type="bibr" rid="ref79">79</xref>,<xref ref-type="bibr" rid="ref81">81</xref></sup><italic>R<sub>f</sub></italic> was obtained on silica gel plates using acetonitrile/KNO<sub>3</sub>(sat)/water = 8/1/1 as previously described.<sup><xref ref-type="bibr" rid="ref81">81</xref></sup> As it is difficult to impossible to fully reproduce the <italic>R</italic><sub><italic>f</italic></sub> values from one experiment to another, we are routinely comparing all the compounds of interest in a single experiment. The log <italic>P</italic><sub>OW</sub> values of the newly synthesized compounds were determined as reported by Kos et al.<sup><xref ref-type="bibr" rid="ref81">81</xref></sup> The log <italic>P</italic><sub>BW</sub> values (the partition between water-saturated n-butanol and n-butanol-saturated water) were determined experimentally using the following equation: log <italic>P</italic><sub>BW</sub> = log(<italic>C</italic><sub>nBuOH</sub>/<italic>C</italic><sub>water</sub>). The log <italic>P</italic><sub>BW</sub> values were converted to log <italic>P</italic><sub>OW</sub> using the equation: log <italic>P</italic><sub>OW</sub> = 1.55 × log <italic>P</italic><sub>BW</sub> – 0.54.<sup><xref ref-type="bibr" rid="ref81">81</xref>,<xref ref-type="bibr" rid="ref82">82</xref></sup> The log <italic>P</italic><sub>OW</sub> values for the most hydrophilic porphyrins, MnTE-2-PyP<sup>5+</sup> and MnTE-3-PyP<sup>5+</sup>, were determined using the following equations: log <italic>P</italic><sub>OW</sub> = 12.207 × <italic>R</italic><sub><italic>f</italic></sub> – 8.521 for <italic>ortho</italic> Mn(III) <italic>N</italic>-alkylpyridyl porphyrins (i.e., MnTE-2-PyP<sup>5+</sup>), and log <italic>P</italic><sub>OW</sub> = 8.764 × <italic>R<sub>f</sub></italic> – 8.198 for <italic>meta</italic> Mn(III) <italic>N</italic>-alkylpyridyl porphyrins (i.e., MnTE-3-PyP<sup>5+</sup>).<sup><xref ref-type="bibr" rid="ref81">81</xref>,<xref ref-type="bibr" rid="ref83">83</xref></sup> The <italic>R<sub>f</sub></italic> and log <italic>P</italic><sub>OW</sub> values are given in Table <xref rid="tbl3" ref-type="other">3</xref>.</p><table-wrap id="tbl3" position="float"><label>Table 3</label><caption><title>Lipophilicity of MnPs Determined in Terms of TLC Retention Factor, <italic>R</italic><sub>f</sub>, and Partition Coefficient between n-Octanol and Water, log <italic>P</italic><sub>OW</sub></title></caption><table frame="hsides" rules="groups" border="0"><colgroup><col align="left"/><col align="center"/><col align="center"/></colgroup><thead><tr><th style="border:none;" align="center"> </th><th colspan="2" align="center">lipophilicity<hr/></th></tr><tr><th style="border:none;" align="center">Mn porphyrin</th><th style="border:none;" align="center"><italic>R<sub><bold>f</bold></sub></italic><xref rid="t3fn1" ref-type="table-fn">a</xref></th><th style="border:none;" align="center">log <italic>P</italic><sub>ow</sub><xref rid="t3fn2" ref-type="table-fn">b</xref></th></tr></thead><tbody><tr><td style="border:none;" align="left">MnTE-2-PyP<sup>5+</sup></td><td style="border:none;" align="justify">0.07</td><td style="border:none;" align="justify">–7.67<xref rid="t3fn3" ref-type="table-fn">c</xref></td></tr><tr><td style="border:none;" align="left">MnTE-3-PyP<sup>5+</sup></td><td style="border:none;" align="justify">0.12</td><td style="border:none;" align="justify">–7.15<xref rid="t3fn3" ref-type="table-fn">c</xref></td></tr><tr><td style="border:none;" align="left">MnTnHexOE-2-PyP<sup>5+</sup></td><td style="border:none;" align="justify">0.50(0.53)</td><td style="border:none;" align="justify">–1.67</td></tr><tr><td style="border:none;" align="left">MnTPhE-2-PyP<sup>5+</sup></td><td style="border:none;" align="justify">0.40(0.47)</td><td style="border:none;" align="justify">–5.90</td></tr><tr><td style="border:none;" align="left">MnTE-2-PyPhP<sup>5+</sup></td><td style="border:none;" align="justify">0.32(0.45)</td><td style="border:none;" align="justify">–5.51</td></tr><tr><td style="border:none;" align="left">MnTnOct-2-PyP<sup>5+</sup></td><td style="border:none;" align="justify">0.48</td><td style="border:none;" align="justify">–2.27</td></tr></tbody></table><table-wrap-foot><fn id="t3fn1"><label>a</label><p>Lipophilicities of porphyrin ligands of the related Mn complexes are given in parentheses. The TLC was done on silica gel plates using acetonitrile/KNO<sub>3</sub>(sat)/water = 8/1/1 as a mobile phase.</p></fn><fn id="t3fn2"><label>b</label><p>Determined experimentally using n-butanol and water biphasic system and converted to log <italic>P</italic><sub>OW</sub> according to the equation log <italic>P</italic><sub>OW</sub> = 1.55 × log <italic>P</italic><sub>BW</sub> – 0.54; <italic>P</italic><sub>BW</sub> is the partition between n-butanol and water.<sup><xref ref-type="bibr" rid="ref81">81</xref>,<xref ref-type="bibr" rid="ref82">82</xref></sup></p></fn><fn id="t3fn3"><label>c</label><p>Data obtained from <italic>R<sub>f</sub></italic> vs log <italic>P</italic><sub>OW</sub> relationships.<sup><xref ref-type="bibr" rid="ref81">81</xref>,<xref ref-type="bibr" rid="ref83">83</xref></sup></p></fn></table-wrap-foot></table-wrap></sec><sec id="sec2.8"><title>Electrochemistry</title><p>Cyclic voltammetry measurements were performed under argon in a glass cell on CH Instruments model 600 voltammetric analyzer, as described previously.<sup><xref ref-type="bibr" rid="ref5">5</xref>,<xref ref-type="bibr" rid="ref84">84</xref></sup> Stock solutions of MnPs were prepared by dissolving solids in deionized water. Working solutions of ∼0.2 mM MnPs were prepared in 0.05 M phosphate buffer (pH = 7.8). The supporting electrolyte in all measurements was 0.1 M NaCl. The pH values were determined on a Denver Instrument Model 250 pH-meter using a glass electrode calibrated with the standard buffers (pH 4.00, 7.00, and 10.00). The concentrations of MnPs were determined spectrophotometrically. All potentials are reported versus the normal hydrogen electrode (NHE). MnTE-2-PyP<sup>5+</sup> with <italic>E</italic><sub>1/2</sub> = +228 mV versus NHE was used as a reference.<sup><xref ref-type="bibr" rid="ref16">16</xref>,<xref ref-type="bibr" rid="ref80">80</xref>,<xref ref-type="bibr" rid="ref81">81</xref></sup> Its voltammetry was performed before and after each series of measurements. The data are presented in Table <xref rid="tbl4" ref-type="other">4</xref>.</p></sec><sec id="sec2.9"><title>Catalysis of O<sub>2</sub><sup>•–</sup> Dismutation (Cytochrome <italic>c</italic> Assay)</title><p>The ability of newly synthesized Mn metalloporphyrins to dismute O<sub>2</sub><sup>•–</sup> was evaluated via cytochrome <italic>c</italic> assay. The validity of assay was proven with pulse radiolysis and stopped-flow methodology.<sup><xref ref-type="bibr" rid="ref84">84</xref>−<xref ref-type="bibr" rid="ref88">88</xref></sup> The cyt <italic>c</italic> assay is based on O<sub>2</sub><sup>•–</sup> production via xanthine/xanthine oxidase reaction and metalloporphyrin ability to compete with ferricytochrome <italic>c</italic> in scavenging O<sub>2</sub><sup>•–</sup>. The experiments were conducted at room temperature (25 ± 1) °C in 0.05 M potassium phosphate buffer, pH 7.8, and 0.1 mM EDTA as previously described in detail.<sup><xref ref-type="bibr" rid="ref84">84</xref></sup> The reduction of cytochrome <italic>c</italic> was followed at 550 nm. MnTE-2-PyP<sup>5+</sup> was used as a standard. Data are summarized in Table <xref rid="tbl4" ref-type="other">4</xref>. Kinetic traces, plots ((<italic>v</italic><sub>0</sub>/<italic>v<sub>i</sub></italic>) −1) vs [MnP]) for the calculations of IC<sub>50</sub>, and the information on the calculation of <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) from such plots are provided in Figures S12 and S13 of <xref rid="notes-1" ref-type="notes">Supporting Information</xref>.</p><table-wrap id="tbl4" position="float"><label>Table 4</label><caption><title>Metal-Centered Reduction Potential, <italic>E</italic><sub>1/2</sub> vs NHE of Mn<sup>III</sup>P/Mn<sup>II</sup>P Redox Couple, Proton Dissociation Constant of First Axial Water, p<italic>K</italic><sub>a1</sub>, log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) for the Catalysis of O<sub>2</sub><sup>•–</sup> Dismutation, log <italic>k</italic><sub>red</sub>(ONOO<sup>–</sup>) for the ONOO<sup>–</sup> Reduction, and Initial Rates for the Catalysis of Ascorbate HA<sup>–</sup> Oxidation with MnPs, <italic>v</italic><sub>0</sub>(HA<sup>–</sup> Oxidation). Relative Molecular Masses, <italic>M</italic><sub>r</sub> are Listed Also</title></caption><table frame="hsides" rules="groups" border="0"><colgroup><col align="left"/><col align="justify"/><col align="justify"/><col align="justify"/><col align="justify"/><col align="justify"/><col align="justify"/></colgroup><thead><tr><th style="border:none;" align="center">compd</th><th style="border:none;" align="center"><italic>M</italic><sub><italic>r</italic></sub></th><th style="border:none;" align="center">p<italic>K</italic><sub>a1</sub></th><th style="border:none;" align="center"><italic>E</italic><sub>1/2</sub>, mV vs NHE<xref rid="t4fn1" ref-type="table-fn">a</xref></th><th style="border:none;" align="center">log <italic>k</italic><sub>cat</sub> (O<sub>2</sub><sup>•–</sup>)<xref rid="t4fn2" ref-type="table-fn">b</xref></th><th style="border:none;" align="center">log <italic>k</italic><sub>red</sub> (ONOO<sup>-</sup>)<xref rid="t4fn3" ref-type="table-fn">c</xref></th><th style="border:none;" align="center"><italic>v</italic><sub>0</sub> (HA<sup>–</sup> oxidation), nM s<sup>-1</sup><xref rid="t4fn4" ref-type="table-fn">d</xref></th></tr></thead><tbody><tr><td style="border:none;" align="left">MnTBAP<sup>3–</sup></td><td style="border:none;" align="justify">842.7</td><td style="border:none;" align="justify">12.6<sup><xref ref-type="bibr" rid="ref4">4</xref></sup></td><td style="border:none;" align="justify">–194<sup><xref ref-type="bibr" rid="ref79">79</xref></sup></td><td style="border:none;" align="justify">3.16<sup><xref ref-type="bibr" rid="ref79">79</xref></sup></td><td style="border:none;" align="justify">5.02<sup><xref ref-type="bibr" rid="ref79">79</xref></sup></td><td style="border:none;" align="justify">2.26</td></tr><tr><td style="border:none;" align="left">MnTE-2-PyPhP<sup>5+</sup></td><td style="border:none;" align="justify">1269.5</td><td style="border:none;" align="justify">12.0<xref rid="t4fn5" ref-type="table-fn">e</xref></td><td style="border:none;" align="justify">–65</td><td style="border:none;" align="justify">5.55</td><td style="border:none;" align="justify">5.93</td><td style="border:none;" align="justify">18.24</td></tr><tr><td style="border:none;" align="left">MnTE-3-PyP<sup>5+</sup></td><td style="border:none;" align="justify">965.1</td><td style="border:none;" align="justify">11.5<sup><xref ref-type="bibr" rid="ref4">4</xref></sup></td><td style="border:none;" align="justify">54<sup><xref ref-type="bibr" rid="ref79">79</xref></sup></td><td style="border:none;" align="justify">6.65<sup><xref ref-type="bibr" rid="ref79">79</xref></sup></td><td style="border:none;" align="justify">6.81</td><td style="border:none;" align="justify">229.96</td></tr><tr><td style="border:none;" align="left">MnTE-2-PyP<sup>5+</sup></td><td style="border:none;" align="justify">965.1</td><td style="border:none;" align="justify">11.0<sup><xref ref-type="bibr" rid="ref4">4</xref></sup></td><td style="border:none;" align="justify">228<sup><xref ref-type="bibr" rid="ref79">79</xref></sup></td><td style="border:none;" align="justify">7.76<sup><xref ref-type="bibr" rid="ref79">79</xref></sup></td><td style="border:none;" align="justify">7.53<sup><xref ref-type="bibr" rid="ref4">4</xref></sup></td><td style="border:none;" align="justify">312.84</td></tr><tr><td style="border:none;" align="left">MnTPhE-2-PyP<sup>5+</sup></td><td style="border:none;" align="justify">1269.5</td><td style="border:none;" align="justify">10.8<xref rid="t4fn5" ref-type="table-fn">e</xref></td><td style="border:none;" align="justify">259</td><td style="border:none;" align="justify">7.66</td><td style="border:none;" align="justify">7.14</td><td style="border:none;" align="justify">147.21</td></tr><tr><td style="border:none;" align="left">MnTnHexOE-2-PyP<sup>5+</sup></td><td style="border:none;" align="justify">1365.8</td><td style="border:none;" align="justify">10.7<xref rid="t4fn5" ref-type="table-fn">e</xref></td><td style="border:none;" align="justify">313</td><td style="border:none;" align="justify">7.92</td><td style="border:none;" align="justify">7.61</td><td style="border:none;" align="justify">76.33</td></tr><tr><td style="border:none;" align="left">MnTnOct-2-PyP<sup>5+</sup></td><td style="border:none;" align="justify">1301.8</td><td style="border:none;" align="justify">10.5<sup><xref ref-type="bibr" rid="ref4">4</xref></sup></td><td style="border:none;" align="justify">340</td><td style="border:none;" align="justify">7.71<sup><xref ref-type="bibr" rid="ref79">79</xref></sup></td><td style="border:none;" align="justify">7.15<sup><xref ref-type="bibr" rid="ref79">79</xref></sup></td><td style="border:none;" align="justify">54.29</td></tr></tbody></table><table-wrap-foot><fn id="t4fn1"><label>a</label><p><italic>E</italic><sub>1/2</sub> of Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple is determined in 0.05 M phosphate buffer (pH 7.8, 0.1 M NaCl).</p></fn><fn id="t4fn2"><label>b</label><p><italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) is determined by cytochrome <italic>c</italic> assay in 0.05 M potassium phosphate buffer [pH 7.8, at (25 ± 1) °C].</p></fn><fn id="t4fn3"><label>c</label><p><italic>k</italic><sub>red</sub>(ONOO<sup>–</sup>) is determined by stopped-flow technique in 0.05 M potassium phosphate buffer [pH 7.4, at (37 ± 0.1) °C].</p></fn><fn id="t4fn4"><label>d</label><p><italic>v</italic><sub>0</sub>, initial rate for HA<sup>–</sup> oxidation, was determined spectrophotometrically under aerobic conditions: 5 μM MnP, 0.15 mM sodium ascorbate, 5 mM EDTA, pH 7.4 maintained with 0.05 M Tris buffer and at (25 ± 1) °C. The mono-deprotonated HA<sup>–</sup> is the main ascorbate species at pH 7.8.</p></fn><fn id="t4fn5"><label>e</label><p>p<italic>K</italic><sub>a1</sub> values were estimated on the basis of the relationship p<italic>K</italic><sub>a1</sub> vs <italic>E</italic><sub>1/2</sub> of Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple published in ref (<xref ref-type="bibr" rid="ref4">4</xref>).</p></fn></table-wrap-foot></table-wrap></sec><sec id="sec2.10"><title>Peroxynitrite Reduction with Mn<sup>III</sup>P</title><p>Oxidation of Mn<sup>III</sup>Ps with peroxynitrite was carried out under pseudo-first-order conditions with peroxynitrite in excess over MnP. In all cases, peroxynitrite (dissolved in a NaOH) was mixed with MnPs dissolved in sodium phosphate buffer. The final concentrations upon mixing were the following: MnPs 0.5 μM (1 μM for MnTnHexOE-2-PyP<sup>5+</sup>), 10-fold excess of peroxynitrite, and 0.05 M sodium phosphate buffer, pH 7.4, 0.1 mM DTPA. The temperature was maintained at (37.0 ± 0.1) °C, and the pH of the reaction mixtures was measured at the outlet of the stopped flow. The reaction was monitored as a change in the absorbance of the Soret band at the following: 456 nm for MnTPhE-2-PyP<sup>5+</sup>, 455 nm for MnTnHexOE-2-PyP<sup>5+</sup>, 467 nm for MnTE-2-PyPhP<sup>5+</sup>, 454 nm for MnTE-2-PyP<sup>5+</sup>, and 460 nm for MnTE-3-PyP<sup>5+</sup>. The pseudo-first-order rate constants, <italic>k</italic><sub>obs</sub> (s<sup>–1</sup>), were determined by fitting the stopped-flow data to a single exponential function. The second-order rate constant was determined from the slope of <italic>k</italic><sub>obs</sub> versus [ONOO<sup>–</sup>] plot. All kinetic runs were performed on a stopped-flow spectrophotometer (Applied Photophysics, SX20). Data are summarized in Table <xref rid="tbl4" ref-type="other">4</xref>. The raw data (kinetic traces, <italic>k</italic><sub>obs</sub> versus [ONOO<sup>–</sup>] plots, and time-resolved equilibrium spectra for all new compounds are provided in <xref rid="notes-1" ref-type="notes">Supporting Information</xref> (Figures S5–S11).</p></sec><sec id="sec2.11"><title>Lipid Peroxidation Assay</title><p>The lipid peroxidation was triggered spontaneously. Rat brains were homogenized on ice in 5 volumes (w/v) of cold 50 mM potassium phosphate buffer, pH 7.00. The 200 μL aliquots were diluted to a final volume of 1.0 mL with 50 mM potassium phosphate buffer and incubated 30 min at 37 °C on a shaking water bath. Under such standardized conditions the 2.2 ± 0.18 μmol/L malondialdehyde (MDA) was produced. If MDA production was not within specified limits, the homogenate was discarded. The level of MDA produced under standardized conditions was taken as 100% lipid peroxidation. In order to measure preformed MDA, as well as the MDA generated during tissue handling and homogenization, butylated hydroxytoluene (BHT) was added before incubation to a final concentration of 60 mM. The MDA content of samples containing BHT did not exceed 0.0254 ± 0.0897 μmol MDA per L homogenate. After 60 min of incubation at 37 °C, BHT (60 mM) was added to all samples, and MDA was initially assessed by colorimetric thiobarbituric acid (TBA) assay.<sup><xref ref-type="bibr" rid="ref89">89</xref></sup> The TBA assay lacks specificity. Thus, all the results were re-evaluated by HPLC analysis as previously described.<sup><xref ref-type="bibr" rid="ref90">90</xref></sup></p></sec><sec id="sec2.12"><title>Catalysis of Ascorbate Oxidation with MnPs</title><p>Initial rates of MnP-catalyzed ascorbate, HA<sup>–</sup>, oxidation to ascorbyl radical, HA<sup>•</sup> (which readily deprotonates to A<sup>•</sup>), were determined with 5 μM metalloporphyrin, 5 mM EDTA, and 0.15 mM sodium ascorbate under aerobic conditions at (25 ± 1)°C and at pH 7.4 maintained with 0.05 M Tris buffer. The buffer was initially treated with Chelex-100 ion-exchange resin (200–400 mesh sodium form, Bio-Rad Life Science) to remove the adventitious metals present in the solution. Ascorbate oxidation was followed at 265 nm on UV–vis spectrophotometer (Shimadzu UV-2550). The molar absorptivity of ascorbate was re-evaluated to be ε<sub>265</sub> = 14 000 M<sup>–l</sup> cm<sup>–1</sup>. The initial rates, <italic>v</italic><sub>0</sub>’s (HA<sup>–</sup> oxidation) (nMs<sup>–1</sup>), which were calculated on the basis of the linear kinetic traces obtained for the first 100 s, are summarized in Table <xref rid="tbl4" ref-type="other">4</xref>. The background rate for noncatalyzed ascorbate oxidation was subtracted from the catalyzed reaction rates.<sup><xref ref-type="bibr" rid="ref77">77</xref></sup></p></sec><sec id="sec2.13"><title>Superoxide-Specific Biological Models. Aerobic Growth of<italic> S. cerevisiae</italic></title><p><italic>S. cerevisiae</italic> strains with mutations in the cytoplasmic CuZnSOD gene (sod1Δ) exhibit amino-acid auxotrophies for lysine and methionine. Only those compounds which are capable of catalyzing the dismutation of superoxide at a rate higher than O<sub>2</sub><sup>•–</sup> self-dismutation substitute for the missing SOD enzyme, thus restoring the aerobic growth of a mutant in a medium lacking lysine or methionine.<sup><xref ref-type="bibr" rid="ref67">67</xref></sup> Such growth impairment makes the SOD-deficient yeast a good system for testing the therapeutic potential of an SOD mimic. The wild type <italic>S. cerevisiae</italic> strain used in this study was EG103, while its corresponding sod1Δ mutant was EG118.<sup><xref ref-type="bibr" rid="ref91">91</xref>,<xref ref-type="bibr" rid="ref92">92</xref></sup> Stock and test cultures were grown as previously described.<sup><xref ref-type="bibr" rid="ref67">67</xref></sup> Tests were performed in 96-well plates in triplicates. Aqueous solutions of MnPs were filter-sterilized (0.22-μm filter, Whatman, Middlesex, U.K.) and added to wells containing 200 μL aliquots of yeast culture in SD medium supplemented with all amino acids except methionine. Cultures in 96-well plates were grown aerobically at 30 °C and 220 rpm on a thermostatic shaker. Since yeast cells tend to clump irrespective of the vigorous shaking, wells were mechanically stirred at regular time intervals using a specifically designed 96-pin sterilized stirrer. In control samples the volume of MnP solution was compensated with sterile distilled water. Growth was followed turbidimetrically at 600 nm using ELISA reader.</p></sec></sec><sec id="sec3"><title>Results and Discussion</title><p>The series of MnPs, MnTnOct-2-PyP<sup>5+</sup>, MnTPhE-2-PyP<sup>5+</sup>, MnTnHexOE-2-PyP<sup>5+</sup>, and MnTE-2-PyPhP<sup>5+</sup>, was synthesized and characterized (Figure <xref rid="fig2" ref-type="fig">2</xref>). With the same porphyrin core, all <italic>meso</italic> substituents have 8 carbon atoms differently organized in either linear or cyclic conformations resulting in compounds of vastly different properties.</p><p>Two porphyrins, MnTnOct-2-PyP<sup>5+</sup> and MnTnHexOE-2-PyP<sup>5+</sup>, bear linear 8 atom-long and 9 atom-long pyridyl substituents, respectively (Figure <xref rid="fig2" ref-type="fig">2</xref>). The latter has one oxygen atom buried so deeply in each of 4 alkylpyridyl chains that the surrounding medium does not sense them. In turn, the lipophilicity of MnTnHexOE-2-PyP<sup>5+</sup> is higher than of MnTnOct-2-PyP<sup>5+</sup>. On the basis of a large amount of lipophilicity measurements reported,<sup><xref ref-type="bibr" rid="ref16">16</xref>,<xref ref-type="bibr" rid="ref80">80</xref>,<xref ref-type="bibr" rid="ref81">81</xref>,<xref ref-type="bibr" rid="ref93">93</xref></sup> we can safely assume that lipophilicity of MnTnHexOE-2-PyP<sup>5+</sup> will be similar to that of MnTnNon-2-PyP<sup>5+</sup> (Mn(III) <italic>meso</italic>-tetrakis(<italic>N</italic>-n-nonylpyridinium-2-yl)porphyrin). Both compounds have equal 9-atom long pyridyl substituents, but the latter does not contain oxygen atoms.</p><p>Two other porphyrins, MnTPhE-2-PyP<sup>5+</sup> and MnTE-2-PyPhP<sup>5+</sup>, bear two cyclic aromatic rings, one phenyl and one pyridyl. The rings are differently organized: in MnTPhE-2-PyP<sup>5+</sup>, the pyridyl ring precedes phenyl, and in MnTE-2-PyPhP<sup>5+</sup> the pyridyl ring follows phenyl ring (Figure <xref rid="fig2" ref-type="fig">2</xref>). In the first case the positively charged quaternary nitrogen atoms are close to the metal site and affect favorably the <italic>E</italic><sub>1/2</sub>. In the second case they are far away, separated from the porphyrin core by the phenyl ring and in turn have minimal impact on <italic>E</italic><sub>1/2</sub>. Besides the effect on <italic>E</italic><sub>1/2</sub>, such distribution of charges affects differentially the shape of the molecule, and in turn the solvation/lipophilicity of these MnPs and their interactions with biotargets.</p><p>Compounds were characterized in terms of the following: (i) elemental analysis; (ii) NMR spectroscopy; (iii) UV–vis spectral properties (Table <xref rid="tbl1" ref-type="other">1</xref>); (iv) electrospray mass spectrometry, ESI-MS (Table <xref rid="tbl2" ref-type="other">2</xref>); (v) lipophilicity in terms of <italic>P</italic><sub>OW</sub> and <italic>R<sub>f</sub></italic> (Table <xref rid="tbl3" ref-type="other">3</xref>); (vi) electrochemistry (metal-centered reduction potential <italic>E</italic><sub>1/2</sub> of Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple) (Table <xref rid="tbl4" ref-type="other">4</xref>); (vii) ability to catalyze O<sub>2</sub><sup>•–</sup> dismutation, <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) (Table <xref rid="tbl4" ref-type="other">4</xref>); (viii) ability to reduce ONOO<sup>–</sup>, <italic>k</italic><sub>red</sub>(ONOO<sup>–</sup>) (Table <xref rid="tbl4" ref-type="other">4</xref>); and (ix) ability to catalyze ascorbate, HA<sup>–</sup> oxidation, described by initial rate, <italic>v</italic><sub>0</sub>(HA<sup>–</sup>) (Table <xref rid="tbl4" ref-type="other">4</xref>, Figure <xref rid="fig7" ref-type="fig">7</xref>).</p><sec id="sec3.1"><title>Lipophilicity of MnPs</title><p>The linear relationship between the chromatographic retention factor, <italic>R<sub>f</sub></italic>, and log <italic>P</italic><sub>OW</sub> has been established for water-soluble cationic Mn <italic>N</italic>-alkylpyridylporphyrins.<sup><xref ref-type="bibr" rid="ref77">77</xref>,<xref ref-type="bibr" rid="ref92">92</xref></sup> It guided us not only in the design and development of lead drug candidates but also in the safe prediction of the partition coefficients of those compounds which are highly hydrophilic and for which log <italic>P</italic><sub>OW</sub> could not be assessed, such as MnTE-2-PyP<sup>5+</sup> and MnTE-3-PyP<sup>5+</sup>.<sup><xref ref-type="bibr" rid="ref81">81</xref>,<xref ref-type="bibr" rid="ref94">94</xref></sup> The log <italic>P</italic><sub>OW</sub> and <italic>R<sub>f</sub></italic> values of the series of ligands and related MnPs are listed in Table <xref rid="tbl3" ref-type="other">3</xref> and Figure <xref rid="fig5" ref-type="fig">5</xref> and are related to other properties of MnPs in Figure <xref rid="fig6" ref-type="fig">6</xref>. While water-soluble MnPs do not distribute readily into n-octanol (as illustrated by highly negative log <italic>P</italic><sub>OW</sub> value), multiple positive charge is a driving force for their distribution into brain and mitochondria.<sup><xref ref-type="bibr" rid="ref83">83</xref></sup> In mitochondria they mimic mitochondrial matrix MnSOD.<sup><xref ref-type="bibr" rid="ref83">83</xref>,<xref ref-type="bibr" rid="ref95">95</xref>−<xref ref-type="bibr" rid="ref97">97</xref></sup></p><fig id="fig5" position="float"><label>Figure 5</label><caption><p>Lipophilicities of Mn(III) porphyrins expressed in terms of chromatographic retention factor, <italic>R</italic><sub><italic>f</italic></sub> (A), and partition coefficient between n-octanol and water, log <italic>P</italic><sub>OW</sub> (B). The <italic>R<sub>f</sub></italic> values are linearly related to log <italic>P</italic><sub>OW</sub> values.<sup><xref ref-type="bibr" rid="ref81">81</xref>,<xref ref-type="bibr" rid="ref93">93</xref></sup> The small differences in <italic>R<sub>f</sub></italic> values translate into large differences in log <italic>P</italic><sub>OW</sub> values.<sup><xref ref-type="bibr" rid="ref81">81</xref>,<xref ref-type="bibr" rid="ref93">93</xref></sup></p></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-01329p_0005" id="gr5" position="float"/></fig><p>The different reorganization of 8 carbon atoms in <italic>meso</italic> substituents resulted in the following observations depicted in Figure <xref rid="fig5" ref-type="fig">5</xref>: (1) lipophilicity of the molecule dropped noticeably by >3.5 log units when 8-carbon atom alkyl chains rearrange into aromatic phenyl or pyridyl substituents; (2) significant increase in lipophilicity was observed when an oxygen atom was introduced into n-octyl chains to form n-hexoxyethyl chains [log <italic>P</italic><sub>OW</sub> (MnTnOct-2-PyP<sup>5+</sup>) < log <italic>P</italic><sub>OW</sub> (MnTnHexOE-2-PyP<sup>5+</sup>)]. We have previously reported that the introduction of one methoxy group at the periphery of each of four hexyl chains reduced significantly the lipophilicity of MnTMOHex-3-PyP<sup>5+</sup> relative to MnTnHex-3-PyP<sup>5+</sup>.<sup><xref ref-type="bibr" rid="ref80">80</xref></sup> Such a drop was considerably minimized when the oxygen atoms were buried deeper into the alkyl chains of MnTnBuOE-2-PyP<sup>5+</sup>.<sup><xref ref-type="bibr" rid="ref16">16</xref></sup> Consequently, this porphyrin has only slightly lower lipophilicity relative to MnTnHex-2-PyP<sup>5+</sup> [log <italic>P</italic><sub>OW</sub> (MnTnBuOE-2-PyP<sup>5+</sup>) = −4.10 versus log <italic>P</italic><sub>OW</sub> (MnTnHex-2-PyP<sup>5+</sup>) = −3.86]. The oxygen atoms in MnTnHexOE-2-PyP<sup>5+</sup> are buried even deeper within the lipophilic n-octyl chains. In turn, the solvation is largely suppressed. We can predict that the chains of a n-hexoxyethyl analog would behave similarly to linear 9-carbon atom substituents in MnTnNon-2-PyP<sup>5+</sup>; the latter is estimated to have partition coefficient log <italic>P</italic><sub>OW</sub> = −1.18 based on reported data for a series of Mn(III) <italic>N</italic>-alkylpyridylporphyrins.<sup><xref ref-type="bibr" rid="ref81">81</xref>,<xref ref-type="bibr" rid="ref94">94</xref></sup></p><p>With phenyl rings at the periphery, such as in MnTPhE-2-PyP<sup>5+</sup>, the compound is more lipophilic than MnTE-2-PyPhP<sup>5+</sup> where the pyridyl cationic charges are exposed at the periphery. As expected, both compounds are much more lipophilic than either MnTE-2-PyP<sup>5+</sup> or MnTE-3-PyP<sup>5+</sup>.</p></sec><sec id="sec3.2"><title>Structure–Activity Relationships among <italic>E</italic><sub>1/2</sub>, p<italic>K</italic><sub>a1</sub>, <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>), and <italic>k</italic><sub>red</sub>(ONOO<sup>–</sup>)</title><p>Cationic Mn(III) porphyrins are among the most potent SOD mimics. They have been tested in numerous oxidative stress related models and have shown remarkable therapeutic potential which is attributed to their ability to interact not only with O<sub>2</sub><sup>•–</sup>, but also with numerous other reactive species, such as ONOO<sup>–</sup>, CO<sub>3</sub><sup>•–</sup>, H<sub>2</sub>O<sub>2</sub>, ClO<sup>–</sup>, ascorbate, lipid reactive species, and thiols, RS<sup>–</sup>.<sup><xref ref-type="bibr" rid="ref2">2</xref></sup> Data, thus far obtained, provide evidence that the ability of MnP to efficiently eliminate O<sub>2</sub><sup>•–</sup> closely parallels its therapeutic efficacy.<sup><xref ref-type="bibr" rid="ref2">2</xref></sup> As already noted in the <xref rid="notes-1" ref-type="notes">Introduction</xref>, this is due to the appropriate electron-deficiency of Mn site which favors reactions with nucleophiles, not only O<sub>2</sub><sup>•–</sup> but other species, some of which are listed above. SOD enzymes have the same thermodynamic property of metal site as MnPs, but steric hindrance imposed by large protein structure provides specificity toward O<sub>2</sub><sup>•–</sup>. Thus, their reactivity toward other species is a few orders of magnitude lower than that of SOD mimics.</p><fig id="fig6" position="float"><label>Figure 6</label><caption><p>Structure–activity relationships between the kinetic parameters, log <italic>k</italic><sub>cat</sub> (O<sub>2</sub><sup>•–</sup>) and log <italic>k</italic><sub>red</sub> (ONOO<sup>–</sup>), and thermodynamic parameters, <italic>E</italic><sub>1/2</sub> for Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple (mV vs NHE), and proton dissociation constant of first axial water, p<italic>K</italic><sub>a1</sub>. (A) log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) vs <italic>E</italic><sub>1/2</sub> for Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple; (B) p<italic>K</italic><sub>a1</sub> vs <italic>E</italic><sub>1/2</sub> for Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple; (C) log <italic>k</italic><sub>red</sub> (ONOO<sup>–</sup>) vs p<italic>K</italic><sub>a1</sub>; (D) log <italic>k</italic><sub>red</sub> (ONOO<sup>–</sup>) vs <italic>E</italic><sub>1/2</sub> for Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple; (E) log <italic>k</italic><sub>red</sub> (ONOO<sup>–</sup>) vs p<italic>K</italic><sub>a1</sub> and log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) vs <italic>E</italic><sub>1/2</sub> for Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple; (F) log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) vs log <italic>k</italic><sub>red</sub> (ONOO<sup>–</sup>) . Numerical values and experimental conditions for <italic>k</italic><sub>cat</sub> (O<sub>2</sub><sup>•–</sup>), <italic>k</italic><sub>red</sub> (ONOO<sup>–</sup>), p<italic>K</italic><sub>a1</sub>, and <italic>E</italic><sub>1/2</sub> (mV vs NHE) are given in Table <xref rid="tbl4" ref-type="other">4</xref>; empty squares in parts B, C, and E are estimated values: (1) MnTBAP<sup>3–</sup>, (2) MnTE-2-PyPhP<sup>5+</sup>, (3) MnTE-3-PyP<sup>5+</sup>, (4) MnTE-2-PyP<sup>5+</sup>, (5) MnTPhE-2-PyP<sup>5+</sup>, (6), MnTnHexOE-2-PyP<sup>5+</sup>, and (7) MnTnOct-2-PyP<sup>5+</sup>. The kinetics of Mn<sup>III</sup>P oxidation to O=Mn<sup>IV</sup>P, involved in reduction of ONOO<sup>–</sup> as well as reduction of lipid reactive species (see Figure <xref rid="fig8" ref-type="fig">8</xref>), relates to the thermodynamics of Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple. For explanation, see text; in brief, the electron transfer from Mn to ONOO<sup>–</sup> is preceded with ONOO<sup>–</sup> ligand binding which is dependent upon the electron-deficiency of Mn site. The latter is described by proton dissociation equilibrium of first axial water, p<italic>K</italic><sub>a1</sub>, which parallels <italic>E</italic><sub>1/2</sub> of Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple and is shown in part B.<sup><xref ref-type="bibr" rid="ref4">4</xref>,<xref ref-type="bibr" rid="ref76">76</xref></sup> There appears to be no difference between the <italic>E</italic><sub>1/2</sub> values for O=Mn<sup>IV</sup>P/Mn<sup>III</sup>P for various structurally diverse metalloporphyrins (<xref rid="notes-1" ref-type="notes">Supporting Information</xref> Table S1).</p></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-01329p_0006" id="gr6" position="float"/></fig><p>In order to mimic the kinetics and thermodynamics of the enzymatic catalysis of O<sub>2</sub><sup>•–</sup> dismutation (eqs <xref rid="eq1" ref-type="disp-formula">1</xref> and <xref rid="eq2" ref-type="disp-formula">2</xref>), the metal-centered reduction potential should be around the midpoint (∼+300 mV vs NHE) between the potential for the oxidation (−180 mV vs NHE) and reduction of O<sub>2</sub><sup>•–</sup> (+890 mV vs NHE).<sup><xref ref-type="bibr" rid="ref98">98</xref>−<xref ref-type="bibr" rid="ref100">100</xref></sup> Such <italic>E</italic><sub>1/2</sub> is controlled by <italic>ortho</italic>-positioned positively charged quaternary pyridyl nitrogens (pyridyls directly linked to porphyrin macrocycle at <italic>meso</italic> positions) which provide equal thermodynamics for both steps of the dismutation process and ensure favorable electrostatics for the approach of negatively charged O<sub>2</sub><sup>•–</sup> molecule to the metal site (eqs <xref rid="eq1" ref-type="disp-formula">1</xref> and <xref rid="eq2" ref-type="disp-formula">2</xref>). The porphyrin solvation impacts the <italic>E</italic><sub>1/2</sub> also.<sup><xref ref-type="bibr" rid="ref13">13</xref></sup> A perfect SOD mimic should oxidize (<italic>k</italic><sub>ox</sub>, eq <xref rid="eq1" ref-type="disp-formula">1</xref>) and reduce (<italic>k</italic><sub>red</sub>, eq <xref rid="eq2" ref-type="disp-formula">2</xref>) O<sub>2</sub><sup>•–</sup> with nearly identical rate constants which for SOD enzyme are <italic>k</italic><sub>ox</sub> ≈ <italic>k</italic><sub>red</sub> ∼10<sup>9</sup> M<sup>–1</sup> s<sup>–1</sup>. We have reported that this is indeed true for MnTE-2-PyP<sup>5+</sup> whose log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) = 7.76 and <italic>E</italic><sub>1/2</sub> = +228 mV versus NHE.<sup><xref ref-type="bibr" rid="ref101">101</xref></sup> It is presumably valid for cationic <italic>ortho</italic> Mn(III) pyridylporphyrins (MnTnOct-2-PyP<sup>5+</sup>, MnTnHexOE-2-PyP<sup>5+</sup>, and MnTPhE-2-PyP<sup>5+</sup>) also, as all of them have <italic>E</italic><sub>1/2</sub> values of ∼+300 mV versus NHE and exhibit a high ability to catalyze O<sub>2</sub><sup>•–</sup> dismutation, log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) ∼ 7.8 (Table <xref rid="tbl4" ref-type="other">4</xref>). In addition to eq <xref rid="eq1" ref-type="disp-formula">1</xref>, Mn<sup>III</sup>P could be reduced to Mn<sup>II</sup>P with ascorbate shown by eq <xref rid="eq1b" ref-type="disp-formula">1b</xref>, and reoxidized with O<sub>2</sub><sup>•–</sup> to Mn<sup>III</sup>P acting as superoxide reductase like rubredoxin oxidoreductase,<sup><xref ref-type="bibr" rid="ref102">102</xref></sup> a likely scenario <italic>in vivo</italic> due to the abundance of ascorbate.<disp-formula id="eq1"><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-01329p_m001" position="anchor"/><label>1</label></disp-formula><disp-formula id="eq1b"><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-01329p_m002" position="anchor"/><label>1b</label></disp-formula><disp-formula id="eq2"><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-01329p_m003" position="anchor"/><label>2</label></disp-formula></p><p>Mn<sup>III</sup>P reduces ONOO<sup>–</sup> via one-electron reaction giving rise to toxic <sup>•</sup>NO<sub>2</sub> (eq <xref rid="eq3" ref-type="disp-formula">3</xref>).<sup><xref ref-type="bibr" rid="ref5">5</xref></sup><italic>In vivo</italic>, the reduction of ONOO<sup>–</sup> by MnP is likely coupled to cellular reductants such as ascorbate.<sup><xref ref-type="bibr" rid="ref5">5</xref></sup> In a first step Mn<sup>III</sup>P gets reduced with ascorbate to Mn<sup>II</sup>P (eq <xref rid="eq1b" ref-type="disp-formula">1b</xref>). In a subsequent step Mn<sup>II</sup>P gets oxidized two-electronically to O=Mn<sup>IV</sup>P while benign nitrite, NO<sub>2</sub><sup>–</sup>, is formed. The rate constant for reaction <xref rid="eq4" ref-type="disp-formula">4</xref> has been estimated for MnTE-2-PyP<sup>5+</sup>, and is equal or higher than for the reaction given by eq <xref rid="eq3" ref-type="disp-formula">3</xref>.<sup><xref ref-type="bibr" rid="ref6">6</xref></sup> The O=Mn<sup>IV</sup>P is a highly oxidizing species. Its damage to biological targets is largely suppressed at the expense of cellular reductants as they readily reduce it to Mn<sup>III</sup>P.<sup><xref ref-type="bibr" rid="ref5">5</xref></sup><disp-formula id="eq3"><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-01329p_m004" position="anchor"/><label>3</label></disp-formula><disp-formula id="eq4"><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-01329p_m005" position="anchor"/><label>4</label></disp-formula></p><p>The <italic>E</italic><sub>1/2</sub> of Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple for the series of compounds studied varies from −194 to +340 mV versus NHE. Strong correlations have been found between <italic>E</italic><sub>1/2</sub> and <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) (eqs <xref rid="eq1" ref-type="disp-formula">1</xref> and <xref rid="eq2" ref-type="disp-formula">2</xref>; Figure <xref rid="fig6" ref-type="fig">6</xref>A). We observed earlier with Mn(III) <italic>N</italic>-alkylpyridylporphyrins,<sup><xref ref-type="bibr" rid="ref4">4</xref></sup> and here with new series of porphyrins, that <italic>k</italic><sub>red</sub>(ONOO<sup>–</sup>) correlates with <italic>E</italic><sub>1/2</sub> for Mn<sup>III</sup>P/Mn<sup>II</sup>P even though the reaction of MnP with peroxynitrite, studied in this work, involves the O=Mn<sup>IV</sup>P/Mn<sup>III</sup>P redox couple. This can be accounted for by a two-step process: (i) binding of ONOO<sup>–</sup> to the Mn site, and (ii) subsequent reduction of ONOO<sup>–</sup> yielding O=Mn<sup>IV</sup>P species. The first step is dependent upon the Mn site electron-deficiency, a property well-described by the <italic>E</italic><sub>1/2</sub> of Mn<sup>III</sup>P/Mn<sup>II</sup>P couple. This <italic>E</italic><sub>1/2</sub> has previously been reported to linearly correlate (i) with metal-free porphyrin protonation equilibria of its inner pyrrolic nitrogens,<sup><xref ref-type="bibr" rid="ref76">76</xref></sup> and (ii) with the protonation equilibria of axial waters of MnPs, depicted herein with the proton dissociation constant of first axial water, p<italic>K</italic><sub>a1</sub> (eq <xref rid="eq5" ref-type="disp-formula">5</xref> and Figure <xref rid="fig6" ref-type="fig">6</xref>B).<sup><xref ref-type="bibr" rid="ref4">4</xref></sup><disp-formula id="eq5"><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-01329p_m006" position="anchor"/><label>5</label></disp-formula></p><p>Therefore, the MnPs of more positive <italic>E</italic><sub>1/2</sub> for Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple and lower p<italic>K</italic><sub>a1</sub> values (Figure <xref rid="fig6" ref-type="fig">6</xref>B) are more electron-deficient and favor binding of an electron-rich ligand (ONOO<sup>–</sup> in this case) which in turn gives rise to higher <italic>k</italic><sub>red</sub>(ONOO<sup>–</sup>) (Figure <xref rid="fig6" ref-type="fig">6</xref>C).<sup><xref ref-type="bibr" rid="ref2">2</xref></sup> Such data explain why the log <italic>k</italic><sub>red</sub>(ONOO<sup>–</sup>) correlates as well with <italic>E</italic><sub>1/2</sub> for Mn<sup>III</sup>P/Mn<sup>II</sup>P (Figure <xref rid="fig6" ref-type="fig">6</xref>D) as does log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) (Figure <xref rid="fig6" ref-type="fig">6</xref>A). It thus explains why there is a linear relationship between log <italic>k</italic><sub>red</sub>(ONOO<sup>–</sup>) and <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) (Figure <xref rid="fig6" ref-type="fig">6</xref>F). The second step of the reaction of Mn<sup>III</sup>P with ONOO<sup>–</sup> is related to the <italic>E</italic><sub>1/2</sub> values for the O=Mn<sup>IV</sup>P/Mn<sup>III</sup>P couple. These, as well as those for (O)<sub>2</sub>Mn<sup>V</sup>P/Mn<sup>III</sup>P redox couples, are essentially identical for different Mn and Fe porphyrins.<sup><xref ref-type="bibr" rid="ref5">5</xref>,<xref ref-type="bibr" rid="ref103">103</xref>−<xref ref-type="bibr" rid="ref110">110</xref></sup> All relevant reduction potentials are listed in <xref rid="notes-1" ref-type="notes">Supporting Information</xref> (Table S1).<sup><xref ref-type="bibr" rid="ref4">4</xref>,<xref ref-type="bibr" rid="ref5">5</xref>,<xref ref-type="bibr" rid="ref76">76</xref>,<xref ref-type="bibr" rid="ref111">111</xref>,<xref ref-type="bibr" rid="ref112">112</xref></sup> For example, the <italic>E</italic><sub>1/2</sub> values for Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple of MnTE-2-PyP<sup>5+</sup>, MnTE-3-PyP<sup>5+</sup>, and MnTnBu-2-PyP<sup>5+</sup> differ by up to 176 mV, while <italic>E</italic><sub>1/2</sub> for O=Mn<sup>IV</sup>P/Mn<sup>III</sup>P redox couple values are +509, +529, and +509 mV versus SHE at pH 11, respectively (<xref rid="notes-1" ref-type="notes">Supporting Information</xref> Table S1).<sup><xref ref-type="bibr" rid="ref112">112</xref></sup> Further, <italic>E</italic><sub>1/2</sub> values for O=Mn<sup>IV</sup>P/Mn<sup>III</sup>P of MnTM-2-PyP<sup>5+</sup>, MnTM-3-PyP<sup>5+</sup>, and MnTM-4-PyP<sup>5+</sup> are +540, +526, and +532 mV versus NHE at pH 11, while <italic>E</italic><sub>1/2</sub> values for their Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couples differ by up to 168 mV (Table <xref rid="tbl4" ref-type="other">4</xref> and <xref rid="notes-1" ref-type="notes">Supporting Information</xref> Table S1).<sup><xref ref-type="bibr" rid="ref5">5</xref></sup> Finally, <italic>E</italic><sub>1/2</sub> values for the (O)<sub>2</sub>Mn<sup>V</sup>P/Mn<sup>III</sup>P redox couple for MnTM-4-PyP<sup>5+</sup>, MnTM-2-PyP<sup>5+</sup>, and MnTDM-2-ImP<sup>5+</sup> (Mn(III) <italic>meso</italic>-tetrakis(<italic>N</italic>,<italic>N</italic>′-dimethylimidazolium-2-yl)porphyrin) at pH 11 are all around +800 mV versus NHE, while their <italic>E</italic><sub>1/2</sub> values for the Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple differ by up to 260 mV (<xref rid="notes-1" ref-type="notes">Supporting Information</xref> Table S1).<sup><xref ref-type="bibr" rid="ref111">111</xref></sup> Thus, the only factor different among these MnPs with respect to their oxidation to oxo-Mn species is the binding of highly oxidizing species (such as ONOO<sup>–</sup>, H<sub>2</sub>O<sub>2</sub>, and lipid reactive species) to Mn, a rate limiting step dependent upon the electron deficiency/richness of Mn site and thus best characterized with proton dissociation constants of either porphyrin pyrrolic nitrogens, or axial waters,<sup><xref ref-type="bibr" rid="ref4">4</xref>,<xref ref-type="bibr" rid="ref76">76</xref></sup> or the <italic>E</italic><sub>1/2</sub> value of the Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple. Of note, as already indicated, the oxidation of MnPs with ONOO<sup>–</sup><italic>in vivo</italic> may involve the O=Mn<sup>IV</sup>P/Mn<sup>II</sup>P redox couple (eq <xref rid="eq4" ref-type="disp-formula">4</xref>), as Mn<sup>III</sup>Ps would likely be readily reduced to Mn<sup>II</sup>Ps by cellular reductants. The <italic>E</italic><sub>1/2</sub> of the O=Mn<sup>IV</sup>P/Mn<sup>II</sup>P redox couple is controlled in part by the Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple and thus differs among MnPs, for MnTE-2-PyP<sup>5+</sup>, MnTE-3-PyP<sup>5+</sup>, and MnTnBuOE-2-PyP<sup>5+</sup>. The <italic>E</italic><sub>1/2</sub> (O=Mn<sup>IV</sup>P/Mn<sup>II</sup>P) values were calculated to be +317, +253, and +343 mV versus SHE,<sup><xref ref-type="bibr" rid="ref112">112</xref></sup> resulting in a higher driving force and therefore thermodynamically favoring the two-electron reduction of ONOO<sup>−</sup>.<sup><xref ref-type="bibr" rid="ref112">112</xref></sup></p><p>An interesting phenomenon has been observed when two plots [log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) vs <italic>E</italic><sub>1/2</sub> and log <italic>k</italic><sub>red</sub>(ONOO<sup>–</sup>) vs p<italic>K</italic><sub>a1</sub>] are overlapped (Figure <xref rid="fig6" ref-type="fig">6</xref>E). The differences observed between the highest and the lowest rate constants for O<sub>2</sub><sup>•–</sup> dismutation and ONOO<sup>–</sup> reduction at identical <italic>E</italic><sub>1/2</sub> values are 4.55 [Δ<italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>)] and 2.13 [Δ<italic>k</italic><sub>red</sub>(ONOO<sup>–</sup>)]. The diagram supports the fact that the reported beneficial effects of MnTBAP<sup>3–</sup><sup><xref ref-type="bibr" rid="ref18">18</xref>−<xref ref-type="bibr" rid="ref42">42</xref></sup> could be rather attributed to its peroxynitrite reducing ability, and not superoxide scavenging.</p><p>The therapeutic effects observed with cationic <italic>ortho</italic> Mn(III) <italic>N</italic>-substituted pyridylporphyrins cannot be safely assigned to a specific reactive species. Implementing multiple approaches, including pharmacological and genetic, along with direct measurements of MnP subcellular localization may allow us to safely identify the location of MnP within tissue and cell/cellular fragments and the nature of reactive species involved in its mode of action.</p></sec><sec id="sec3.3"><title>MnP-Catalyzed Ascorbate Oxidation</title><p>Understanding the reactivity of MnPs toward ascorbate is biologically relevant due to the: (i) high intracellular ascorbate concentrations; (ii) high ability of MnPs to oxidize ascorbate; (iii) coupling of ascorbate with O<sub>2</sub><sup>•–</sup> and ONOO<sup>–</sup>; reduction of Mn<sup>III</sup>P to Mn<sup>II</sup>P with ascorbate <italic>in vivo</italic> is likely a first step in its redox cycling with O<sub>2</sub><sup>•–</sup> and ONOO<sup>–</sup> - in such scenario MnP acts as O<sub>2</sub><sup>•–</sup> reductase rather than superoxide dismutase;<sup><xref ref-type="bibr" rid="ref113">113</xref></sup> and (iv) therapeutic potential of MnP/ascorbate as a ROS generator for tumor therapy.</p><p>We have herein demonstrated that the ability of MnPs to catalyze ascorbate oxidation (eq <xref rid="eq1b" ref-type="disp-formula">1b</xref>), described as initial rate, <italic>v</italic><sub>0</sub>(HA<sup>–</sup>), depends upon the electron deficiency of the metal center, <italic>E</italic><sub>1/2</sub>. The bell-shape curve was established for MnPs in the range <italic>E</italic><sub>1/2</sub> −194 to +340 mV versus NHE (Figure <xref rid="fig7" ref-type="fig">7</xref>). The highest rate of ascorbate oxidation was reached for MnTE-2-PyP<sup>5+</sup> at <italic>E</italic><sub>1/2</sub> of +228 mV versus NHE, and dropped afterward. The <italic>k</italic><sub>cat</sub> (O<sub>2</sub><sup>•–</sup>) and <italic>k</italic><sub>red</sub> (ONOO<sup>–</sup>), though, reached a plateau at MnTE-2-PyP<sup>5+</sup> but did not drop afterward (Figure <xref rid="fig6" ref-type="fig">6</xref>A,D). The ratio of the stabilities of Mn +2 and +3 oxidation states has larger impact on the catalysis of ascorbate oxidation than it has on O<sub>2</sub><sup>•–</sup> dismutation, where the interplay of solvation and lipophilicity of alkyl chains results in similarly high <italic>k</italic><sub>cat</sub> (O<sub>2</sub><sup>•–</sup>) of MnTnOct-2-PyP<sup>5+</sup> and MnTE-2-PyP<sup>5+</sup> (Figure <xref rid="fig6" ref-type="fig">6</xref>A)<sub>.</sub><sup><xref ref-type="bibr" rid="ref13">13</xref></sup></p><fig id="fig7" position="float"><label>Figure 7</label><caption><p>MnP-catalyzed ascorbate oxidation. (A) Kinetic traces for different MnPs. (B) Initial rates of ascorbate oxidation/consumption expressed in nM s<sup>–1</sup>. (C) Regions where <italic>E</italic><sub>1/2</sub> Mn is stabilized in +2 and +3 oxidation states; MnTE-2-PyP<sup>5+</sup> is the most optimized MnP in terms of H<sub>2</sub>O<sub>2</sub> production. It has equally stabilized +2 and +3 oxidation states.<sup><xref ref-type="bibr" rid="ref101">101</xref></sup> It gets readily reduced with ascorbate but also reoxidized back to Mn<sup>III</sup>P with either O<sub>2</sub><sup>•–</sup> or O<sub>2</sub>, whichever is <italic>in vivo</italic> in excess. Those MnPs with negative potentials do not favor reduction, while those with too positive potential do not favor reoxidation of Mn<sup>II</sup>P. (D) Redox cycling of MnP with ascorbate, which involves the reoxidation of Mn<sup>II</sup>P with O<sub>2</sub> (preferred over O<sub>2</sub><sup>•–</sup> due to its higher <italic>in vivo</italic> levels) to close the catalytic cycle. The conditions are 5 μM MnP, 0.15 mM sodium ascorbate at pH 7.4 maintained with 0.05 M Tris buffer with 5 mM EDTA, (25 ± 1) °C. The numerical assignments in part C are identical to those described in the Figure <xref rid="fig6" ref-type="fig">6</xref> caption.</p></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-01329p_0007" id="gr7" position="float"/></fig><p>We have shown that <italic>k</italic><sub>red</sub>(O<sub>2</sub><sup>•–</sup>) = <italic>k</italic><sub>ox</sub>(O<sub>2</sub><sup>•–</sup>) for MnTE-2-PyP<sup>5+</sup> with <italic>E</italic><sub>1/2</sub> = +228 mV vs NHE.<sup><xref ref-type="bibr" rid="ref101">101</xref></sup> Thus, both +3 and +2 oxidation states are equally stabilized. With MnTnOct-2-PyP<sup>5+</sup>, at +340 mV versus NHE, the Mn +2 oxidation state is more stabilized and disfavors reoxidation with either O<sub>2</sub> or O<sub>2</sub><sup>•–</sup> (whichever species predominates <italic>in vivo</italic> in MnP neighborhood), suppressing in turn the cycling of MnP with ascorbate. In addition to ascorbate, glutathione and cysteine (and likely protein thiols based on their exposure) may reduce MnP also. The magnitude of the reoxidation of Mn<sup>II</sup>P, resulting eventually in a peroxide production, may distinguish which compound would produce higher levels of H<sub>2</sub>O<sub>2</sub> and be more efficient in employing it subsequently in oxidation of biological targets. The relevance of the differential impact of MnP/peroxide on cancer versus normal cell is discussed in the next paragraph.</p></sec><sec id="sec3.4"><title>Differential Impact of MnP/Peroxide on Cancer versus Normal Cell</title><p>The interaction of MnP with peroxide produced in its cycling with ascorbate (or when combined with radiation<sup><xref ref-type="bibr" rid="ref114">114</xref></sup> or other chemotherapies such as dexamethasone)<sup><xref ref-type="bibr" rid="ref10">10</xref>,<xref ref-type="bibr" rid="ref11">11</xref>,<xref ref-type="bibr" rid="ref114">114</xref>,<xref ref-type="bibr" rid="ref115">115</xref></sup> will eventually cause cancer cell death, while either sparing or protecting normal cell. While little is still known on the differential biology of cancer versus normal cells, the prevailing opinion is that this is largely based on the differential redox environments of those cells; in turn, the differential impact of MnP/peroxide is dependent upon such differences also. It has been established that cancer relative to normal cell is under increased oxidative stress. While cancer cell often up-regulates MnSOD in efforts to control oxidative stress, this seems frequently not to be accompanied by up-regulation of appropriate levels of peroxide removing enzymes.<sup><xref ref-type="bibr" rid="ref116">116</xref>−<xref ref-type="bibr" rid="ref120">120</xref></sup> Thus, an increase in circulating MnSOD frequently results in increased peroxide levels and is positively correlated with tumor reoccurrence.<sup><xref ref-type="bibr" rid="ref30">30</xref>,<xref ref-type="bibr" rid="ref83">83</xref></sup> Moreover, down-regulation of a number of peroxide removing enzymes, such as thioredoxin reductase, peroxyredoxin, catalase, and glutathione peroxide, was reported;<sup><xref ref-type="bibr" rid="ref116">116</xref>−<xref ref-type="bibr" rid="ref120">120</xref></sup> in turn, the peroxide levels get increased as tumor progresses. Malignant properties were reportedly reversed by up-regulation of catalase.<sup><xref ref-type="bibr" rid="ref121">121</xref></sup> Under such conditions of high oxidative stress, any addition of a redox-active compound such as MnP, that further enhances the levels of RS via cycling with cellular reductants such as ascorbate (added exogenously), will further increase the levels of superoxide/peroxide and will enhance cancer cell death, the observation we have frequently demonstrated.<sup><xref ref-type="bibr" rid="ref2">2</xref>,<xref ref-type="bibr" rid="ref77">77</xref>,<xref ref-type="bibr" rid="ref78">78</xref></sup> Such enhancement of oxidative stress via radiation or chemotherapy has been regularly used as therapeutic modality. Thus, the enhancement of the anticancer effect in a lymphoma cellular study by the joint action of MnP and dexamethasone has been reported.<sup><xref ref-type="bibr" rid="ref11">11</xref></sup> The cancer cell killing by MnP in a lymphoma model occurred via MnP/peroxide-driven oxidation of thiols of antiapoptotic transcription factor NF-κB with subsequent suppression of its transcription.<sup><xref ref-type="bibr" rid="ref10">10</xref>,<xref ref-type="bibr" rid="ref11">11</xref>,<xref ref-type="bibr" rid="ref55">55</xref>,<xref ref-type="bibr" rid="ref115">115</xref></sup> The inactivation of mitochondrial complexes I and III, and the impact on the glycolysis by MnP/dexamethasone, has been implicated in cancer cell death also.<sup><xref ref-type="bibr" rid="ref10">10</xref></sup></p><p>In a normal cell, though, the dismutation of O<sub>2</sub><sup>•–</sup> catalyzed by MnP and MnP cycling with ascorbate, both giving rise to peroxide, has no significant toxic impact as peroxide is readily removed by abundant peroxide-removing enzymes maintaining physiological redox balance. If anything, and in diseased cell, the MnP may suppress excessive inflammation which would have otherwise lead to death of a normal cell. This could occur via suppression of NF-κB transcription by oxidation of its thiols, yet to a limited extent as levels of peroxide are much lower than those in a cancer cell. We have indeed frequently reported on the differential effects of MnP on normal versus cancer cells.<sup><xref ref-type="bibr" rid="ref17">17</xref>,<xref ref-type="bibr" rid="ref114">114</xref>,<xref ref-type="bibr" rid="ref122">122</xref>,<xref ref-type="bibr" rid="ref123">123</xref></sup> Please see for further discussion Miriyala et al.<sup><xref ref-type="bibr" rid="ref83">83</xref></sup>and Batinic-Haberle et al.<sup><xref ref-type="bibr" rid="ref2">2</xref></sup></p><p>The type of cell, cancer or normal, will control the suitability of particular MnP as a therapeutic of choice: MnTE-2-PyP<sup>5+</sup> and MnTE-3-PyP<sup>5+</sup> would be preferred when applied along with therapeutic doses of ascorbate to destroy cancer cells due to the highest rate of catalysis of ascorbate oxidation with subsequent peroxide production. Equally active and more lipophilic MnPs, such as MnTnOct-2-PyP<sup>5+</sup> and MnTnHexOE-2-PyP<sup>5+</sup>, may be selected for the application in normal tissue oxidative stress related models as they are less efficacious in catalyzing ascorbate oxidation. However, much is still needed to fully understand therapeutic effects of redox-active drugs as they depend not only on their redox properties but on cellular and subcellular accumulation and colocalization with targeted species, many of those likely not yet identified.</p></sec><sec id="sec3.5"><title>Inhibition of Lipid Peroxidation by MnPs</title><p>Lipid peroxidation, i.e., the oxidative damage to polyunsaturated fatty acids, is initiated by the attack of reactive oxygen species, such as hydrogen peroxide, singlet oxygen, and hydroxyl radical.<sup><xref ref-type="bibr" rid="ref124">124</xref>−<xref ref-type="bibr" rid="ref126">126</xref></sup> This gives rise to lipid peroxyl, ROO<sup>•</sup>, and alkoxyl RO<sup>•</sup> radicals and lipid hydroperoxides which propagate the lipid peroxidation. As most of the proteins are closely associated with membranes, the lipid peroxidation damages not only lipids but proteins also. Lipid peroxidation is involved in pathogenesis of a number of diseases such as cancer, atherosclerosis, diabetes, Alzheimer’s disease, and Parkinson’s disease, etc.<sup><xref ref-type="bibr" rid="ref127">127</xref>−<xref ref-type="bibr" rid="ref132">132</xref></sup> The peroxidation of arachidonic, linolenic, and docosahexanoic acids gives rise to malondialdehyde, MDA. MDA is also formed enzymatically during eicosanoid metabolism.<sup><xref ref-type="bibr" rid="ref133">133</xref></sup> Due to the intrinsic aldehyde instability, the MDA is reactive toward DNA and amino acids, in particular lysine. HPLC-based thiobarbituric acid (TBA)-assay eliminates most of the interference that plagues the simple TBA assay and is therefore useful in screening the biological tissues on lipid peroxidation.<sup><xref ref-type="bibr" rid="ref133">133</xref></sup> Figure <xref rid="fig8" ref-type="fig">8</xref> shows the magnitude of spontaneous lipid peroxidation affected by MnPs and measured as MDA with HPLC method. The data are expressed as percentage of peroxidation in control samples, which was taken as 100%.</p><fig id="fig8" position="float"><label>Figure 8</label><caption><p>Attenuation of lipid peroxidation by various MnPs as a function of their of metal-centered reduction potentials. (A) The ability of MnPs to prevent lipid peroxidation of rat brain homogenates in terms of malondialdehyde, MDA, expressed as % of control (taken as 100% of lipid peroxidation) measured by HPLC method. Butylated hydroxytoluene (BHT) was used as positive control which prevented ∼90% of lipid peroxidation. The <italic>E</italic><sub>1/2</sub> of Mn<sup>III</sup>P/Mn<sup>II</sup>P governs the ability of MnPs to attenuate lipid peroxidation. The possible reasons why the oxidation of Mn<sup>III</sup>P with lipid reactive species relates to the <italic>E</italic><sub>1/2</sub> of Mn<sup>III</sup>P/Mn<sup>II</sup>P have been discussed in Figure <xref rid="fig6" ref-type="fig">6</xref> and in text in the <xref rid="sec3.2" ref-type="other">Structure–Activity Relationships</xref> section. The bulkiness of the molecule, i.e., the steric hindrance toward lipid reactive species plays a minimal role. The impact of <italic>E</italic><sub>1/2</sub> was better visualized in plot B where the percent of lipid peroxidation was plotted vs <italic>E</italic><sub>1/2</sub> at 5 μM MnP. At that concentration, no inhibition of lipid peroxidation was observed with MnTBAP<sup>3–</sup> (1) and MnTE-2-PyPhP<sup>5+</sup> (2). As <italic>E</italic><sub>1/2</sub> increases from MnTBAP<sup>3–</sup> and MnTE-2-PyPhP<sup>5+</sup> to MnTE-3-PyP<sup>5+</sup>, the inhibition of lipid peroxidation increases (3) and reaches maximum at ∼+300 mV vs NHE with MnTE-2-PyP<sup>5+</sup> (4), MnTPhE-2-PyP<sup>5+</sup> (5), and MnTnHexOE-2-PyP<sup>5+</sup> (6). The somewhat lower inhibition with MnTnOct-2-PyP<sup>5+</sup> (7) is likely due to the steric hindrance imposed by long <italic>N</italic>-pyridyl substituents toward the approach of lipid reactive species.</p></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-01329p_0008" id="gr8" position="float"/></fig><p>The data in Figure <xref rid="fig8" ref-type="fig">8</xref> demonstrate a direct link between the <italic>E</italic><sub>1/2</sub> of Mn<sup>III</sup>P/Mn<sup>II</sup>P and log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>), i.e., SOD-like activity, and the ability of MnPs to suppress lipid peroxidation. As <italic>E</italic><sub>1/2</sub> parallels the log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>), it in turn correlates with the ability of MnP to inhibit lipid peroxidation. Such a relationship [among the <italic>E</italic><sub>1/2</sub>, log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>), and inhibition of lipid peroxidation] is related to the electron deficiency of Mn site which controls ligand binding (in this case binding of lipid reactive species). See related discussion under Figure <xref rid="fig6" ref-type="fig">6</xref> and in the text in the <xref rid="sec3.2" ref-type="other">Structure–Activity Relationships</xref> section.</p><p>MnTBAP<sup>3–</sup>, with very negative <italic>E</italic><sub>1/2</sub> = −194 mV vs NHE, did not suppress lipid peroxidation at up to 200 μM concentration. With less negative <italic>E</italic><sub>1/2</sub> = −65 mV versus NHE, MnTE-2-PyPhP<sup>5+</sup> fully suppressed the lipid peroxidation at ≥50 μM. Compounds with <italic>E</italic><sub>1/2</sub> > +228 mV versus NHE were the most efficacious. Although MnTE-2-PyP<sup>5+</sup> and MnTE-3-PyP<sup>5+</sup> demonstrate similar superoxide scavenging ability in SOD-deficient <italic>E. coli</italic>(<xref ref-type="bibr" rid="ref67">67</xref>) and <italic>S. cerevisiae</italic> assays, a clear difference between two of them was observed in lipid peroxidation assay. As cells are disrupted during preparation of brain homogenates, <italic>E</italic><sub>1/2</sub> of MnPs, but not lipophilicity/bioavailability, gains controls over the magnitude of lipid peroxidation. At 1 μM concentration, MnTE-2-PyP<sup>5+</sup> was very efficacious, whereas MnTE-3-PyP<sup>5+</sup> did not inhibit lipid peroxidation. Despite similar <italic>E</italic><sub>1/2</sub> values, the efficacy of larger molecule, MnTPhE-2-PyP<sup>5+</sup>, was slightly lower relative to MnTE-2-PyP<sup>5+</sup> and MnTnHexOE-2-PyP<sup>5+</sup> (Figure <xref rid="fig8" ref-type="fig">8</xref>A). This effect may be due to the steric hindrance toward lipid reactive species imposed by the bulkier substituents of MnTPhE-2-PyP<sup>5+</sup> and MnTnHexOE-2-PyP<sup>5+</sup> (Figure <xref rid="fig8" ref-type="fig">8</xref>A). The oxygen-bearing MnTnHexOE-2-PyP<sup>5+</sup> demonstrated higher potency than MnTnOct-2-PyP<sup>5+</sup>. We have assigned this to the favorable interactions of polar oxygen atoms with lipid reactive species, guiding them toward Mn site. The impact of higher <italic>E</italic><sub>1/2</sub> of MnTnOct-2-PyP<sup>5+</sup> may not be excluded.</p></sec><sec id="sec3.6"><title>Effect of MnPs on the Aerobic Growth of SOD-Deficient <italic>S. cerevisiae</italic></title><p>The aerobic growth of SOD-deficient <italic>S. cerevisiae</italic>, which lacks CuZnSOD, is an excellent <italic>in vivo</italic> model for the evaluation of the therapeutic potential of relatively lipophilic compounds within a class of water-soluble MnPs.<sup><xref ref-type="bibr" rid="ref67">67</xref></sup> It is also O<sub>2</sub><sup>•–</sup> specific <italic>in vivo</italic> model of oxidative stress.</p><p>The combined impact of <italic>E</italic><sub>1/2</sub>, lipophilicity, and bulkiness (size, shape) was demonstrated in yeast study. The most lipophilic compounds MnTnOct-2-PyP<sup>5+</sup> and MnTnHexOE-2-PyP<sup>5+</sup> are the most efficacious MnPs, presumably due to their higher accumulation in the cell and the higher <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) values relative to other MnPs. At 1–5 μM both compounds allow SOD-deficient yeast to grow as well as wild type (Figure <xref rid="fig9" ref-type="fig">9</xref>). At higher concentrations both MnPs become toxic. The lipophilic, but bulkier, MnTPhE-2-PyP<sup>5+</sup> has lower accumulation and is thus less efficient, but less toxic also. As seen before in <italic>E. coli</italic> assay,<sup><xref ref-type="bibr" rid="ref2">2</xref></sup> the higher lipophilicity of MnTE-3-PyP<sup>5+</sup> compensates for its lower <italic>E</italic><sub>1/2</sub> and lower <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) and is thus equally efficacious as MnTE-2-PyP<sup>5+</sup>. Both MnTE-3-PyP<sup>5+</sup> and MnTE-2-PyP<sup>5+</sup> become efficacious at concentrations above 5 μM. The compounds with very negative <italic>E</italic><sub>1/2</sub> lacking SOD-like activity, MnTBAP<sup>3–</sup> and MnTE-2-PyPhP<sup>5+</sup>, are not protective to SOD-deficient <italic>S. cerevisiae</italic>.</p><fig id="fig9" position="float"><label>Figure 9</label><caption><p>Aerobic growth of the wild type SOD-proficient (EG 103) and SOD-deficient (EG118) <italic>S. cerevisiae</italic> in the presence and absence of MnPs. Yeast grew in a restricted medium where the impact of MnPs is enhanced. All samples were run in triplicate. Growth was followed turbidimetrically by measuring the absorbance at 600 nm using ELISA reader. Inset: The lipophilicity, <italic>R</italic><sub><italic>f</italic></sub>, and the SOD-like activity, described by log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>), are plotted to demonstrate their impact on the growth of SOD-deficient yeast. The plots show that compounds of high lipophilicity (bioavailability) and high log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) are the most efficacious in protecting SOD-deficient yeast and in turn bear the highest therapeutic potential.</p></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-01329p_0009" id="gr9" position="float"/></fig></sec></sec><sec id="sec4"><title>Concluding Remarks</title><p>A series of MnPs with a wide range of metal-centered reduction potentials (<italic>E</italic><sub>1/2</sub>, from −190 to +340 mV vs NHE) and lipophilicities (log <italic>P</italic><sub>OW</sub>, from −7.67 to −1.67) have been synthesized and evaluated for their redox activities [<italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>), <italic>k</italic><sub>red</sub>(ONOO<sup>–</sup>), and <italic>v</italic><sub>0</sub>(HA<sup>–</sup> oxidation to A<sup>•–</sup>, ascorbyl radical)] and <italic>in vitro</italic> (lipid peroxidation) and <italic>in vivo</italic> therapeutic potential (aerobic growth of SOD-deficient <italic>S. cerevisiae</italic>). Those porphyrins could be divided in 3 groups by their <italic>E</italic><sub>1/2</sub> [which parallels log <italic>k</italic><sub>cat</sub> (O<sub>2</sub><sup>•–</sup>)] and lipophilicity as shown in Figure <xref rid="fig10" ref-type="fig">10</xref>.</p><fig id="fig10" position="float"><label>Figure 10</label><caption><p>Schematic representations of the dominant properties of MnPs which control their therapeutic potential: <italic>E</italic><sub>1/2</sub>, log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•<bold>–</bold></sup>), and log <italic>P</italic><sub>OW</sub>. MnPs could be divided into 3 groups: (1) lipophilic and SOD-inactive [(of negative <italic>E</italic><sub>1/2</sub> and log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>)], the latter being lower than 5.7, situated in the right part of the figure, (2) lipophilic and SOD-active, situated in the left part of the figure, and (3) hydrophilic and SOD-active situated in the middle, i.e., in the minimum of the lipophilicity plot (Figure <xref rid="fig5" ref-type="fig">5</xref>). Those MnPs that are lipophilic, SOD-active, and of positive <italic>E</italic><sub>1/2</sub> are the most efficacious in <italic>in vivo S. cerevisiae</italic> assay and therefore bear the highest therapeutic potential.</p></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-01329p_0010" id="gr10" position="float"/></fig><p>Two of those groups contain MnPs which are similarly lipophilic, yet in one group are the MnPs with negative <italic>E</italic><sub>1/2</sub> and in the other with positive <italic>E</italic><sub>1/2</sub>. MnPs with negative <italic>E</italic><sub>1/2</sub>, despite high bioavailability, were inferior or nonefficacious in both <italic>in vitro</italic> and <italic>in vivo</italic> assays. MnPs with positive <italic>E</italic><sub>1/2</sub> are the most efficacious ones. The third group comprises very hydrophilic <italic>ortho</italic> and <italic>meta</italic> MnTE-2-PyP<sup>5+</sup> and MnTE-3-PyP<sup>5+</sup> which are efficacious but several-fold less than lipophilic analogs of equal <italic>E</italic><sub>1/2</sub>. In summary our observations are the following: (1) The <italic>E</italic><sub>1/2</sub> for the Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple, dominated by the electron deficiency of porphyrin and its metal site, not only controls the ability of MnPs to eliminate O<sub>2</sub><sup>•–</sup> and ONOO<sup>–</sup> but also the ability to prevent lipid peroxidation. The SOD-like activity appears to be proportional to the efficacy of MnP in preventing lipid peroxidation. The MnPs with highly positive <italic>E</italic><sub>1/2</sub> (≥+228) and high SOD-like activity demonstrated strong inhibition of lipid peroxidation. MnTBAP<sup>3–</sup> cannot inhibit lipid peroxidation, and MnTE-2-PyPhP<sup>5+</sup> shows activity only at high concentration (≥50 μM). (2) The catalysis of ascorbate oxidation which involves reduction of Mn<sup>III</sup>P is controlled by the thermodynamics of Mn<sup>III</sup>P/Mn<sup>II</sup>P redox couple. A bell shaped curve was observed for the MnP-driven catalysis of ascorbate oxidation with the highest <italic>v</italic><sub>0</sub>(HA<sup>–</sup>) observed at <italic>E</italic><sub>1/2</sub> = +228 mV versus NHE (MnTE-2-PyP<sup>5+</sup>). The data on ascorbate oxidation by MnP leading to a cytotoxic peroxide production (Figure <xref rid="fig7" ref-type="fig">7</xref>) indicate the superiority of MnTE-2-PyP<sup>5+</sup>, while the data on lipid peroxidation and <italic>S. cerevisiae</italic> suggest that lipophilic MnTnOct-2-PyP<sup>5+</sup> and MnTnHexOE-2-PyP<sup>5+</sup> may be superior as therapeutics. Hence, the latter compounds may be preferably applied in normal tissue injuries with oxidative stress background, whereas MnTE-2-PyP<sup>5+</sup>, in combination with exogenous ascorbate, would be a therapeutic of choice for tumor treatment. (3) Only MnPs which disproportionate O<sub>2</sub><sup>•–</sup> with a rate constant higher than the one for noncatalyzed, O<sub>2</sub><sup>•–</sup> self-dismutation, i.e., log <italic>k</italic><sub>self-dismutation</sub>(O<sub>2</sub><sup>•–</sup>) = 5.70, mimic the SOD enzyme in protecting SOD-deficient yeast. MnTE-2-PyPhP<sup>5+</sup> [log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) = 5.55] or MnTBAP<sup>3–</sup> [log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) = 3.16] did not show any beneficial effect as they are not true SOD mimics. While <italic>E</italic><sub>1/2</sub> controls the efficacy of MnPs in aqueous solution [i.e., log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>) and log <italic>k</italic><sub>red</sub>(ONOO<sup>–</sup>)], the lipophilicity plays critical role <italic>in vivo</italic> also, as it governs the cellular and intracellular distribution of MnPs. Therefore, the compounds of somewhat lower SOD-like potency, such as MnTE-3-PyP<sup>5+</sup>, still support the yeast growth as good as the MnTE-2-PyP<sup>5+</sup> of higher log <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>), as their cellular uptake is enhanced (Figure <xref rid="fig9" ref-type="fig">9</xref>). (4) Enhanced toxicity of MnPs to SOD-deficient yeast is observed (Figure <xref rid="fig9" ref-type="fig">9</xref>) with very lipophilic MnPs as they accumulate within a cell to higher levels and tend to localize in membranes disrupting their integrity. While fully protective in the region 1–5 μM, the MnTnOct-2-PyP<sup>5+</sup> and MnTnHexOE-2-PyP<sup>5+</sup> were already toxic at 20 μM. While MnTnHexOE-2-PyP<sup>5+</sup> is a better inhibitor of lipid peroxidation, it is somewhat inferior to MnTnOct-2-PyP<sup>5+</sup> in protecting <italic>S. cerevisiae</italic> which is likely due to the polar interactions between oxygen atoms and membrane structures. Jointly, high <italic>k</italic><sub>cat</sub>(O<sub>2</sub><sup>•–</sup>), high lipophilicity, and lower bulkiness contribute to the therapeutic potential of MnPs in <italic>S. cerevisiae</italic>. (5) We have originally developed MnPs as SOD mimics. Over the past decade we have shown that MnPs are involved in other actions, some of which may even predominate <italic>in vivo</italic>. Yet, thus far the experimental evidence (provided herein and elsewhere) indicates that the higher the SOD-like activity, the higher the therapeutic potential MnPs possess, even when reactions in question do not involve O<sub>2</sub><sup>•–</sup> elimination. We can therefore safely conclude that the modification of a porphyrin molecule to enhance its SOD-like activity may still comprise the best experimental strategy in the design of redox-active drugs. (6) None of the data obtained here on MnTBAP<sup>3–</sup> explain therapeutic effects reported elsewhere. Yet its <italic>in vivo</italic> efficacy justifies its future exploration. Its ONOO<sup>–</sup>-related chemistry is ∼100-fold slower than that of Mn(III) <italic>N</italic>-substituted pyridylporphyrins (Table <xref rid="tbl4" ref-type="other">4</xref>).<sup><xref ref-type="bibr" rid="ref2">2</xref></sup> When compared to cationic pyridylporphyrins, MnTBAP<sup>3–</sup> cannot be <italic>in vivo</italic> reduced to Mn<sup>II</sup>P by cellular reductants, such as ascorbate, in order to produce peroxide in subsequent reoxidation step (Figure <xref rid="fig7" ref-type="fig">7</xref>). No reactivity toward lipid reactive species was observed (Figure <xref rid="fig8" ref-type="fig">8</xref>). The lack of its SOD-like activity was proven here with SOD-deficient <italic>S. cerevisiae</italic> (Figure <xref rid="fig9" ref-type="fig">9</xref>). It has insignificant catalase-like activity (Maia et al., submitted). Reactivity toward H<sub>2</sub>O<sub>2</sub> may thus not play a major role in its actions. The role of reactive nitrogen species, other than ONOO<sup>–</sup>, awaits further explorations.</p></sec>
Liver fibrosis is strongly associated with an enhanced level of immunosuppressive tryptophan catabolism independently of HCV viremia in ART-treated HIV/HCV co-infected patients	Could not extract abstract	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Jenabian</surname><given-names>Mohammad-Ali</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref></contrib><contrib contrib-type="author" id="A2"><name><surname>Kema</surname><given-names>Ido</given-names></name><xref ref-type="aff" rid="I3">3</xref></contrib><contrib contrib-type="author" id="A3"><name><surname>Ramirez</surname><given-names>Robert Paulino</given-names></name><xref ref-type="aff" rid="I4">4</xref></contrib><contrib contrib-type="author" id="A4"><name><surname>Saeed</surname><given-names>Sahar</given-names></name><xref ref-type="aff" rid="I5">5</xref></contrib><contrib contrib-type="author" id="A5"><name><surname>Rollet</surname><given-names>Kathleen</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A6"><name><surname>Vyboh</surname><given-names>Kishanda</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A7"><name><surname>Tejada</surname><given-names>Jean-Carlos</given-names></name><xref ref-type="aff" rid="I6">6</xref></contrib><contrib contrib-type="author" id="A8"><name><surname>Gilmore</surname><given-names>Norbert</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref></contrib><contrib contrib-type="author" id="A9"><name><surname>Klein</surname><given-names>Marina B</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref></contrib><contrib contrib-type="author" id="A10"><name><surname>Routy</surname><given-names>Jean-Pierre</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref></contrib>	BMC Infectious Diseases	<sec><title>Background</title><p>HCV infection induces hepatic and extra-hepatic damage that includes kidney and neurocognitive dysfunction. Tryptophan (Trp) is catabolized into immunosuppressive kynurenine (Kyn) by indoleamine 2,3-dioxygenase (IDO) and tryptophan 2,3 dioxegenase (TDO). Increased Trp catabolism measured by Kyn/Trp ratio has been associated with neurocognitive impairment and immune dysfunction in HIV mono-infection. Here, we assessed the contribution of Trp catabolism in HCV/HIV co-infected patients.</p></sec><sec sec-type="methods"><title>Methods</title><p>Plasma samples were collected from ART-treated (HIV RNA <40 copies/ml) HCV/HIV co-infected patients with or without liver fibrosis (n=20 per group), HBV/HIV co-infected patients (n=25), ART-treated and untreated HIV-mono-infected patients and 30 healthy subjects (HS), (n=30 per group). Furthermore, 17 additional HCV/HIV INF-α/ribavirin treated patients were longitudinally assessed before and 6 months after sustained virological response (SVR). IDO and TDO enzymatic activity (Kyn/Trp ratio) was measured by isotope dilution tandem mass spectrometry. Statistical analyses were performed using Anova, unpaired or paired t-tests and Spearman correlation tests.</p></sec><sec sec-type="results"><title>Results</title><p>Among HCV/HIV patients, those having fibrosis compared with non-fibrosis had higher APRI scores (2.48±0.23 vs 0.36±0.018, p<0.0001) and elevated Kyn levels (2.6±0.24 vs. 1.97±0.15 µmol/L, p=0.038). For HBV/HIV co-infected, Kyn level was also elevated (2.1±0.16 µmol/L). The Kyn/Trp ratio was equally elevated in all HCV and HBV co-infected groups, similar to the untreated mono-infected HIV group. Importantly, HCV/HIV fibrotic and HBV/HIV groups but not the non-fibrotic group had higher Kyn/Trp ratios compared to the ART-treated and HS groups. Unlike HIV viremia, HCV viremia was not correlated with the Kyn/Trp ratio. However, in all HCV/HIV co-infected patients, Kyn/Trp ratio was correlated with the APRI score (p=0.027). Successful HCV treatment improved APRI score (0.89±0.13 vs. 0.4±0.04, p=0.001), contrasting with unchanged elevated Kyn/Trp ratios six months after SVR.</p></sec><sec sec-type="conclusions"><title>Conclusion</title><p>ART-treated HCV/HIV and HBV/HIV co-infected patients presented with elevated immunosuppressive Kyn/Trp ratios when compared to mono-infected HIV-treated patients and reached a ratio similar to the untreated HIV mono-infected patients. In ART-treated patients, liver fibrosis on its own, but not HCV viremia, was associated with an enhanced level of immunosuppressive Tryptophan catabolism. These findings suggest that a necrotico-inflammatory liver syndrome persists even after SVR, and subsequently induces a systemic immune activation by increasing tryptophan catabolism.</p></sec>
Small Bifunctional Chelators That Do Not Disaggregate Amyloid β Fibrils Exhibit Reduced Cellular Toxicity	<p content-type="toc-graphic"><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-00926c_0013" id="ab-tgr1"/></p><p>Multifunctional metal chelators that can modulate the amyloid β (Aβ) peptide aggregation and its interaction with metal ions such as copper and zinc hold considerable promise as therapeutic agents for Alzheimer’s disease (AD). However, specific rather than systemic metal chelation by these compounds is needed in order to limit any side effects. Reported herein are two novel small bifunctional chelators, 2-[2-hydroxy-4-(diethylamino)phenyl]benzothiazole (L1) and 2-(2-hydroxy-3-methoxyphenyl)benzothiazole (L2), in which the metal-binding donor atoms are integrated within a molecular framework derived from the amyloid-binding fluorescent dye thioflavin T (ThT). The metal-binding properties of L1 and L2 were probed by pH spectrophotometric titrations to determine their p<italic>K</italic><sub>a</sub> values and the corresponding metal complex stability constants, and the isolated metal complexes were structurally characterized. The amyloid-fibril-binding properties of L1 and L2 were investigated by fluorescence titrations and ThT competition assays. Interestingly, L1 and L2 do not lead to the formation of neurotoxic Aβ<sub>42</sub> oligomers in the presence or absence of metal ions, as observed by native gel electrophoresis, Western blotting, and transmission electron microscopy. In addition, L1 and L2 were able to reduce the cell toxicity of preformed Aβ<sub>42</sub> oligomers and of the copper-stabilized Aβ<sub>42</sub> oligomers. Given their ability to reduce the toxicity of soluble Aβ<sub>42</sub> and Cu-Aβ<sub>42</sub> species, L1 and L2 are promising lead compounds for the development of chemical agents that can control the neurotoxicity of soluble Aβ<sub>42</sub> species in AD.</p>	<contrib contrib-type="author" id="ath1"><name><surname>Sharma</surname><given-names>Anuj K.</given-names></name><xref rid="aff1" ref-type="aff">†</xref></contrib><contrib contrib-type="author" id="ath2"><name><surname>Kim</surname><given-names>Jaekwang</given-names></name><xref rid="aff2" ref-type="aff">‡</xref></contrib><contrib contrib-type="author" id="ath3"><name><surname>Prior</surname><given-names>John T.</given-names></name><xref rid="aff1" ref-type="aff">†</xref></contrib><contrib contrib-type="author" id="ath4"><name><surname>Hawco</surname><given-names>Nicholas J.</given-names></name><xref rid="aff1" ref-type="aff">†</xref></contrib><contrib contrib-type="author" id="ath5"><name><surname>Rath</surname><given-names>Nigam P.</given-names></name><xref rid="aff3" ref-type="aff">§</xref></contrib><contrib contrib-type="author" id="ath6"><name><surname>Kim</surname><given-names>Jungsu</given-names></name><xref rid="aff2" ref-type="aff">‡</xref></contrib><contrib contrib-type="author" corresp="yes" id="ath7"><name><surname>Mirica</surname><given-names>Liviu M.</given-names></name><xref rid="cor1" ref-type="other">*</xref><xref rid="aff1" ref-type="aff">†</xref></contrib><aff id="aff1"><label>†</label>Department of Chemistry, <institution>Washington University</institution>, One Brookings Drive, St. Louis, Missouri 63130-4899, <country>United States</country></aff><aff id="aff2"><label>‡</label>Department of Neurology, <institution>Washington University School of Medicine</institution>, St. Louis, Missouri 63108, <country>United States</country></aff><aff id="aff3"><label>§</label>Department of Chemistry and Biochemistry, <institution>University of Missouri—St. Louis</institution>, One University Boulevard, St. Louis, Missouri 63121-4400, <country>United States</country></aff>	Inorganic Chemistry	<sec sec-type="intro" id="sec1"><title>Introduction</title><p>More than 5 million people in the United States and 24 million worldwide are affected by Alzheimer’s disease (AD), a disease characterized by memory loss and neurodegeneration.<sup><xref ref-type="bibr" rid="ref1">1</xref>−<xref ref-type="bibr" rid="ref3">3</xref></sup> Insoluble aggregates of the 42 and 40 amino acid long amyloid β peptides (Aβ<sub>42</sub> and Aβ<sub>40</sub>, respectively) are believed to be intimately involved in the onset of AD. However, while amyloid plaques, the ultimate product of Aβ aggregation, have been proposed to promote neurodegeneration and dementia,<sup><xref ref-type="bibr" rid="ref4">4</xref></sup> recent studies have shown that the soluble Aβ oligomers are more neurotoxic both in vitro and in vivo.<sup><xref ref-type="bibr" rid="ref5">5</xref>−<xref ref-type="bibr" rid="ref8">8</xref></sup> In addition, while Aβ<sub>40</sub> is present in larger amounts in amyloid plaques, the longer and more hydrophobic Aβ<sub>42</sub> isoform has a higher tendency to generate neurotoxic oligomeric species.<sup><xref ref-type="bibr" rid="ref9">9</xref>,<xref ref-type="bibr" rid="ref10">10</xref></sup></p><p>Interestingly, unusually high concentrations of copper, iron, and zinc have been found within the amyloid deposits in AD-affected brains.<sup><xref ref-type="bibr" rid="ref11">11</xref></sup> These metal ions have been shown to affect Aβ aggregation, as well as lead to the formation of reactive oxygen species (ROS).<sup><xref ref-type="bibr" rid="ref12">12</xref>−<xref ref-type="bibr" rid="ref16">16</xref></sup> Although many studies have investigated the involvement of metal ions in Aβ aggregation and plaque formation, the molecular mechanisms of metal–Aβ species interactions, especially for the more neurotoxic Aβ<sub>42</sub>, are still not completely understood. Exley et al. have used fluorescence assays and transmission electron microscopy (TEM) to show that both sub- and superstoichiometric concentrations of Cu<sup>2+</sup> prevent aggregation of Aβ<sub>42</sub> into thioflavin T (ThT)-positive β-sheet-rich amyloid fibrils.<sup><xref ref-type="bibr" rid="ref17">17</xref>−<xref ref-type="bibr" rid="ref19">19</xref></sup> We have also recently reported that Cu<sup>2+</sup> ions slow down fibrillization of the Aβ<sub>42</sub> peptide and stabilize the soluble Aβ<sub>42</sub> species, as observed by native gel electrophoresis/Western blotting, TEM, cell viability studies,<sup><xref ref-type="bibr" rid="ref20">20</xref>,<xref ref-type="bibr" rid="ref21">21</xref></sup> and pulsed hydrogen/deuterium exchange mass spectrometry studies.<sup><xref ref-type="bibr" rid="ref22">22</xref></sup> In this regard, metal chelators have been shown to reduce metal-mediated Aβ aggregation, ROS formation, and neurotoxicity in vitro.<sup><xref ref-type="bibr" rid="ref23">23</xref>−<xref ref-type="bibr" rid="ref26">26</xref></sup> Recent efforts to control abnormal Aβ–metal interactions have focused on small molecules, bifunctional chelators (BFCs), that have binding affinity for both metal ions and Aβ aggregates.<sup><xref ref-type="bibr" rid="ref20">20</xref>,<xref ref-type="bibr" rid="ref25">25</xref>,<xref ref-type="bibr" rid="ref27">27</xref>−<xref ref-type="bibr" rid="ref30">30</xref></sup> However, the majority of these BFCs have been studied for their effect on aggregation of the Aβ<sub>40</sub> peptide,<sup><xref ref-type="bibr" rid="ref25">25</xref>,<xref ref-type="bibr" rid="ref31">31</xref></sup> and fewer studies have focused on their interaction with the more neurotoxic Aβ<sub>42</sub> peptide.<sup><xref ref-type="bibr" rid="ref20">20</xref>,<xref ref-type="bibr" rid="ref21">21</xref></sup> Compounds that inhibit metal-mediated Aβ<sub>40</sub> aggregation or promote disaggregation of amyloid fibrils were shown to lead to increased cell viability.<sup><xref ref-type="bibr" rid="ref25">25</xref>,<xref ref-type="bibr" rid="ref31">31</xref>,<xref ref-type="bibr" rid="ref32">32</xref></sup> However, this approach may not be optimal for the Aβ<sub>42</sub> peptide, given the increased toxicity observed for soluble Aβ<sub>42</sub> oligomers.<sup><xref ref-type="bibr" rid="ref20">20</xref>,<xref ref-type="bibr" rid="ref21">21</xref></sup> Our current efforts are aimed at the development of BFCs that control the metal–Aβ<sub>42</sub> neurotoxicity in vivo and take into consideration the formation of neurotoxic soluble Aβ<sub>42</sub> oligomers and their proposed role in AD neuropathogenesis. Reported herein are two new BFCs, 2-[2-hydroxy-4-(diethylamino)phenyl]benzothiazole (L1) and 2-(2-hydroxy-3-methoxyhenyl)benzothiazole (L2), that were designed following a previously reported strategy of incorporating metal-binding atoms into the structural framework of an Aβ-interacting compound.<sup><xref ref-type="bibr" rid="ref23">23</xref>,<xref ref-type="bibr" rid="ref33">33</xref>,<xref ref-type="bibr" rid="ref34">34</xref></sup> L1 and L2 have a modified 2-(2-hydroxyphenyl)benzothiazole (HPB) molecular framework reminiscent of the amyloid-binding fluorescent dye ThT (Scheme <xref rid="sch1" ref-type="scheme">1</xref>), which was shown to be promising in the design of new Aβ-binding compounds.<sup><xref ref-type="bibr" rid="ref33">33</xref></sup> These BFCs fulfill the druglike criteria for further biological in vivo assays. The bifunctional character of the two compounds was confirmed by metal-chelating and Aβ-binding studies. The Cu<sup>2+</sup> complexes of L1 and L2 were isolated and characterized spectroscopically and by X-ray structural determination. The amyloid-binding properties of L1 and L2 were explored by fluorescent titrations and by ThT competition assays. While L1 exhibits an acceptable binding affinity for Aβ fibrils, with the higher affinity binding site corresponding to a <italic>K</italic><sub>i</sub> of 260 nM, L2 binds very strongly to Aβ fibrils with a <italic>K</italic><sub>i</sub> of 15 nM. In addition, L1 and L2 moderately inhibit Aβ<sub>42</sub> aggregation in the absence of metal ions, yet they promote aggregation in the presence of metal ions, as observed by native gel/Western blotting and TEM. In addition, L1 and L2 were able to reduce the cell toxicity of preformed Aβ<sub>42</sub> oligomers as well as that of soluble copper-stabilized Aβ<sub>42</sub> oligomers, which were shown recently to be neurotoxic.<sup><xref ref-type="bibr" rid="ref20">20</xref>,<xref ref-type="bibr" rid="ref21">21</xref></sup> These promising properties of L1 and L2 make them suitable candidates for further in vivo investigation.</p><fig id="sch1" position="float"><label>Scheme 1</label><caption><title>Molecular Structures of ThT, Pittsburgh Compound B (PiB), Clioquinol (CQ), and the BFCs L1 and L2 Described in This Work</title></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-00926c_0011" id="gr1" position="float"/></fig></sec><sec id="sec2"><title>Experimental Section</title><sec id="sec2.1"><title>General Methods</title><p>All reagents were purchased from commercial sources and used as received unless stated otherwise. Solvents were purified prior to use by passing through a column of activated alumina using an MBraun Solvent Purification System. All solutions and buffers were prepared using metal-free Millipore water that was treated with Chelex overnight and filtered through a 0.22 μm nylon filter. <sup>1</sup>H (300.121 MHz) and <sup>13</sup>C (75 MHz) NMR spectra were recorded on a Varian Mercury-300 spectrometer. Chemical shifts are reported in ppm and referenced to residual solvent resonance peaks. UV–vis spectra were recorded on a Varian Cary 50 Bio spectrophotometer and are reported as λ<sub>max</sub>, nm (ε, M<sup>–1</sup> cm<sup>–1</sup>). Electrospray ionization mass spectrometry (ESI-MS) experiments were performed using a Bruker M-axis QTOF mass spectrometer with an ESI source. ESI-MS was provided by the Washington University Mass Spectrometry NIH Resource (Grant P41RR0954), and elemental analyses were carried out at Intertek Chemical and Pharmaceuticals testing and analysis services. TEM analysis was performed at the Nano Research Facility at Washington University.</p></sec><sec id="sec2.2"><title>X-ray Crystallography</title><p>Suitable crystals of appropriate dimensions were mounted in a Bruker Kappa Apex II CCD X-ray diffractometer equipped with an Oxford Cryostream LT device and a fine-focus Mo Kα radiation X-ray source (λ = 0.71073 Å). Preliminary unit cell constants were determined with a set of 36 narrow frame scans. Typical data sets consist of combinations of ϖ and ϕ scan frames with a typical scan width of 0.5° and a counting time of 15–30 s/frame at a crystal-to-detector distance of ∼4.0 cm. The collected frames were integrated using an orientation matrix determined from the narrow frame scans. <italic>Apex II</italic> and <italic>SAINT</italic> software packages<sup><xref ref-type="bibr" rid="ref35">35</xref></sup> were used for data collection and data integration. Final cell constants were determined by the global refinement of reflections from the complete data set. Data were corrected for systematic errors using <italic>SADABS</italic>.<sup><xref ref-type="bibr" rid="ref35">35</xref></sup> Structure solutions and refinement were carried out using the <italic>SHELXTL-PLUS</italic> software package.<sup><xref ref-type="bibr" rid="ref36">36</xref></sup> The structures were refined with full-matrix least-squares refinement by minimizing ∑<italic>w</italic>(<italic>F</italic><sub>o</sub><sup>2</sup> – <italic>F</italic><sub>c</sub><sup>2</sup>)<sup>2</sup>. All non-hydrogen atoms were refined anisotropically to convergence. All hydrogen atoms were added in the calculated position and refined using appropriate riding models (AFIX m<sup>3</sup>). Additional crystallographic details can be found in the <xref rid="notes-1" ref-type="notes">Supporting Information</xref> (SI).</p></sec><sec id="sec2.3"><title>Acidity and Stability Constant Determination</title><p>UV–vis pH titrations were employed for determination of the acidity constants of L1 and L2 and their stability constants with Cu<sup>2+</sup> and Zn<sup>2+</sup>. For the acidity constants, solutions of BFCs (25 μM for L1 and 50 μM for L2, 0.1 M NaCl, pH 3) were titrated with small aliquots of 0.1 M NaOH at room temperature. At least 30 UV–vis spectra were collected in the pH 3–11 range. Because of the limited solubility of L1 and L2 in water, methanol (MeOH) stock solutions (10 mM) were used and titrations were performed in a MeOH–water mixture in which MeOH did not exceed 20% (v/v). Similarly, the stability constants for the L1–copper system were determined by titrating a solution of L1 (25 μM) and Cu(ClO<sub>4</sub>)<sub>2</sub>·6H<sub>2</sub>O (12.5 μM) with small aliquots of 0.1 M NaOH at room temperature. At least 30 UV–vis spectra were collected in the pH 3–11 range. The acidity and stability constants were calculated using the <italic>HypSpec</italic> computer program (Protonic Software GmbH, U.K.).<sup><xref ref-type="bibr" rid="ref37">37</xref></sup> Speciation plots of the compounds and their metal complexes were calculated using the program <italic>HySS2009</italic> (Protonic Software GmbH, U.K.).<sup><xref ref-type="bibr" rid="ref38">38</xref></sup></p></sec><sec id="sec2.4"><title>Aβ Peptide Experiments</title><p>Aβ monomeric films were prepared by dissolving commercial Aβ<sub>42</sub> (or Aβ<sub>40</sub> for Aβ-fibril-binding studies) peptide (Keck Biotechnology Resource Laboratory, Yale University) in HFIP (1 mM) and incubating for 1 h at room temperature.<sup><xref ref-type="bibr" rid="ref39">39</xref></sup> This solution was then aliquoted out, and HFIP was allowed to evaporate overnight. The aliquots were vacuum-centrifuged, and the resulting monomeric films were stored at −80 °C. Aβ fibrils were generated by dissolving monomeric Aβ films in dimethyl sulfoxide (DMSO), diluting into the appropriate buffer, and incubating for 24 h at 37 °C with continuous agitation (the final DMSO concentration was <2%). For metal-containing fibrils, the corresponding metal ions were added before initiation of the fibrillization conditions. For inhibition studies, BFCs (50 μM, DMSO stock solutions) were added to Aβ solutions (25 μM) in the absence or presence of metal salts (CuCl<sub>2</sub> or ZnCl<sub>2</sub>, 25 μM) and incubated for 24 h at 37 °C with constant agitation. For disaggregation studies, the preformed Aβ fibrils in the absence or presence of metal ions were treated with BFCs and further incubated for 24 h at 37 °C with constant agitation. For the preparation of soluble Aβ<sub>42</sub> oligomers, a literature protocol was followed.<sup><xref ref-type="bibr" rid="ref7">7</xref>,<xref ref-type="bibr" rid="ref39">39</xref></sup> A monomeric film of Aβ<sub>42</sub> was dissolved in anhydrous DMSO, followed by the addition of DMEM-F12 media [1:1 (v/v), without phenol red; Invitrogen]. The solution (50–100 μM) was incubated at 4 °C for 24 h and then centrifuged at 10000<italic>g</italic> for 10 min. The supernatant was used as a solution of soluble Aβ<sub>42</sub> oligomers.</p></sec><sec id="sec2.5"><title>Fluorescence Measurements</title><p>All fluorescence measurements were performed using a SpectraMax M2e plate reader (Molecular Devices). For ThT fluorescence studies, samples were diluted to a final concentration of 2.5 μM Aβ in phosphate-buffered saline (PBS) containing 10 μM ThT and the fluorescence was measured at 485 nm (λ<sub>ex</sub> = 435 nm). For Aβ-fibril-binding studies, a 5 μM Aβ fibril solution was titrated with small amounts of compound and their fluorescence intensity measured (λ<sub>ex</sub>/λ<sub>em</sub> = 330/450 nm). For ThT competition assays, a 5 μM Aβ fibril solution with 2 μM ThT was titrated with small amounts of compound and the ThT fluorescence measured (λ<sub>ex</sub>/λ<sub>em</sub> = 435/485 nm). For calculation of <italic>K</italic><sub>i</sub> values, a <italic>K</italic><sub>d</sub> value of 1.17 μM was used for the binding of ThT to Aβ<sub>40</sub> fibrils.<sup><xref ref-type="bibr" rid="ref20">20</xref>,<xref ref-type="bibr" rid="ref21">21</xref></sup></p></sec><sec id="sec2.6"><title>TEM</title><p>Glow-discharged grids (Formar/Carbon 300 mesh, Electron Microscopy Sciences) were treated with Aβ samples (25 μM, 5 μL) for 2–3 min at room temperature. Excess solution was removed using filter paper, and grids were rinsed twice with water (5 μL). Grids were stained with uranyl acetate (1% w/v, water, 5 μL) for 1 min, blotted with filter paper, and dried for 15 min at room temperature. Images were captured using a FEI G2 Spirit Twin microscope (60–80 kV, 6500–97000× magnification).</p></sec><sec id="sec2.7"><title>Native Gel Electrophoresis and Western Blotting</title><p>All gels, buffers, membranes, and other reagents were purchased from Invitrogen and used as directed except where otherwise noted. Samples were separated on 10–20% gradient Tris-tricine minigels. The gel was transferred to a nitrocellulose membrane in an ice bath, and the protocol was followed as suggested except that the membrane was blocked overnight at 4 °C. After blocking, the membrane was incubated in a solution (1:2000 dilutions) of 6E10 anti-Aβ primary antibody (Covance) for 3 h. Invitrogen’s Western Breeze Chemiluminescent kit was used to visualize the bands. An alkaline/phosphatase antimouse secondary antibody was used, and the protein bands were imaged using a Fujifilm LAS-1000CH luminescent image analyzer.</p></sec><sec id="sec2.8"><title>Cytotoxicity Studies (Alamar Blue Assay)</title><p>Mouse neuroblastoma Neuro2A (N2A) cell lines were purchased from the American Type Culture Collection (ATCC). Cells were grown in DMEM/10% FBS, which is the regular growth media for N2A cells. N2A cells were plated to each well of a 96-well plate (2.5 × 10<sup>4</sup>/well) with DMEM/10% FBS. The media was changed to DMEM/N2 media 24 h later. After 1 h, the reagents (20 μM Aβ<sub>42</sub> species, compounds, and metals) were added. Because of the poor solubility of the compounds in water or media, the final amount of DMSO used was 1% (v/v). After an additional incubation of 40 h, the Alamar Blue solution was added in each well and the cells were incubated for 90 min at 37 °C. The absorbance was measured at 570 nm (control optical density = 600 nm). For these toxicity studies, three types of Aβ<sub>42</sub> species were tested: freshly made monomeric Aβ<sub>42</sub> (MAβ<sub>42</sub>), Aβ<sub>42</sub> oligomers (OAβ<sub>42</sub>), and Aβ<sub>42</sub> fibrils (FAβ<sub>42</sub>). These Aβ<sub>42</sub> species were prepared as described above.</p><sec id="sec2.8.1"><title>Synthesis of L1</title><p>A mixture of 2-aminothiophenol (2.5 g, 20 mmol) and (diethylamino)salicylaldehyde (3.86 g, 20 mmol) in DMSO (20 mL) was heated with stirring at 150 °C for 6 h under dinitrogen. Then it was cooled to 0 °C for 8 h, and the orange solid crystallized out, was collected by filtration, and was washed with hexane (2.97 g, yield 50%). <sup>1</sup>H NMR (CDCl<sub>3</sub>): δ 7.85 (d, 1H, ArH), 7.81 (d, 1H, ArH), 7.46 (d, 1H, ArH), 7.43 (d, 1H, ArH), 7.30 (d, 1H, phenol H), 6.30 (d, 1H, phenol H), 6.27 (s, 1H, phenol H), 3.41 (q, 2H, C<italic>H</italic><sub>2</sub>CH<sub>3</sub>), 1.22 (t, 3H, CH<sub>2</sub>C<italic>H</italic><sub>3</sub>). <sup>13</sup>C NMR (CDCl<sub>3</sub>): δ 169.78, 159.98, 152.43, 151.52, 132.08, 130.04, 126.42, 124.35, 121.40, 121.18, 105.96, 104.34, 98.053, 44.72, 12.86. UV–vis [λ<sub>max</sub>, nm, (ε, M<sup>–1</sup> cm<sup>–1</sup>)]: in MeCN, 262 (10300), 380 (43000); in PBS, 272 (9500), 375 (18000), 405 (19500). HRMS. Calcd for [M + H]<sup>+</sup>: <italic>m</italic>/<italic>z</italic> 299.1218. Found: <italic>m</italic>/<italic>z</italic> 299.1225.</p></sec><sec id="sec2.8.2"><title>Synthesis of L2</title><p>A mixture of <italic>o</italic>-vanilin (2.0 g, 0.0131 mol) and 2-aminothiophenol (1.64 g, 0.0131 mol) in ethanol (15 mL) was refluxed with stirring for 24 h. Then it was cooled to room temperature and poured in water. The sticky yellow solid was crystallized from MeOH to obtain yellow crystals (1.15 g, yield 37%). <sup>1</sup>H NMR (CDCl<sub>3</sub>): δ 8.02 (d, 1H, ArH), 7.91 (d, 1H, ArH), 7.52 (t, 1H, ArH), 7.42 (t, 1H, ArH), 7.33 (d, 1H, ArH), 7.01 (d, 1H, ArH), 6.99 (t, 1H, ArH), 3.97 (s, 3H, OCH<sub>3</sub>). <sup>13</sup>C NMR (CDCl<sub>3</sub>): δ 169.28, 151.50, 148.93, 148.19, 148.17, 132.59, 126.70, 125.56, 122.16, 121.45, 119.98, 119.14, 116.71, 114.06, 56.22. UV–vis [λ<sub>max</sub>, nm, (ε, M<sup>–1</sup> cm<sup>–1</sup>)]: in MeCN, 307 (br, 10200), 340 (sh, 5500); in PBS, 305 (3500), 340 (sh, 1800), 390 (800). HRMS. Calcd for [M + H]<sup>+</sup>: <italic>m</italic>/<italic>z</italic> 258.0588. Found: <italic>m</italic>/<italic>z</italic> 258.0584.</p></sec><sec id="sec2.8.3"><title>Synthesis of [(L1)<sub>2</sub>Cu<sup>II</sup>]·0.5H<sub>2</sub>O (<bold>1</bold>)</title><p>A solution of CuCl<sub>2</sub>·2H<sub>2</sub>O (0.011 g, 0.084 mmol) in MeCN (2 mL) was added to the stirring mixture of L1 (0.050 g, 0.167 mmol) and Et<sub>3</sub>N (0.036 g, 0.334 mmol) in MeCN (5 mL). The brown precipitate immediately obtained was filtered, washed with ether, and dried under vacuum (0.040 g, yield 54%). UV–vis [λ<sub>max</sub>, nm, (ε, M<sup>–1</sup>cm<sup>–1</sup>)]: in MeCN, 273 (sh, 18000), 391 (56800), 404 (sh, 50500), 540 (900). HRMS. Calcd for [M + H]<sup>+</sup>: <italic>m</italic>/<italic>z</italic> 658.1497. Found: <italic>m</italic>/<italic>z</italic> 658.1492. Anal. Calcd for C<sub>34</sub>H<sub>34</sub>N<sub>4</sub>O<sub>2</sub>S<sub>2</sub>Cu·0.5H<sub>2</sub>O: C, 61.52; H, 5.25; N, 8.44. Found: C, 61.63; H, 6.11; N, 8.44.</p></sec><sec id="sec2.8.4"><title>Synthesis of [(L1)Zn<sup>II</sup>(Cl<sub>2</sub>)](Et<sub>3</sub>NH) (<bold>2</bold>)</title><p>A solution of ZnCl<sub>2</sub> (0.011 g, 0.084 mmol) in MeCN (2 mL) was added to the stirring mixture of L4 (0.050 g, 0.167 mmol) and Et<sub>3</sub>N (0.018 g, 0.167 mmol) in MeCN (5 mL). The red solution was kept for slow evaporation, and a few single crystals were obtained within 2 days (0.052 g, yield 58%). UV–vis [λ<sub>max</sub>, nm, (ε, M<sup>–1</sup>cm<sup>–1</sup>)]: in MeCN, 265 (19900), 391 (51000). ESI-MS. Calcd for [(L1)ZnCl + H]<sup>+</sup>: <italic>m</italic>/<italic>z</italic> 397.01. Found: <italic>m</italic>/<italic>z</italic> 397.01. Satisfactory elemental analysis data could not be obtained because of the presence of a small amount of free ligand.</p></sec><sec id="sec2.8.5"><title>Synthesis of [(L2)<sub>2</sub>Cu<sup>II</sup>]·0.5H<sub>2</sub>O (<bold>3</bold>)</title><p>A solution of CuCl<sub>2</sub>·2H<sub>2</sub>O (0.013 g, 0.097 mmol) in MeCN (2 mL) was added to the stirring mixture of L2 (0.050 g, 0.194 mmol) and Et<sub>3</sub>N (0.078 g, 0.776 mmol) in MeOH (10 mL). The brown precipitate immediately obtained was filtered, washed with ether, and dried under vacuum (0.051 g, yield 91%). UV–vis [λ<sub>max</sub>, nm, (ε, M<sup>–1</sup>cm<sup>–1</sup>)]: in MeCN, 398 (26200), 515 (sh, 800), 700 (350). ESI-MS. Calcd for [M + Na]<sup>+</sup>: <italic>m</italic>/<italic>z</italic> 598.0. Found: <italic>m</italic>/<italic>z</italic> 598.1. Anal. Calcd for C<sub>28</sub>H<sub>20</sub>N<sub>2</sub>O<sub>4</sub>S<sub>2</sub>Cu·0.5H<sub>2</sub>O: C, 57.47; H, 3.62; N, 4.79. Found: C, 57.34; H, 3.62; N, 4.83.</p></sec></sec></sec><sec id="sec3"><title>Results and Discussion</title><sec id="sec3.1"><title>Design, Synthesis, and Characterization of L1 and L2</title><p>Herein we employed the strategy of developing metal BFCs with amyloid-binding properties as potential therapeutic agents for AD. The fluorescent dye ThT is well-known for its amyloid-binding properties because of the 2-phenylbenzothiazole aromatic framework. Previously, an iodinated HPB and a few related compounds were shown to have potential as metal-chelating agents by Rodríguez-Rodríguez et al.<sup><xref ref-type="bibr" rid="ref33">33</xref></sup> However, in that report, only the solution coordination properties of those compounds were investigated in detail, while the Aβ-binding and antiaggregation properties were only qualitatively evaluated through epifluorecense microscopy and turbidity assays. In contrast, in addition to the solution coordination chemistry, we have also determined the solid-state structure of the copper and zinc complexes of our compounds, and we have also quantitatively determined their Aβ-binding affinities through fluorescence titrations. In addition, we have used both TEM and gel electrophoresis/Western blotting to provide a complete picture of the effect of our compounds on Aβ aggregation and formation of various sized Aβ aggregates, in both the absence and presence of metal ions. Most importantly, we have correlated the in vitro studies with cellular toxicity studies, thus making this work a comprehensive study of these compounds in the context of developing new compounds that can control the toxicity and aggregation of various Aβ species. We have also recently reported 2-phenylbenzothiazole/vanillin derivatives as BFCs that showed a strong affinity for Aβ fibrils.<sup><xref ref-type="bibr" rid="ref20">20</xref></sup> Herein we designed two new molecules, L1 and L2, that contain a 2-phenylbenzothiazole moiety for Aβ recognition and oxygen- and nitrogen-atom donors needed for metal chelation. Synthesis of L1 and L2 was accomplished upon oxidative cyclization of the corresponding arylaldehyde and 2-aminothiophenol (Scheme S1 in the <xref rid="notes-1" ref-type="notes">SI</xref>).<sup><xref ref-type="bibr" rid="ref40">40</xref></sup> These compounds exhibit UV absorption in MeCN at 260 and 380 nm for L1 and at 305 and 345 nm for L2 (Figures S1 and S2 in the <xref rid="notes-1" ref-type="notes">SI</xref>).<sup><xref ref-type="bibr" rid="ref40">40</xref></sup> While both L1 and L2 are 2-phenylbenzothiazole derivatives, L1 exhibits low-intensity emission at 450 nm and a shoulder at 525 nm, yet L2 shows a more intense emission at 525 nm in PBS (Figures S3 and S4 in the <xref rid="notes-1" ref-type="notes">SI</xref>).<sup><xref ref-type="bibr" rid="ref40">40</xref></sup> The parent molecule HPB has emission properties similar to those of L2 (Figure S4 in the <xref rid="notes-1" ref-type="notes">SI</xref>), which suggests that having a diethylamino group in L1 strongly affects its emission properties.</p><p>Another important aspect of the design of new drug molecules for central nervous system diseases is their ability to cross the blood brain barrier (BBB).<sup><xref ref-type="bibr" rid="ref13">13</xref>,<xref ref-type="bibr" rid="ref26">26</xref>,<xref ref-type="bibr" rid="ref41">41</xref></sup> Both L1 and L2 are small in size, and they also satisfy Lipinski’s rules for crossing the BBB (Table <xref rid="tbl1" ref-type="other">1</xref>). The ability of a compound to cross the BBB can be evaluated by determining the blood–brain partitioning parameter log BB that is determined by taking into account the molecular weight of the molecule, charge, lipophilicity, and its hydrogen-bonding capability. Molecules having log BB values higher than −0.3 are generally expected to cross the BBB.<sup><xref ref-type="bibr" rid="ref33">33</xref></sup> The calculated log BB values for L1 and L2 are +0.33 and +0.087, respectively, suggesting that these BFCs are expected to cross the BBB.</p><table-wrap id="tbl1" position="float"><label>Table 1</label><caption><title>Lipinski Parameters for L1 and L2</title></caption><table frame="hsides" rules="groups" border="0"><colgroup><col/><col/><col/><col/></colgroup><thead><tr><th style="border:none;" align="center">property</th><th style="border:none;" align="center">L1</th><th style="border:none;" align="center">L2</th><th style="border:none;" align="center">Lipinski’s rule<xref rid="t1fn1" ref-type="table-fn">a</xref></th></tr></thead><tbody><tr><td style="border:none;">MW</td><td style="border:none;">298.41</td><td style="border:none;">257.31</td><td style="border:none;">≤500</td></tr><tr><td style="border:none;"><italic>c</italic> log <italic>P</italic></td><td style="border:none;">4.853</td><td style="border:none;">3.834</td><td style="border:none;">≤5</td></tr><tr><td style="border:none;">HBD</td><td style="border:none;">1</td><td style="border:none;">1</td><td style="border:none;">≤5</td></tr><tr><td style="border:none;">HBA</td><td style="border:none;">3</td><td style="border:none;">3</td><td style="border:none;">≤10</td></tr><tr><td style="border:none;">PSA</td><td style="border:none;">36.36</td><td style="border:none;">42.35</td><td style="border:none;">≤70 Å<sup>2</sup></td></tr><tr><td style="border:none;">log BB</td><td style="border:none;">0.330</td><td style="border:none;">0.087</td><td style="border:none;">>−0.3</td></tr></tbody></table><table-wrap-foot><fn id="t1fn1"><label>a</label><p>MW = molecular weight; <italic>c</italic> log <italic>P</italic> = calculated octanol–water partition coefficient; HBD = hydrogen-bonding donor atoms; HBA = hydrogen-bonding acceptor atoms; PSA = polar surface area; log BB = −0.0148PSA + 0.152 <italic>c</italic> log <italic>P</italic> + 0.130 (Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P. J. <italic>Adv. Drug Delivery Rev.</italic><bold>1997</bold>, 23, 3–25; Clark, D. E.; Pickett, S. D. <italic>Drug Discovery Today</italic><bold>2000</bold>, 5, 49–58).</p></fn></table-wrap-foot></table-wrap></sec><sec id="sec3.2"><title>Acidity Constants of L1 and L2</title><p>Because both L1 and L2 contain functional groups that can undergo protonation, the acidity constants (p<italic>K</italic><sub>a</sub>) were determined by UV–vis spectrophotometric titrations. For both L1 and L2, UV–vis titrations from pH 3.0 to 11.0 reveal several changes in the spectra (Figures <xref rid="fig1" ref-type="fig">1</xref> and <xref rid="fig2" ref-type="fig">2</xref>). For L1, the best fit to the data was obtained with p<italic>K</italic><sub>a</sub> values of 3.556(2), 8.427(3), and 10.108(2) (Table <xref rid="tbl2" ref-type="other">2</xref>). The first and second p<italic>K</italic><sub>a</sub> values can be assigned to deprotonation of the nitrogen atoms of the benzothiazole and diethylamino groups, while the largest p<italic>K</italic><sub>a</sub> value is likely due to phenol deprotonation.<sup><xref ref-type="bibr" rid="ref25">25</xref>,<xref ref-type="bibr" rid="ref42">42</xref></sup> For L2, the best fit to the data was obtained with two p<italic>K</italic><sub>a</sub> values of 3.154(2) and 9.360(1) (Table <xref rid="tbl2" ref-type="other">2</xref>). These p<italic>K</italic><sub>a</sub> values can be assigned to deprotonation of the nitrogen atoms of the benzothiazole and phenol groups, respectively.</p><fig id="fig1" position="float"><label>Figure 1</label><caption><p>Variable-pH UV spectra of L1 (25 μM, 25 °C, <italic>I</italic> = 0.1 M NaCl) and the species distribution plot.</p></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-00926c_0002" id="gr3" position="float"/></fig><fig id="fig2" position="float"><label>Figure 2</label><caption><p>Variable-pH UV spectra of L2 (50 μM, 25 °C, <italic>I</italic> = 0.1 M NaCl) and the species distribution plot.</p></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-00926c_0003" id="gr4" position="float"/></fig><table-wrap id="tbl2" position="float"><label>Table 2</label><caption><title>Acidity Constants (p<italic>K</italic><sub>a</sub> Values) of L1 and L2 Determined by Spectrophotometric Titrations (Errors Are for the Last Digit)<xref rid="tbl2-fn1" ref-type="table-fn">a</xref></title></caption><table frame="hsides" rules="groups" border="0"><colgroup><col align="left"/><col align="char" char="."/><col align="char" char="."/></colgroup><thead><tr><th style="border:none;" align="center">reaction</th><th style="border:none;" align="center" char=".">L1</th><th style="border:none;" align="center" char=".">L2</th></tr></thead><tbody><tr><td style="border:none;" align="left">[H<sub>3</sub>L]<sup>2+</sup> = [H<sub>2</sub>L]<sup>+</sup> + H<sup>+</sup> (p<italic>K</italic><sub>a1</sub>)</td><td style="border:none;" align="char" char=".">3.556(2)</td><td style="border:none;" align="char" char="."> </td></tr><tr><td style="border:none;" align="left">[H<sub>2</sub>L]<sup>+</sup> = [HL] + H<sup>+</sup> (p<italic>K</italic><sub>a2</sub>)</td><td style="border:none;" align="char" char=".">8.427(3)</td><td style="border:none;" align="char" char=".">3.154(2)</td></tr><tr><td style="border:none;" align="left">[HL] = [L]<sup>−</sup> + H<sup>+</sup> (p<italic>K</italic><sub>a3</sub>)</td><td style="border:none;" align="char" char=".">10.108(2)</td><td style="border:none;" align="char" char=".">9.360(1)</td></tr></tbody></table><table-wrap-foot><fn id="tbl2-fn1"><label>a</label><p>[HL] represents the neutral form of L1 and L2.</p></fn></table-wrap-foot></table-wrap></sec><sec id="sec3.3"><title>Characterization of Metal Complexes</title><p>The BFCs L1 and L2 were explored for their ability to chelate metal ions. The stoichiometry of L1–Cu<sup>2+</sup> and L1–Zn<sup>2+</sup> complexes in solution was determined by Job’s plot analysis.<sup><xref ref-type="bibr" rid="ref43">43</xref></sup> For L1–Cu<sup>2+</sup>, a break in the plot at 0.4 mole fraction of Cu<sup>2+</sup> suggests the formation of both 1:1 and 2:1 L1–Cu<sup>2+</sup> complexes in solution (Figure S4 in the <xref rid="notes-1" ref-type="notes">SI</xref>).<sup><xref ref-type="bibr" rid="ref40">40</xref></sup> For L1–Zn<sup>2+</sup>, the break occurs at ∼0.3 mole fraction of Zn<sup>2+</sup>, suggesting the formation of a 2:1 L1–Zn<sup>2+</sup> complex (Figure S5 in the <xref rid="notes-1" ref-type="notes">SI</xref>).<sup><xref ref-type="bibr" rid="ref40">40</xref></sup> By contrast to L1, no Job’s plot analysis could be performed for L2–Cu<sup>2+</sup> because of the limited solubility of this complex, while no significant spectral change was observed for the L2–Zn<sup>2+</sup> system. The Cu<sup>2+</sup> complexes of L1 or L2 were synthesized by reacting CuCl<sub>2</sub>·2H<sub>2</sub>O with the corresponding ligand in MeCN in the presence of an equivalent amount of Et<sub>3</sub>N. The 2:1 complex formation was confirmed by the ESI-MS signal at <italic>m</italic>/<italic>z</italic> 658.1 corresponding to [M + H]<sup>+</sup>, which was further supported by X-ray structural determination (see below). The UV–vis spectrum of the L1–Cu<sup>2+</sup> complex <bold>1</bold> reveals a d–d transition at ∼700 nm and a more intense band at 540 nm, while a phenolate-to-Cu<sup>2+</sup> charge-transfer band is observed at ∼400 nm (Figure S6 in the <xref rid="notes-1" ref-type="notes">SI</xref>).<sup><xref ref-type="bibr" rid="ref40">40</xref></sup> Similarly, the UV–vis spectrum of the L2–Cu<sup>2+</sup> complex <bold>3</bold> shows characteristic d–d transition bands at 700 and 515 nm as well as a phenolate-to-Cu charge-transfer band at ∼400 nm (Figure S7 in the <xref rid="notes-1" ref-type="notes">SI</xref>).<sup><xref ref-type="bibr" rid="ref40">40</xref></sup> Although a single crystal suitable for X-ray crystallography could be obtained from the reaction mixture of L1 and Zn<sup>2+</sup>, the isolation of the zinc complexes of L1 or L2 did not lead to pure complexes to allow for their complete characterization.</p><p>Spectrophotometric titrations were also performed to determine the stability constants and solution speciation of Cu<sup>2+</sup> with L1. The p<italic>K</italic><sub>a</sub> values of the ligands were included in the calculations,<sup><xref ref-type="bibr" rid="ref42">42</xref></sup> and the best fit to the data shows that L1 binds Cu<sup>2+</sup> with a stability constant of 11.99(2) for the M + L = ML equilibrium and a value of 20.99(2) for the ML + L = ML<sub>2</sub> equilibrium (L represents the deprotonated form of L1). The speciation diagram for the Cu–L1 system is shown in Figure <xref rid="fig3" ref-type="fig">3</xref>. In addition, the concentration of free Cu<sup>2+</sup> (pCu = −log [Cu<sub>free</sub>]) can be determined at specific pH values and total ion concentrations, with pCu values of 6.7 and 8.2 being determined for the Cu–L1 system at pH 6.6 and 7.4, respectively. These pCu values are lower compared to those for the analogous HPB compound reported previously,<sup><xref ref-type="bibr" rid="ref33">33</xref></sup> likely because of the strong electron-withdrawing effect of the diethylamine group upon protonation. Unfortunately, the spectrophotometric titration of L2 in the presence of Cu<sup>2+</sup> revealed few spectral changes at micromolar concentrations, while the use of higher concentrations led to precipitation of the L2–Cu<sup>2+</sup> complex. However, it is expected that L2 exhibits a higher binding affinity to Cu<sup>2+</sup> than L1 because of the electron-donating <italic>o</italic>-methoxy group present in L2. In addition, almost no changes were observed in the spectra of both L1 and L2 upon the addition of Zn<sup>2+</sup>; therefore, the corresponding stability constants could not be determined. Attempts were then made to determine the apparent stability constants for the Zn–L1 and Zn–L2 complexes by using competition assays with a zinc chelator, zincon. Zincon was used previously by Faller et al. in Zn–Aβ interaction studies and exhibits a <italic>K</italic><sub>d</sub> ≈ 10 μM for zinc.<sup><xref ref-type="bibr" rid="ref44">44</xref></sup> In our case, the addition of zincon to solutions of Zn–L1 and Zn–L2 complexes results in quantitative formation of the Zn–zincon complex. In addition, L1 and L2 do not compete significantly with zincon for zinc binding, even when present in large excess versus zincon, suggesting that they exhibit affinities for zinc that are at least 10 times lower than that of zincon.</p><fig id="fig3" position="float"><label>Figure 3</label><caption><p>Variable-pH UV spectra of the L1–Cu<sup>2+</sup> system ([L1] = 25 μM; [Cu<sup>2+</sup>] = 12.5 μM, 25 °C, <italic>I</italic> = 0.1 M NaCl) and the species distribution plot.</p></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-00926c_0004" id="gr5" position="float"/></fig></sec><sec id="sec3.4"><title>X-ray Structure of Metal Complexes</title><p>To further characterize the metal-binding properties of compounds L1 and L2, single-crystal X-ray structures were determined for complexes <bold>1</bold>–<bold>3</bold> (Figure <xref rid="fig4" ref-type="fig">4</xref> and Tables S1–S3 in the <xref rid="notes-1" ref-type="notes">SI</xref>).<sup><xref ref-type="bibr" rid="ref40">40</xref></sup> Complex <bold>1</bold> affords single crystals by Et<sub>2</sub>O diffusion into a CH<sub>2</sub>Cl<sub>2</sub> solution of <bold>1</bold> that crystallized out in the <italic>C</italic>2/<italic>c</italic> space group. An oxygen atom and a nitrogen atom from two ligand molecules coordinate to Cu<sup>2+</sup>. The Cu–O1 and Cu–N1 bond lengths are 1.980(3) and 1.876(2) Å, respectively, and the trans angles O1–Cu1–O1#1 and N1#1–Cu1–N1 are 153.3(2)° and 154.2(2)°, respectively, with these parameters suggesting a distorted square-planar geometry for Cu<sup>2+</sup>. For complex <bold>2</bold>, the Zn<sup>2+</sup> ion exhibits a distorted tetrahedral geometry and is coordinated to the benzothiazole nitrogen and phenol oxygen atoms of L1 and two chloride ions. The copper complex <bold>3</bold> afforded single crystals by Et<sub>2</sub>O diffusion into a CH<sub>2</sub>Cl<sub>2</sub> solution and has structural properties similar to those of <bold>1</bold>, yet it crystallized out in the <italic>P</italic>2/<italic>n</italic> space group. As shown for <bold>1</bold>, in <bold>3</bold> an oxygen atom and a nitrogen atom from two ligand molecules coordinate to Cu<sup>2+</sup>. The Cu–O1 and Cu–N1 bond lengths are 1.873(4) and 1.965(4) Å, respectively, and the trans angles O1–Cu1–O1#1 and N1#1–Cu1–N1 are 145.1(2)° and 146.2(2)°, respectively, reflecting a distorted square-planar geometry for Cu<sup>2+</sup>.</p><fig id="fig4" position="float"><label>Figure 4</label><caption><p>ORTEP representations (50% probability ellipsoids) of complexes <bold>1</bold>–<bold>3</bold>. All hydrogen atoms, counterions, and solvent molecules are omitted for clarity. Selected bond distances: <bold>1</bold>, Cu1–N1 1.980(3), Cu1–O1 1.876(2); <bold>2</bold>, Zn–N1 2.005(1), Zn–O1 1.956(1), Zn–Cl(1) 2.233(1), Zn–Cl(2) 2.222(1); <bold>3</bold>, Cu1–N1/N2 1.965(4), Cu1–O1/O3 1.873(4).</p></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-00926c_0005" id="gr6" position="float"/></fig></sec><sec id="sec3.5"><title>Interaction of L1 and L2 with Aβ Species</title><p>Both compounds L1 and L2 contain a 2-phenylbenzothiazole fragment reminiscent of the fluorescent dye ThT used to detect the β-sheet structure of fibrillar Aβ aggregates.<sup><xref ref-type="bibr" rid="ref45">45</xref>,<xref ref-type="bibr" rid="ref46">46</xref></sup> In this context, the affinity of L1 and L2 toward Aβ fibrils was investigated by fluorescence assays. These studies were performed with Aβ<sub>40</sub> fibrils, which form more homogeneous fibrillar structures without any nonfibrillar aggregates.<sup><xref ref-type="bibr" rid="ref45">45</xref>,<xref ref-type="bibr" rid="ref47">47</xref></sup> Titration of L1 with Aβ fibrils was performed, and a saturation behavior is observed. The data were fitted with a one-site binding model to give a <italic>K</italic><sub>d</sub> of 4.70 ± 0.30 μM (Figure <xref rid="fig5" ref-type="fig">5</xref>a), which suggests a moderate affinity of L1 for the Aβ fibrils. ThT fluorescence competition assays were also performed by the addition of L1 to a solution of Aβ fibrils in the presence of ThT. L1 shows a decrease in ThT fluorescence upon the addition of nanomolar amounts. Competitive-binding curve fitting yields a <italic>K</italic><sub>i</sub> value of 260 ± 40 nM (Figure <xref rid="fig5" ref-type="fig">5</xref>b). It should be noted that the <italic>K</italic><sub>i</sub> value for L1 is smaller than the <italic>K</italic><sub>d</sub> value, which is most likely due to the fact that ThT binds with different affinities to more than one site to the Aβ fibrils.<sup><xref ref-type="bibr" rid="ref47">47</xref></sup> As such, we propose that L1 replaces ThT only from the higher affinity binding site. By constrast, L2 did not show any significant increase in the fluorescence intensity in the presence of Aβ fibrils yet replaces ThT from the Aβ fibrils and exhibits a <italic>K</italic><sub>i</sub> value of 15 ± 10 nM, indicating a very strong binding affinity to Aβ fibrils (Figure <xref rid="fig5" ref-type="fig">5</xref>c).</p><fig id="fig5" position="float"><label>Figure 5</label><caption><p>(a) Fluorescence titration assay of L1 with Aβ fibrils ([Aβ] = 5 μM; λ<sub>ex</sub>/λ<sub>em</sub> = 330/450 nm). ThT fluorescence competition assays of Aβ fibrils with (b) L1 and (c) L2 ([Aβ] = 2 μM; [ThT] = 1 μM; λ<sub>ex</sub>/λ<sub>em</sub> = 435/485 nm).</p></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-00926c_0006" id="gr7" position="float"/></fig></sec><sec id="sec3.6"><title>Effect of L1 and L2 on Aβ<sub>42</sub> Aggregation</title><p>To investigate the effect of L1 and L2 on Aβ aggregation, we performed both inhibition and disaggregation studies (Scheme <xref rid="sch3" ref-type="scheme">3</xref>). Importantly, the Aβ<sub>42</sub> peptide was employed in these studies because it was shown to form neurotoxic soluble Aβ oligomers.<sup><xref ref-type="bibr" rid="ref5">5</xref>,<xref ref-type="bibr" rid="ref8">8</xref>,<xref ref-type="bibr" rid="ref48">48</xref></sup> For inhibition experiments, freshly prepared Aβ<sub>42</sub> fibrils were treated with metal ions, BFCs, or both, and the reactions was monitored by ThT fluorescence, native gel electrophoresis/Western blotting analysis, and TEM.</p><fig id="sch3" position="float"><label>Scheme 3</label><caption><title>Schematic Description of the Performed Inhibition and Disaggregation Experiments</title></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-00926c_0012" id="gr2" position="float"/></fig><p>In inhibition experiments, a reduced ThT fluorescence was observed for the Cu<sup>2+</sup>- and Zn<sup>2+</sup>-containing fibrils, as reported previously (Figure <xref rid="fig6" ref-type="fig">6</xref>).<sup><xref ref-type="bibr" rid="ref21">21</xref>,<xref ref-type="bibr" rid="ref49">49</xref>,<xref ref-type="bibr" rid="ref50">50</xref></sup> In the absence of metal ions, the presence of L1 leads to a low ThT fluorescence intensity, while L2 has a negligible effect on Aβ<sub>42</sub> fibrillization, suggesting that L1 can inhibit fibrillization more efficiently than L2. In the presence of Cu<sup>2+</sup> and either L1 or L2, a very low ThT fluorescence intensity is observed, which could be due to either limited Aβ<sub>42</sub> fibril formation or fluorescence quenching by the paramagnetic Cu<sup>2+</sup> ions. By comparison, the presence of Zn<sup>2+</sup> and either L1 or L2 shows almost no effect on the ThT fluorescence intensity (Figure <xref rid="fig6" ref-type="fig">6</xref>). Because the ThT fluorescence assays do not show a complete picture of the effect of BFCs on Aβ<sub>42</sub> aggregation, native gel electrophoresis/Western blotting analysis and TEM were employed in order to characterize in more detail the various Aβ<sub>42</sub> species formed during these aggregation studies. As such, the Aβ<sub>42</sub> peptide forms well-defined Aβ<sub>42</sub> fibrils, as confirmed by TEM (Figure <xref rid="fig7" ref-type="fig">7</xref>, panel 1), while Aβ<sub>42</sub> aggregation in the presence of Cu<sup>2+</sup> shows the formation of a dramatically reduced number of Aβ<sub>42</sub> fibrils (Figure <xref rid="fig7" ref-type="fig">7</xref>, panel 2), thus supporting the low ThT fluorescence intensity observed above. By comparison, aggregation of Aβ<sub>42</sub> in the presence of Zn<sup>2+</sup> leads to the formation of large amorphous aggregates (Figure <xref rid="fig7" ref-type="fig">7</xref>, panel 3). Native gel/Western blotting analysis shows that aggregation of Aβ<sub>42</sub> in the absence of metal ions also forms high-molecular-weight oligomers, along with the insoluble aggregates that are observed at the top of the blot because they do not enter the gel (Figure <xref rid="fig7" ref-type="fig">7</xref>, lane 1). As reported previously,<sup><xref ref-type="bibr" rid="ref20">20</xref>,<xref ref-type="bibr" rid="ref21">21</xref></sup> the presence of Cu<sup>2+</sup> ions inhibits the formation of large soluble and insoluble Aβ<sub>42</sub> aggregates (Figure <xref rid="fig7" ref-type="fig">7</xref>, lane 2).</p><fig id="fig6" position="float"><label>Figure 6</label><caption><p>Normalized ThT fluorescence of the inhibition of Aβ fibrillization, measured upon incubaton at 37 °C for 24 h. Samples are as indicated on top of the lanes (PBS; [Aβ] = 25 μM; [M<sup>2+</sup>] = 25 μM; [compound] = 50 μM).</p></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-00926c_0007" id="gr8" position="float"/></fig><fig id="fig7" position="float"><label>Figure 7</label><caption><p>Top: TEM images of the inhibition of Aβ<sub>42</sub> aggregation by L1 and L2, in the presence or absence of metal ions ([Aβ<sub>42</sub>] = [M<sup>2+</sup>] = 25 μM; [compound] = 50 μM; PBS, 37 °C, 24 h, scale bar = 500 nm). Bottom: Native gel electrophoresis/Western blotting analysis. Panels and lanes are as follows: (1) Aβ<sub>42</sub>; (2) Aβ<sub>42</sub> + Cu<sup>2+</sup>; (3) Aβ<sub>42</sub> + Zn<sup>2+</sup>; (4) Aβ<sub>42</sub> + L1; (5) Aβ<sub>42</sub> + L1 + Cu<sup>2+</sup>; (6) Aβ<sub>42</sub> + L1 + Zn<sup>2+</sup>; (7) Aβ<sub>42</sub> + L2; (8) Aβ<sub>42</sub> + L2 + Cu<sup>2+</sup>; (9) Aβ<sub>42</sub> + L2 + Zn<sup>2+</sup>; (10) MW marker.</p></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-00926c_0008" id="gr9" position="float"/></fig><p>The presence of either L1 or L2 in the absence of metal ions does not seem to significantly inhibit Aβ<sub>42</sub> aggregation, although some morphological changes were observed for the Aβ<sub>42</sub> fibrils (Figure <xref rid="fig7" ref-type="fig">7</xref>, panels 4 and 7); the observed decrease in the ThT fluorescence intensity may thus be due to these morphological changes of the Aβ<sub>42</sub> aggregates. Native gel/Western blotting analysis reveals a limited effect of L1 and L2 on Aβ<sub>42</sub> aggregation (Figure <xref rid="fig7" ref-type="fig">7</xref>, lanes 4 and 7). Interestingly, the presence of either L1 or L2 has a dramatic effect on the Cu<sup>2+</sup>-mediated oligomerization of Aβ<sub>42</sub> and promotes the formation of larger Aβ<sub>42</sub> aggregates, as observed by TEM (Figure <xref rid="fig7" ref-type="fig">7</xref>, panels 5 and 8). Importantly, native gel/Western blotting analysis reveals bands at the top of the gel, suggesting the presence of insoluble Aβ<sub>42</sub> aggregates (Figure <xref rid="fig7" ref-type="fig">7</xref>, lanes 5 and 8), which were not observed in the presence of Cu<sup>2+</sup> alone (Figure <xref rid="fig7" ref-type="fig">7</xref>, lane 2). Thus, compounds L1 and L2 reduce the amount of soluble Aβ<sub>42</sub> oligomers formed in the presence of Cu<sup>2+</sup> by promoting the formation of insoluble Aβ<sub>42</sub> aggregates. In addition, the presence of either L1 or L2 does not dramatically affect the formation of insoluble Aβ<sub>42</sub> aggregates in the presence of Zn<sup>2+</sup> (Figure <xref rid="fig7" ref-type="fig">7</xref>, panels 6 and 9), which is supported by native gel/Western blotting analysis showing no formation of high-molecular-weight soluble Aβ<sub>42</sub> oligomers, although insoluble aggregates are observed at the top of the gel (Figure <xref rid="fig7" ref-type="fig">7</xref>, lanes 6 and 9). Overall, these studies suggest that the presence of either L1 or L2, in both the presence or absence of metal ions, does not promote the formation of soluble Aβ<sub>42</sub> oligomers. Because the soluble Aβ<sub>42</sub> oligomers are the most neurotoxic species, these compounds are expected to control the neurotoxicity of Aβ<sub>42</sub> species in the presence of metal ions (vide infra).</p></sec><sec id="sec3.7"><title>Effect of L1 and L2 on Aβ<sub>40</sub> Aggregation</title><p>Although it was shown in previous studies<sup><xref ref-type="bibr" rid="ref9">9</xref>,<xref ref-type="bibr" rid="ref10">10</xref>,<xref ref-type="bibr" rid="ref21">21</xref></sup> that the Aβ<sub>42</sub> peptide is more prone to form neurotoxic soluble Aβ oligomers than the Aβ<sub>40</sub> peptide, we have also investigated the effect of L1 and L2 on the metal-mediated aggregation of Aβ<sub>40</sub>. TEM analysis shows that Aβ<sub>40</sub> forms well-defined fibrils either in the absence or presence of L1 or L2 (Figure S9 in the <xref rid="notes-1" ref-type="notes">SI</xref>, panels 1 and 4). Importantly, the presence of Cu<sup>2+</sup> and Zn<sup>2+</sup> does not inhibit Aβ<sub>40</sub> aggregation,<sup><xref ref-type="bibr" rid="ref21">21</xref></sup> yet they change the morphology of the insoluble aggregates that become more amorphous (Figure S9 in the <xref rid="notes-1" ref-type="notes">SI</xref>, panels 2 and 3). Moreover, in the presence of Cu<sup>2+</sup> and either L1 or L2, a small amount of Aβ<sub>40</sub> fibrils are still formed (Figure S9 in the <xref rid="notes-1" ref-type="notes">SI</xref>, panels 5 and 8), while the presence of Zn<sup>2+</sup> and either L1 or L2 leads to the formation of a larger amount of fibrillar or amorphous aggregates, respectively (Figure S9 in the <xref rid="notes-1" ref-type="notes">SI</xref>, panels 6 and 9). The TEM results are also supported by native gel electrophoresis/Western blotting analysis (Figure S10 in the <xref rid="notes-1" ref-type="notes">SI</xref>), although this is less informative than in the case of the Aβ<sub>42</sub> peptide because of the lack of formation of an appreciable amount of soluble Aβ oligomers under any experimental conditions. Therefore, because we are particularly interested in evaluating the role of metal ions and bifunctional compounds in controlling the formation of the neurotoxic soluble Aβ oligomers, the Aβ<sub>42</sub> peptide was employed in the additional studies described below.</p></sec><sec id="sec3.8"><title>Disaggregation of Aβ<sub>42</sub> Fibrils</title><p>The effect of L1 and L2 on preformed Aβ<sub>42</sub> fibrils was also investigated in both the absence and presence of metal ions (Scheme <xref rid="sch3" ref-type="scheme">3</xref>). For disaggregation studies, Aβ<sub>42</sub> fibrils were prepared by incubation for 24 h at 37 °C, in both the presence and absence of metal ions. Then, L1 or L2 was added and further incubated for an additional 24 h at 37 °C. TEM and native gel electrophoresis/Western blotting characterization was performed on all of these samples. First, either L1 or L2 in metal-free conditions does not disaggregate the preformed Aβ<sub>42</sub> fibrils to an appreciable extent. Only some morphological changes were observed even after 24 h of incubation, and a significant amount of Aβ<sub>42</sub> aggregates were still observed by TEM (Figure <xref rid="fig8" ref-type="fig">8</xref>, panels 4 and 7). Native gel electrophoresis/Western blotting analysis reveals high-molecular-weight soluble Aβ<sub>42</sub> species and insoluble aggregates, although to a lesser extent in the presence of L1 (Figure <xref rid="fig8" ref-type="fig">8</xref>, lanes 4 and 7). In the presence of Cu<sup>2+</sup>, either L1 or L2 partially inhibits the Cu<sup>2+</sup>-promoted formation of small soluble aggregates and allows for enhanced Aβ<sub>42</sub> aggregation (Figure <xref rid="fig8" ref-type="fig">8</xref>, panels 5 and 8 vs panel 2). This is supported by native gel/Western blotting, which shows that while there is a slightly increased amount of low-molecular-weight (10–80 kDa) soluble Aβ<sub>42</sub> species, more insoluble aggregates are also observed at the top of the gel (Figure <xref rid="fig8" ref-type="fig">8</xref>, lanes 5 and 8). In the presence of Zn<sup>2+</sup>, either L1 or L2 seems to have a limited effect on the morphology of the insoluble Aβ<sub>42</sub> aggregates, as observed by both TEM and Western blotting (Figure <xref rid="fig8" ref-type="fig">8</xref>, panels/lanes 6 and 9). Overall, all of these in vitro studies suggest that compounds L1 and L2 should control the formation of Cu<sup>2+</sup>-stabilized soluble Aβ<sub>42</sub> species and thus should control their neurotoxicity (vide infra).</p><fig id="fig8" position="float"><label>Figure 8</label><caption><p>Top: TEM images of the disaggregation of Aβ<sub>42</sub> aggregation by L1 and L2, in the presence or absence of metal ions ([Aβ<sub>42</sub>] = [M<sup>2+</sup>] = 25 μM; [compound] = 50 μM; PBS, 37 °C, 24 h for fibrilization and an additional 24 h for disaggregation, scale bar = 500 nm). Bottom: Native gel electrophoresis/Western blotting analysis. Panels and lanes are as follows: (1) Aβ<sub>42</sub>; (2) Aβ<sub>42</sub> + Cu<sup>2+</sup>; (3) Aβ<sub>42</sub> + Zn<sup>2+</sup>; (4) Aβ<sub>42</sub> + L1; (5) Aβ<sub>42</sub> + L1 + Cu<sup>2+</sup>; (6) Aβ<sub>42</sub> + L1 + Zn<sup>2+</sup>; (7) Aβ<sub>42</sub> + L2; (8) Aβ<sub>42</sub> + L2 + Cu<sup>2+</sup>; (9) Aβ<sub>42</sub> + L2 + Zn<sup>2+</sup>; (10) MW marker.</p></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-00926c_0009" id="gr10" position="float"/></fig></sec><sec id="sec3.9"><title>Effect of L1 and L2 on the Aβ<sub>42</sub> Neurotoxicity</title><p>Because Cu<sup>2+</sup>–Aβ species have been shown to be neurotoxic,<sup><xref ref-type="bibr" rid="ref20">20</xref>,<xref ref-type="bibr" rid="ref21">21</xref>,<xref ref-type="bibr" rid="ref31">31</xref>,<xref ref-type="bibr" rid="ref51">51</xref>,<xref ref-type="bibr" rid="ref52">52</xref></sup> the development of metal-binding compounds that control the metal-dependent Aβ toxicity is desired. In this regard, the effect of L1 and L2 on the Aβ<sub>42</sub> neurotoxicity in N2A cells was investigated using an Alamar Blue cell viability assay.<sup><xref ref-type="bibr" rid="ref53">53</xref></sup> First, we examined the toxicity of both L1 and L2 along with ethylenediaminetetraacetic acid (EDTA) and CQ at various concentrations ranging from 0.2 to 20 μM (Figure S11 in the <xref rid="notes-1" ref-type="notes">SI</xref>).<sup><xref ref-type="bibr" rid="ref40">40</xref></sup> Both L1 and L2 show no appreciable cell toxicity (>80% cell survival) even up to 20 μM concentration. By comparison, the clinically tested compound CQ was quite toxic to cells even at 2 μM concentration, while EDTA does not show any toxicity up to 20 μM.<sup><xref ref-type="bibr" rid="ref25">25</xref>,<xref ref-type="bibr" rid="ref31">31</xref></sup> For the preformed Aβ<sub>42</sub> fibrils, no significant cell toxicity was observed in either the absence or presence of Cu<sup>2+</sup> (Figure <xref rid="fig9" ref-type="fig">9</xref>, lanes 1 and 2), supporting the previously reported diminished toxicity of Aβ<sub>42</sub> fibrils.<sup><xref ref-type="bibr" rid="ref5">5</xref>,<xref ref-type="bibr" rid="ref7">7</xref>,<xref ref-type="bibr" rid="ref54">54</xref>−<xref ref-type="bibr" rid="ref56">56</xref></sup> Moreover, compounds L1 and L2 do not affect the cell viability when added to the N2A cells in the presence of preformed Aβ<sub>42</sub> fibrils (Figure <xref rid="fig9" ref-type="fig">9</xref>, lanes 3 and 4). Even in the presence of preformed Aβ<sub>42</sub> fibrils and Cu<sup>2+</sup>, the cell viability is not affected by either L1 or L2 (Figure <xref rid="fig9" ref-type="fig">9</xref>, lanes 5 and 6). By contrast, CQ shows quite high cell toxicity to cells in the presence of Cu<sup>2+</sup> and Aβ<sub>42</sub> fibrils with a cell survival of only 26 ± 3% (Figure <xref rid="fig9" ref-type="fig">9</xref>, lane 7). Thus, these cellular toxicity data suggest that L1 and L2 do not lead to any cell toxicity in the presence or absence of Aβ<sub>42</sub> fibrils and Cu<sup>2+</sup> ions, suggesting that these compounds do not disaggregate Aβ<sub>42</sub> fibrils to generate neurotoxic soluble Aβ<sub>42</sub> species.</p><p>As shown previously,<sup><xref ref-type="bibr" rid="ref20">20</xref>,<xref ref-type="bibr" rid="ref21">21</xref></sup> the preformed soluble Aβ<sub>42</sub> oligomers show a limited cell survival of 45 ± 6% (Figure <xref rid="fig9" ref-type="fig">9</xref>, lane 8). Also, we have shown previously that while Cu<sup>2+</sup> by itself is not toxic to N2A cells, the presence of Cu<sup>2+</sup> and either monomeric Aβ<sub>42</sub> or soluble Aβ<sub>42</sub> oligomers exhibits enhanced cell toxicity.<sup><xref ref-type="bibr" rid="ref17">17</xref>,<xref ref-type="bibr" rid="ref18">18</xref></sup> However, the addition of either L1 or L2 to the N2A cells along with the preformed Aβ<sub>42</sub> oligomers drastically reduces the neurotoxicity of soluble Aβ<sub>42</sub> species to 100 ± 15% and 103 ± 15% cell viability, respectively (Figure <xref rid="fig9" ref-type="fig">9</xref>, lanes 9 and 10). Even in the presence of Cu<sup>2+</sup>, which has been shown to stabilize the soluble Aβ<sub>42</sub> oligomers,<sup><xref ref-type="bibr" rid="ref22">22</xref></sup> L1 and L2 were able to eliminate the neurotoxicity of soluble Aβ<sub>42</sub> oligomers (Figure <xref rid="fig9" ref-type="fig">9</xref>, lanes 9 and 10). These results are particularly exciting because, to the best of our knowledge, these are the first BFCs that reduce the toxicity of soluble Aβ<sub>42</sub> oligomers, in both the presence or absence of Cu<sup>2+</sup> ions. These BFCs thus represent promising lead compounds for the development of potential therapeutic agents that can control the metal-mediated neurotoxicity of soluble Aβ<sub>42</sub> species in AD and are currently the focus of our ongoing in vivo studies.</p><fig id="fig9" position="float"><label>Figure 9</label><caption><p>Cell viability (% DMSO control) of N2A cells determined by the Alamar Blue assay, upon incubation with (1) Aβ<sub>42</sub> fibrils, (2) Aβ<sub>42</sub> fibrils + Cu<sup>2+</sup>, (3) Aβ<sub>42</sub> fibrils + L1, (4) Aβ<sub>42</sub> fibrils + L2, (5) Aβ<sub>42</sub> fibrils + Cu<sup>2+</sup> + L1, (6) Aβ<sub>42</sub> fibrils + Cu<sup>2+</sup> + L2, (7) Aβ<sub>42</sub> fibrils + Cu<sup>2+</sup> + CQ, (8) Aβ<sub>42</sub> oligomers, (9) Aβ<sub>42</sub> oligomers + L1, (10) Aβ<sub>42</sub> oligomers + L2, (11) Aβ<sub>42</sub> oligomers + L1 + Cu<sup>2+</sup>, and (12) Aβ<sub>42</sub> oligomers + L2 + Cu<sup>2+</sup>. Conditions: [compound] = 20 μM; [Cu<sup>2+</sup>] = 20 μM; [Aβ<sub>42</sub>] = 20 μM. The <italic>t</italic>-test analysis reveals values of <italic>p</italic> < 0.001 for the pairs of data sets marked with asterisks.</p></caption><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="ic-2014-00926c_0010" id="gr11" position="float"/></fig></sec></sec><sec sec-type="conclusions" id="sec4"><title>Conclusion</title><p>In summary, two small BFCs, L1 and L2, were developed to specifically target the interactions of Cu<sup>2+</sup> and Zn<sup>2+</sup> ions with the Aβ<sub>42</sub> peptide and to control the formation of neurotoxic soluble Aβ<sub>42</sub> oligomers. The bifunctionality of L1 and L2 (i.e., metal chelation and Aβ interaction) was established through various spectroscopic studies such as spectrophotometric titrations, structural characterization, and fluorescent assays. We have previously shown that BFCs, which have high affinity for both metal ions and Aβ fibrils, can cause significant oligomerization via disaggregation of Aβ<sub>42</sub> fibrils and lead to enhanced neurotoxicity.<sup><xref ref-type="bibr" rid="ref20">20</xref></sup> However, the BFCs described herein bind to metal ions and Aβ fibrils only with moderate affinity and are having the desired opposite effect of promoting fibrillization of Cu<sup>2+</sup>-stabilized soluble Aβ<sub>42</sub> oligomers, and most importantly they drastically reduce the neurotoxicity of soluble Aβ<sub>42</sub> species, in both the presence or absence Cu<sup>2+</sup> ions. Because, to the best of our knowledge, these BFCs are the first to control the metal-mediated neurotoxicity of Aβ<sub>42</sub> oligomers, L1 and L2 are promising lead compounds for the development of chemical agents that can control the neurotoxicity of soluble Aβ<sub>42</sub> species in AD.</p></sec>
Impact of the level of pre-ART CD4+ T cells in blood on the rectal HIV reservoir in long-term treated men (VIRECT study)	Could not extract abstract	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Vergnon-Miszczycha</surname><given-names>Delphine</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A2"><name><surname>Girard</surname><given-names>Alexandre</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A3"><name><surname>Depincé</surname><given-names>Anne</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A4"><name><surname>Roblin</surname><given-names>Xavier</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I3">3</xref></contrib><contrib contrib-type="author" id="A5"><name><surname>Tedesco</surname><given-names>Emilie Del</given-names></name><xref ref-type="aff" rid="I3">3</xref></contrib><contrib contrib-type="author" id="A6"><name><surname>Frésard</surname><given-names>Anne</given-names></name><xref ref-type="aff" rid="I2">2</xref></contrib><contrib contrib-type="author" id="A7"><name><surname>Guglielminotti</surname><given-names>Claire</given-names></name><xref ref-type="aff" rid="I2">2</xref></contrib><contrib contrib-type="author" id="A8"><name><surname>Lambert</surname><given-names>Claude</given-names></name><xref ref-type="aff" rid="I4">4</xref></contrib><contrib contrib-type="author" id="A9"><name><surname>Pozzetto</surname><given-names>Bruno</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I5">5</xref></contrib><contrib contrib-type="author" id="A10"><name><surname>Lucht</surname><given-names>Frédéric</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref></contrib><contrib contrib-type="author" id="A11"><name><surname>Paul</surname><given-names>Stéphane</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I4">4</xref></contrib><contrib contrib-type="author" id="A12"><name><surname>Bourlet</surname><given-names>Thomas</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I5">5</xref></contrib>	BMC Infectious Diseases	<sec><title/><p>Gut associated lymphoid tissue (GALT) represents the largest reservoir of HIV-1. The aim of the study is: 1) to establish the viral and cellular characteristics of the HIV-1 rectal reservoir from chronically infected men under antiretroviral therapy (ART) for more than one year and less than five years, and, 2) to correlate the size of the rectal reservoir and the cellular composition of rectal mucosa to the level of CD4+ T cells in blood at the start of ART.</p><p>For each patient, whole blood and six rectal biopsies were collected. Total HIV DNA and cell-associated HIV RNA were quantified in PBMCs and in rectal samples by rtPCR assays. The cellular composition of blood and rectal samples in CD4+ and CD8+ T cells, Th17 lymphocytes, regulatory T cells (Treg), CD4+/p24+ and CD8+/ PD-1+ T cells was established by flow cytometry.</p><p>Up to now, 12 patients were enrolled: 4 with pre-ART CD4+ T cell count above 350/mm3, 3 between 200 and 350/mm3 and 5 under 200/mm3. The mean HIV DNA level in rectal cells was lower in patients who initiated ART with a blood CD4+ T cell count > 350/mm3, compared to those who started ART with a blood CD4+ T cell count < 200/mm3 (3.49 vs 3.74 log10 copies/106 cells). Moreover, the rate of rectal CD8+ T cells exhibiting an exhausted phenotype (CD8+/ PD-1+ T cells) tended to be negatively correlated to the pre-ART blood CD4+ T cell count (p = 0.06, rho = -0.61) and to the rectal [Th17/Treg] ratio (p = 0.058, rho = -0.66).</p><p>These preliminary results suggest that initiating ART with a high blood CD4+ T cells (> 350/mm3) limits the size of the HIV rectal reservoir and preserve local immunity, even after long-term therapy. This could participate to decrease local inflammation and viral replication. The study is still ongoing to increase the size of the effective and to complete these results by detecting viral integrated and 2LTR DNA forms, investigating co-infections agents and studying the inflammatory environment.</p></sec>
Next-generation LTR-specific Tre-recombinase targets a majority of HIV-1 isolates	Could not extract abstract	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Hauber</surname><given-names>Joachim</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A2"><name><surname>Karpinski</surname><given-names>Janet</given-names></name><xref ref-type="aff" rid="I2">2</xref></contrib><contrib contrib-type="author" id="A3"><name><surname>Hauber</surname><given-names>Ilona</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A4"><name><surname>Chemnitz</surname><given-names>Jan</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A5"><name><surname>Hofmann-Sieber</surname><given-names>Helga</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A6"><name><surname>Nagel</surname><given-names>Claus-Henning</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A7"><name><surname>Beschorner</surname><given-names>Niklas</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A8"><name><surname>Schäfer</surname><given-names>Carola</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A9"><name><surname>Buchholz</surname><given-names>Frank</given-names></name><xref ref-type="aff" rid="I2">2</xref></contrib>	BMC Infectious Diseases	<sec sec-type="intro"><title>Introduction</title><p>HIV-1 integrates into the host chromosome and persists as a provirus flanked by long terminal repeats (LTR). To date, treatment regimens primarily target the virus enzymes, virus attachment or virus-cell fusion, but not the integrated provirus. Thus, current antiretroviral therapies (i.e. cART) cannot eradicate HIV-1, a fact that highlights the urgency of pursuing new strategies to find a cure for HIV/AIDS.</p><p>Previously, we engineered an experimental LTR-specific recombinase (Tre-recombinase) that can effectively excise integrated HIV-1 proviral DNA from infected human cell cultures (Sarkar et al. 2007 Science 316:1912). Subsequently, we demonstrated highly significant antiviral activity of this HIV-1 subtype A-specific Tre in humanized mice (Hauber et al. 2013 PLOS pathogens 9:e1003587). Broad clinical application, however, requires availability of a tre-recombinase that recognizes a majority of clinical HIV-1 isolates.</p></sec><sec sec-type="materials\|methods"><title>Materials and methods</title><p>Here we report LTR target site identification as well as the engineering and functional analysis of a next-generation Tre-recombinase that recognizes the vast majority (e.g. >93% clade B and >80% clade A) of clinical HIV-1 isolates.</p></sec><sec sec-type="results"><title>Results</title><p>It is shown that the HIV-1 LTR harbours a conserved region that may serve as a universal tre recognition site for provirus excision. In fact, targeting this site by next-generation tre-recombinase demonstrates pronounced antiviral activity in the absence of cellular toxicity.</p></sec><sec sec-type="conclusions"><title>Conclusion</title><p>The presented data suggest that next-generation Tre technology may be a valuable component of future antiretroviral therapies to reverse infection and thereby providing a cure for HIV/AIDS.</p></sec>
Towards gene therapy against HIV-1: new therapeutic target in gag RNA accessible to ribozymes and RNA interference molecules	Could not extract abstract	<contrib contrib-type="author" corresp="yes" id="A1"><name><surname>Gatignol</surname><given-names>A</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A2"><name><surname>Scarborough</surname><given-names>RJ</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A3"><name><surname>Lévesque</surname><given-names>MV</given-names></name><xref ref-type="aff" rid="I2">2</xref></contrib><contrib contrib-type="author" id="A4"><name><surname>Boudrias-Dalle</surname><given-names>E</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A5"><name><surname>Chute</surname><given-names>IC</given-names></name><xref ref-type="aff" rid="I3">3</xref></contrib><contrib contrib-type="author" id="A6"><name><surname>Daniels</surname><given-names>SM</given-names></name><xref ref-type="aff" rid="I1">1</xref></contrib><contrib contrib-type="author" id="A7"><name><surname>Ouellette</surname><given-names>RJ</given-names></name><xref ref-type="aff" rid="I3">3</xref></contrib><contrib contrib-type="author" id="A8"><name><surname>Perreault</surname><given-names>J-P</given-names></name><xref ref-type="aff" rid="I2">2</xref></contrib>	BMC Infectious Diseases	<sec><title>Aim</title><p>Antisense molecules targeting HIV-1 RNA have the potential to be used as part of combination gene or drug therapy to treat HIV-1 infection to reach a functional or complete cure. Only a small number of extremely active molecules currently exist and a treatment option has not yet been identified in clinical trials. We have previously developed new hepatitis delta virus (HDV)-derived ribozymes (Rzs) called "switch on-off adaptor" (SOFA) to target HIV-1 RNA [<xref ref-type="bibr" rid="B1">1</xref>]. Our aim is to develop highly active RNA-based molecules with complementary mechanisms, which are able to reach a large number of HIV-1 variants.</p></sec><sec sec-type="methods"><title>Methods</title><p>We screened HIV-1 RNA to identify conserved target sites for new HDV-Rzs [<xref ref-type="bibr" rid="B2">2</xref>]. We designed new SOFA-HDV-Rzs against the Gag RNA and developed a rapid test to evaluate the inhibition of HIV-1 production. We designed small interfering (si) RNAs targeting the same region and tested their activity on HIV-1 replication.</p></sec><sec sec-type="results"><title>Results</title><p>We identified 13 conserved regions in the gag RNA and constructed the corresponding Rzs. We transfected HEK293T cells with these Rzs and HIV-1 molecular clones. We identified one Rz that was particularly efficacious. We then constructed siRNAs and short hairpin (sh) RNAs targeting the same sequence. The shRNA was very active against HIV-1 clades B, C and A/G. Neither the Rz, nor the shRNA disturbs the cellular transcriptome, suggesting no toxicity. In lymphocytic cell lines, both the Rz and the shRNA inhibit long-term HIV-1 replication [<xref ref-type="bibr" rid="B2">2</xref>].</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>We identified new SOFA-HDV-Rz, siRNA and shRNA targeting HIV-1 Gag RNA. The shRNA is as active as the only shRNA that has advanced to clinical trials and targets more strains. Long-term inhibitory activity of these molecules shows that this site is particularly accessible to other antisense molecules. These molecules have a high potential to be used in combination gene therapy or as drugs with appropriate delivery methods.</p></sec>